Swarn Tantram

Investigating the simultaneous influence of
intrinsic and extrinsic cues:
An examination of the interaction between
country of origin, price and selected sensory variables
By
Roberta Veale
A thesis submitted in fulfillment of
the requirements for the degree of
Doctor of Philosophy
(Commerce)
July 2007
Abstract
The research was undertaken to quantify the power of selected extrinsic (price and country
of origin) and intrinsic cues (acid in chardonnay and fat in brie) on consumer evaluations of
both experienced and expected product quality, and further, to measure the respective
influences of objective knowledge, subjective knowledge and self-confidence on these
quality assessments. The study also seeks to determine if a survey conducted measuring
expectations of quality would yield comparable results with quality assessments based on
actual product experience. The study was comprised of sensory experiments using full
profile conjoint analysis experimental design to measure quality perception, followed by a
survey where only product description profiles were provided. The analysis from both
stages show findings that are remarkably similar in most respects.
Results of the sensory experiment for chardonnay show both extrinsic cues tested to be
more important than acid levels, while results for the survey show price maintained the
strongest influence, with comparable expectations regarding the importance of country of
origin and acid. For brie (both stages) consumer opinions were consistent; with price
found the most influential; and while country of origin was considered relatively important,
fat levels were more influential for both groups. Whilst for chardonnay (both stages)
respondents held consistent beliefs regarding each acid level tested, for brie respondents
experiencing the highest level of fat held an opposite view to respondents assessing
quality based on their expected liking for this type of product. The influence of knowledge
(objective and subjective) and self-confidence was found to be sporadic and weak, likely
due to respondents general lack of objective knowledge in both stages of the survey. In
the case of self-confidence, results are surprising given that respondents in both studies
exhibited reasonably healthy degrees of self-confidence. The research provides important
information to marketers seeking to exploit the most attractive aspects of their products
and platform for a number of subsequent studies.
i
Acknowledgements
There are many people that must be acknowledged as important contributors to the
completion of this study. Firstly, I would like to thank my primary supervisor Professor
Pascale Quester. She was my honors supervisor and told me I think you can do it, and if
she thought so, then I imagined that I could. In the 3 years that have passed since we had
that meeting she has also become a valued friend in addition to supervisor and mentor. My
co-supervisor, Dr Amal Karunaratna, has also been a great source of knowledge,
encouragement and ideas. He was particularly instrumental in the development of the
conjoint analysis design and initial data testing. We spent more than a few hours working
through some difficult and challenging issues. We were helped with this by Dr Hume
Winzor, whose expertise in the use of conjoint analysis was invaluable. Another source of
invaluable expertise was Professor Jordan Louviere, who took the time to meet with me to
discuss my research and offered much needed advice regarding the methodology
employed. This research would not have been possible without the guidance and support
of industry product experts, many thanks to Jim Smith, Michael Ross and Louise Elder. I
would also like to acknowledge my work colleagues in 2 locations. Firstly, the TAFESA
City Campus where I received ongoing encouragement and support from my direct reports
Dr. Rodger Thomas and Matt Stanton backed up by my fellow lecturers and executive
management. Secondly, my work mates at the School of Commerce (University of
Adelaide) who provided advice and friendship. I would also like to thank specifically,
Professor Lee Parker and Associate Professor Barry Burgan for helping me whenever they
could and providing the environment I needed to just get it done. Finally, I would like to
thank my good friend and tireless research assistant, Emma Parker, for numbering all
those glasses and pouring all those samples (in addition to about 100 other things).
This research was made possible by funding by the
Wine and Grape Research Development Board.
ii
Table of Contents
Chapter page
ABSTRACT i
AKNOWLEDGEMENTS ii
TABLE OF CONTENTS iii
LIST OF TABLES xi
List OF FIGURES xxi
1 INTRODUCTION 1
1.1 BACKGROUND TO THE RESEARCH 1
1.2 RESEARCH JUSTIFICATION 3
1.3 OBJECTIVES AND CONTRIBUTION OF RESEARCH 5
1.4 THESIS OUTLINE 6
1.5 SUMMARY 8
2 LITERATURE REVIEW 9
2.1 CONSUMER USE OF INTRINSIC AND EXTRINSIC CUES 9
2.1.1 Extrinsic cues and sensory perceptions 11
2.2 COO AS AN EXTRINSIC CUE 12
2.2.1 Introduction to the COO construct 12
2.2.2 How country images are developed 15
2.2.3 Building country equity through international brands 17
2.2.4 Consumer bias against products from developing countries 18
2.2.5 The challenge of marketing 'hybrid' products 21
2.2.6 Effect of market specific attributes on use of CI 24
2.2.7 Effect of CI on commercial customers 28
2.2.8 Effect of consumer knowledge on use of CI 29
iii
2.2.9 COO summary 30
2.3 PRICE AS AN EXTRINSIC CUE 35
2.3.1 The role of price in consumer buying decision making 35
2.3.2 Effect of consumer knowledge on use of price 37
2.3.3 Price summary 37
2.4 CONSUMER EXPERTISE AND SELF-CONFIDENCE 38
2.4.1 Two dimensions of knowledge 38
2.4.1.1 Objective knowledge 40
2.4.1.2 Subjective knowledge 42
2.4.2 Consumer self-confidence 44
2.4.3 Consumer knowledge and self-confidence summary 45
2.5 GAPS IN THE EXISTING LITERATURE 46
2.5.1 Towards a conceptual framework 47
2.6 SUMMARY 48
3 CAUSAL MODEL AND HYPOTHESIS DEVELOPMENT 49
3.1 INTRODUCTION 49
3.2 RESEARCH FRAMEWORK 49
3.2.1 Testing the power of extrinsic cues 49
3.2.2 Testing the influence of consumer characteristics 50
3.2.3 Use of sensory experiments 51
3.2.4 Suitability of conjoint analysis experimental design 51
3.2.4.1 Formulating a conjoint analysis design 52
3.2.5 Expected quality vs perceived quality 54
3.2.6 Development of causal model and hypotheses 54
3.2.6.1 Causal model and hypotheses 54
iv
3.2.6.2 Hypotheses 57
3.2.7 Research paradigm 58
3.2.8 Justification for the model 59
3.2.9 Stimuli used 60
3.2.9.1 Wine and cheese 60
3.2.9.2 COO and price 61
3.2.9.3 Hypotheses summary 62
3.3 OVERVIEW OF RESEARCH DESIGN 64
3.4 STAGE 1 QUALITATIVE STUDY 66
3.4.1 Sampling 66
3.4.2 Data collection 67
3.4.3 Results 68
3.4.3.1 Taste testing 72
3.4.3.2 Country rankings (survey of students) 73
3.4.4 Qualitative data analysis summary 75
3.5 SUMMARY 76
4 QUANTITATIVE METHODOLOGY 77
4.1 INTRODUCTION 77
4.2 SAMPLING METHODS 77
4.3 DATA COLLECTION INSTRUMENTS 78
4.3.1 Use of full profile conjoint analysis 78
4.3.2 Orthogonal fractional factorial design 79
4.3.3 Training respondents 80
4.4 MEASURES USED FOR CONSUMER CHARACTERISTICS 80
4.4.1 Subjective knowledge 80
v
4.4.2 Objective knowledge 81
4.4.3 Self-confidence 86
4.5 SURVEY DATA ANALYSIS 87
4.5.1 Correlations and factor analysis 88
4.5.2 Objective knowledge 88
4.5.3 Conjoint analysis 89
4.6 DETERMINING ATTRIBUTE IMPORTANCE 89
4.6.1 Non-parametric tests 91
4.7 VALIDITY OF INSTRUMENTS 92
4.7.1 Normality testing 92
4.7.2 Subjective knowledge and self-confidence scale 92
4.7.3 Objective knowledge questions 93
4.7.4 Conjoint analysis 93
4.8 SUMMARY 94
5 QUANTITATIVE CONJOINT PILOT STUDY 95
5.1 INTRODUCTION 95
5.2 CONJOINT ANALYSIS SURVEY (PILOT) 95
5.2.1 Sample (pilot) 95
5.2.2 Data collection instrument (pilot questionnaire) 96
5.2.3 Intrinsic and extrinsic cues - wine 97
5.2.4 Intrinsic and extrinsic cues - cheese 98
5.2.5 Questionnaire pre-test 101
5.3 VALIDATION OF RESEARCH INSTRUMENTS (PILOT) 103
5.3.1 Subjective knowledge and self-confidence scales 103
5.3.2 Objective knowledge tests (pilot) 109
vi
5.3.3 Conjoint analysis fractional factorial design (pilot) 110
5.4 RESULTS AND DISCUSSION (PILOT) 110
5.4.1 Sample (pilot) 110
5.4.2 Conjoint analysis results (pilot) 113
5.4.3 Results knowledge and self-confidence (pilot) 115
5.5 PILOT STUDY RESULTS SUMMARY 116
5.5.1 Needed changes to questionnaire (pilot) 116
6 SENSORY EXPERIMENT METHODOLOGY 118
6.1 INTRODUCTION 118
6.2 SAMPLE (SENSORY) 118
6.3 DATA COLLECTION INSTRUMENT (SENSORY) 119
6.3.1 Intrinsic and extrinsic cues wine 120
6.3.2 Triangle and paired sample tests 120
6.3.3 Intrinsic and extrinsic cues cheese 123
6.3.4 Change of rating scale anchors 125
6.3.5 Knowledge and self-confidence 126
6.3.6 Changes to questionnaire layout and format 127
6.3.7 Group briefings 127
6.3.8 Conducting the experiment 129
6.4 VALIDATION OF RESEARCH INSTRUMENTS (SENSORY) 132
6.4.2 Objective knowledge tests 137
6.4.3 Conjoint analysis fractional factorial design 137
6.5 SUMMARY 138
7 SENSORY DATA ANALYSIS RESULTS 139
vii
7.2 SAMPLE PROFILE 139
7.3 CONJOINT ANALYSIS RESULTS (SENSORY) 140
7.3.1 Chardonnay 140
7.3.1.1 Chardonnay profiles and likelihood of purchase 141
7.3.1.2 Knowledge and self-confidence levels 142
7.3.1.3 Influence of knowledge and self-confidence 143
7.3.1.4 Objective knowledge segments (chardonnay) 147
7.3.1.5 Subjective knowledge segments (chardonnay) 148
7.3.1.6 Self-confidence segments (chardonnay) 150
7.3.1.7 Chardonnay summary 152
7.3.2 Brie 153
7.3.2.1 Brie profiles and the likelihood of purchase 155
7.3.2.2 Knowledge levels brie 155
7.3.2.3 Influence of knowledge and self-confidence (brie) 156
7.3.2.4 Objective knowledge segments (brie) 159
7.3.2.5 Brie subjective knowledge segments 162
7.3.2.6 Brie self-confidence segments (sensory) 164
7.3.2.7 Summary of test results Brie 165
7.4 SENSORY ANALYSIS SUMMARY 166
8 CONJOINT ANALYSIS SURVEY AND RESULTS FOR CHARDONNAY 168
8.2 METHODOLOGY 168
8.2.1 Sample (conjoint survey) 168
8.2.2 Data collection instrument (conjoint survey) 170
viii
8.3 VALIDATION OF RESEARCH INSTRUMENTS (SURVEY) 171
8.3.2 Objective knowledge tests 172
8.3.3 Conjoint analysis fractional factorial design 173
8.4 CONJOINT SURVEY DATA ANALYSIS RESULTS 173
8.4.1 Sample profile 173
8.4.2 Chardonnay conjoint analysis results 174
8.4.3 Chardonnay profiles and likelihood of purchase 176
8.4.4 Knowledge and self-confidence levels 177
8.4.4.1 Influence of knowledge and self-confidence 178
8.4.5 Objective knowledge segments (chardonnay) 181
8.4.6 Subjective knowledge segments (chardonnay) 184
8.4.7 Self-confidence segments (chardonnay) 187
8.5 CHARDONNAY SUMMARY 189
9 CONJOINT ANALYSIS SURVEY RESULTS FOR BRIE 192
9.2 BRIE CONJOINT ANALYSIS RESULTS 192
9.2.1 Brie profiles and likelihood of purchase 194
9.2.2 Knowledge levels brie 194
9.2.2.1 Influence of knowledge and self-confidence (brie) 195
9.2.3 Objective knowledge segments (brie) 198
9.2.4 Subjective knowledge segments (brie) 200
9.2.5 Self-confidence segments (brie) 203
9.3 BRIE SUMMARY 206
10 CONCLUSIONS, LIMITATIONS AND IMPLICATIONS 209
ix
10.2 OVERVIEW OF THE RESEARCH AND FINDINGS 209
10.3 CONTRIBUTION TO THE LITERATURE 213
10.3.1 Consumer reliance on extrinsic cues: COO and price 214
10.3.2 Knowledge and self-confidence 215
10.4 IMPLICATIONS FOR MARKETERS 217
10.5 LIMITATIONS 220
10.6 DIRECTIONS FOR FUTURE RESEARCH 222
10.7 SUMMARY 223
11 REFERENCES 225
12 APPENDICES 242
1 FOCUS GROUP DISCUSSION GUIDE 242
2 PILOT STUDY QUESTIONNAIRE 245
3 NORMALITY ASSUMPTIONS AND TESTING 257
3.1 ASSUMPTIONS FOR NORMALITY 258
3.2 PILOT STUDY DATA 259
3.3 SENSORY EXPERIMENT DATA 260
3.4 CONJOINT SURVEY DATA 261
4 SCALE RELIABILITY AND VALIDITY 262
4.5 RELIABILITY ANALYSIS FOR ALL SCALES 263
4.6 CONSTRUCT VALIDITY 264
4.6.1 Assumptions 265
4.6.2 Subjective knowledge scale test results (Chardonnay) 268
4.6.3 Subjective knowledge scale test results (camembert/brie) 271
x
4.6.4 Self-confidence scale 274
4.6.5 Discriminant validity 278
5 INFORMATION PACK AND REGISTRATION FORM 282
6 TASTING REGISTRATION SPREADSHEET AND
RESPONDENT CONFIRMATION NOTICE 285
6.1 SPREADSHEET FOR REGISTRATIONS 286
6.2 CONFIRMATION NOTICE 287
7 SENSORY EXPERIMENT QUESTIONNAIRE 288
8 CORRELATION MATRICES (CHARDONNAY) 307
8.1 SENSORY EXPERIMENT 308
8.2 CONJOINT SURVEY (CHARDONNAY) 314
9 CORRELATION MATRICES (BRIE) 320
9.1 SENSORY EXPERIMENT 321
9.2 CONJOINT SURVEY (BRIE) 327
10 INVITATIONS TO PARTICIPATE IN SURVEY 333
10.1 ELECTRONIC INVITATION TO PARTICIPATE 334
10.2 CLASS ROOM INVITATION TO PARTICIPATE
(TRANSPARENCY SHOWN BY LECTURERS) 336
11 CONJOINT SURVEY QUESTIONNAIRE 337
xi
List of Tables
Table 3.1 Hypotheses tested 63
Table 3.2 Demographic profile of focus group participants 67
Table 3.3 Important product attributes for wine 68
Table 3.4 important product attributes for cheese 69
Table 3.5 Countries suggested by participants as indicators of quality for
wine and cheese 69
Table 3.6 Taste tests - wine 72
Table 3.7 Taste tests - cheese 73
Table 3.8 Country rankings for chardonnay 74
Table 3.9 Country rankings for camembert 75
Table 4.1 Subjective knowledge scale items 81
Table 4.2 Objective knowledge questions wine/chardonnay 84
Table 4.3 Objective knowledge questions cheese/camembert 85
Table 4.4 Self-confidence scale items 87
Table 5.1 Specification of chardonnay attributes and levels 97
Table 5.2 Fractional factorial design chardonnay 97
Table 5.3 Specification of camembert attributes and levels 99
Table 5.4 Fractional factorial design camembert 99
Table 5.5 Total variance explained (pre-test) 101
Table 5.6 Reliability coefficients of scales (pre-test) 101
Table 5.7 Summary of part worths for chardonnay (pre-test) 102
Table 5.8 Summary of part worths for camembert (pre-test) 103
Table 5.9 Conjoint analysis internal validity tests (pre-test) 103
xii
Table 5.10 Reliability coefficients of scales (pilot) 104
Table 5.11 Construct reliability for subjective knowledge and self-confidence 104
Table 5.12 Subjective knowledge chardonnay (pilot) 105
Table 5.13 Subjective knowledge camembert (pilot) 106
Table 5.14 Self confidence factors (pilot) 107
Table 5.15 Wilcoxon Signed Ranks Test Self-confidence 108
Table 5.16 Conjoint analysis internal validity tests (pilot) 110
Table 5.17 Sample demographic profile 112
Table 5.18 Summary of part worths for chardonnay (pilot) 114
Table 5.19 Summary of part worths for camembert (pilot) 114
Table 5.20 Equivalent mean scores for knowledge and self-confidence 115
Table 6.1 Specification of chardonnay attributes and levels 123
Table 6.2 Fractional factorial design chardonnay 123
Table 6.3 Specification of brie attributes and levels 124
Table 6.4 Fractional factorial design brie 124
Table 6.5 Reliability coefficients of scales (sensory) 132
Table 6.6 Construct reliability for scales (sensory) 132
Table 6.7 Self-confidence factors (sensory) 133
Table 6.8 Reliability tests of self-confidence items (sensory) 134
Table 6.9 Descriptive statistics self-confidence items (sensory) 135
Table 6.10 Comparison of item groupings (sensory) 135
Table 6.11 Significance testing of average item scores 136
Table 6.12 Conjoint analysis internal validity tests (sensory) 137
Table 7.1 Profile demographic profile (sensory) 139
Table 7.2 Summary of part worths and utilities chardonnay (sensory) 140
xiii
Table 7.3 Average values per profile chardonnay (sensory) 141
Table 7.4 Average chardonnay profile values by purchase intentions 142
Table 7.5 Mean scores for knowledge and self-confidence (sensory) 142
Table 7.6 Spearmans rho Average importance chardonnay 143
Table 7.7 Spearman's rho - Utilities chardonnay 144
Table 7.8 Quartile values for knowledge and self-confidence (chardonnay) 147
Table 7.9 Comparison of part worths and utilities objective knowledge 147
Table 7.10 Utility comparison between high and low objective knowledge groups 147
Table 7.11 Comparison of part worths and utilities subjective knowledge 149
Table 7.12 Utility comparison between high and low subjective knowledge groups 150
Table 7.13 Comparison of part worths and utilities self-confidence 152
Table 7.14 Utility comparison between high and low self-confidence groups 152
Table 7.15 Summary of part worths and utilities brie (sensory) 154
Table 7.16 Average values per profile brie (sensory) 154
Table 7.17 Average brie profile values by purchase intentions 155
Table 7.18 Mean scores for knowledge (brie) 156
Table 7.19 Spearmans rho Average importance brie 157
Table 7.20 Spearmans rho Utilities brie 157
Table 7.21 Quartile values for knowledge and self-confidence (brie) 158
Table 7.22 Comparison of part worths and utilities brie objective knowledge 161
Table 7.24 Comparison of part worths and utilities brie subjective knowledge 163
Table 7.26 Comparison of part worths and utilities brie - self-confidence 165
xiv
Table 8.1 Reliability coefficients of scales (survey) 171
Table 8.2 Construct reliability for scales (survey) 172
Table 8.3 Conjoint analysis internal validity tests (survey) 173
Table 8.4 Profile demographic profile (survey) 173
Table 8.5 Summary of part worths and utilities chardonnay 175
Table 8.6 Comparative average values per profile chardonnay 176
Table 8.7 Average chardonnay profile values by purchase intentions 177
Table 8.8 Mean scores for knowledge and self-confidence (sensory and survey) 178
Table 8.9 Mann-Whitney U test for comparison of knowledge and self-confidence 178
Table 8.10 Spearman's rho - Average importance chardonnay 179
Table 8.11 Spearman's rho - Utilities chardonnay 180
Table 8.13 Comparison of part worths and utilities objective knowledge chardonnay
182
Table 8.15 Objective knowledge summary comparison of significant correlations 183
Table 8.18 Subjective knowledge summary comparison of significant correlations 186
Table 8.19 Comparison of part worths and utilities self-confidence 187
Table 8.21 Self-confidence summary comparison of significant correlations 188
Table 8.22 Hypotheses testing outcomes (chardonnay) 191
Table 9.1 Summary of part worths and utilities brie 193
Table 9.2 Comparative average values per profile brie (sensory and survey) 193
xv
Table 9.3 Average brie profile values by purchase intentions (survey) 194
Table 9.4 Mean scores for knowledge (sensory and survey) 195
Table 9.5 Mann-Whittney U test for comparison of knowledge between groups 195
Table 9.6 Spearmans rho Average attribute importance brie 196
Table 9.7 Spearmans rho Utilities brie 197
Table 9.9 Comparison of part worths and utilities objective knowledge 198
Table 9.11 Objective knowledge summary comparison of significant correlations (brie)
200
Table 9.13 Comparison between high and low subjective knowledge groups 201
Table 9.14 Subjective knowledge summary comparison of significant correlations (brie)
203
Table 9.15 Self-confidence summary comparison of significant correlations (brie) 204
Table 9.17 Hypotheses testing outcomes (brie) 205
TABLES FROM APPENDICES
Table A 1 Normality tests (pilot) 259
Table A 2 Normality tests (sensory) 260
Table A 3 Normality tests (survey) 261
Table A 4 Reliability Coefficients of Scales 263
Table A 5 Construct validity summary for all scales 265
Table A 6 Factorability of subjective knowledge scale chardonnay (pilot) 268
Table A 7 Spearmans r correlation coefficients - Subjective knowledge scale
chardonnay (pilot) 268
xvi
Table A 8 Factorability of subjective knowledge scale chardonnay (sensory) 269
Table A 9 Spearmans r correlation coefficients - Subjective knowledge scale
chardonnay (sensory) 269
Table A 10 Factorability of subjective knowledge scale chardonnay (survey) 270
Table A 11 Spearmans r correlation coefficients Subjective knowledge scale
chardonnay (survey) 270
Table A 12 Factorability of Subjective knowledge scale camembert (pilot) 271
brie (survey) 271
Table A 14 Factorability of Subjective knowledge scale brie (sensory) 272
brie (sensory) 272
Table A 16 Factorability of Subjective knowledge scale brie (survey) 273
brie (survey) 273
Table A 18 Factorability of self-confidence scale (pilot) 274
Table A 19 Spearmans r correlation coefficients Self-confidence scale (pilot) 274
Table A 20 Factorability of self-confidence scale (sensory) 275
Table A 21 Spearmans r correlation coefficients Self-confidence scale (sensory)
275
Table A 22 Factorability of reduced self confidence scale (sensory) 276
Table A 23 Spearmans r correlation coefficients Reduced self-confidence scale
(sensory) 276
Table A 24 Factorability of reduced self confidence scale (survey) 277
(survey) 277
xvii
Table A 26 Discriminant validity chardonnay and self confidence (pilot) 278
Table A 27 Discriminant validity brie and self confidence (pilot) 279
Table A 28 Discriminant validity chardonnay and self confidence (sensory) 279
Table A 29 Discriminant validity brie and self confidence (sensory) 280
Table A 30 Discriminant validity chardonnay and self confidence (survey) 280
Table A 31 Discriminant validity brie and self confidence (survey) 281
Table A 32 High objective knowledge and average importance chardonnay 308
Table A 33 High objective knowledge and utility values 308
Table A 34 Low objective knowledge and average importance chardonnay 309
Table A 35 Low objective knowledge and utility values 309
Table A 36 High subjective knowledge and average importance chardonnay 310
Table A 37 High subjective knowledge and utility values 310
Table A 38 Low subjective knowledge and average importance chardonnay 311
Table A 39 Low subjective knowledge and utility values 311
Table A 40 High self-confidence and average importance chardonnay 312
Table A 41 High self-confidence and utility values 312
Table A 42 Low self-confidence and average importance chardonnay 313
Table A 43 Low self-confidence and utility values 313
Table A 44 High objective knowledge and average importance chardonnay 314
Table A 46 Low objective knowledge and average importance chardonnay 315
Table A 48 High subjective knowledge and average importance chardonnay 316
Table A 50 Low subjective knowledge and average importance chardonnay 317
xviii
Table A 52 High self-confidence and average importance chardonnay 318
Table A 54 Low self-confidence and average importance chardonnay 319
Table A 56 High objective knowledge and average importance brie 321
Table A 58 Low objective knowledge and average importance brie 322
Table A 60 High subjective knowledge and average importance brie 323
Table A 62 Low subjective knowledge and average importance brie 324
Table A 64 High self-confidence and average importance brie 325
Table A 66 Low self-confidence and average importance brie 326
Table A 68 High objective knowledge and average importance brie 327
Table A 70 Low objective knowledge and average importance brie 328
Table A 72 High subjective knowledge and average importance brie 329
Table A 74 Low subjective knowledge and average importance brie 330
xix
Table A 76 High self-confidence and average importance brie 331
Table A 78 Low self-confidence and average importance brie 332
xx
List of Figures
Figure 2.1 Conceptual framework- effect of consumer characteristics on assessment
of product quality 47
Figure 3.1 Causal model- moderating effects of consumer characteristics on product
cue usage 56
Figure 3.2 Stages of the research 64
Figure 5.1 Pilot study questionnaire format 96
Figure 5.2 Example of wine product profile (pilot) 100
Figure 5.3 Example of cheese product profile (pilot) 100
Figure 6.1 Example of wine product profile (sensory) 126
Figure 6.2 Example of cheese product profile (sensory) 126
Figure 6.3 Amended questionnaire structure: 129
Figure 6.4 Numbering wine glasses 130
Figure 6.5 Pouring wine samples 130
Figure 6.6 Treated and untreated chardonnay 130
Figure 6.7 Preparing brie samples 130
Figure 6.8 Preparing trays for a tasting session 131
Figure 6.9 Sample order 131
Figure 6.10 Tray with samples 131
Figure 6.11 Experiment in progress 131
Figure 8.1 Example of wine product profile (survey) 170
Figure 8.2 Example of cheese product profile (survey) 171
xxi
1 Introduction
1.1 Background to the research
Products and services are comprised of both intrinsic and extrinsic cues used by
consumers to shape their opinions of expected or experienced product quality. An
intrinsic cue can be described as any product attribute inherent to the product itself,
such as engine capacity for a car or flavor for a soft drink. An extrinsic cue is a product
attribute (or closely associated attribute) that is not a fundamental part of the product
itself, for example, price, brand, place of purchase or country of origin. Consumers are
not always able to accurately evaluate these cues prior to purchase or judge quality in a
post-purchase evaluation (Alba 2000; Kardes, Kim et al. 2001). Reasons for this
include: lack of understanding, lack of self-confidence, misinterpretation or
inaccessibility to information (Roper 1969; Olson 1972; Szbillo and Jacoby 1974;
Monroe 1976; Rao and Olson 1990; Northen 2000; Teas and Agarwal 2000; Wansink,
Park et al. 2000; Siu and Wong 2002; Kardes, Cronley et al. 2004). Managers should
understand the respective influence of extrinsic cues in the product quality assessment
process to ensure that marketing efforts are focused towards enhancing those attributes
most likely to favorably influence consumers opinions regarding quality.
As an extrinsic cue, Country of Origin (COO) has been studied extensively since the
1960s for its effect on consumers' product evaluation (Bilkey and Nes 1982). COO is
generally described as the 'source country' for a product or service provider, which may
be different from the country of brand or country of assembly or design. Country Image
(CI) is one aspect of this complex construct. CI profiles are akin to a 'brand profile' for
individual countries and are used by consumers as extrinsic cues for forming product
quality expectations (Badri, Davis et al. 1995; Nebanzahl and Jaffe 1996; Kotler and
Gertner 2002; Papadopoulos and Heslop 2002; Srikatanyoo and Gnoth 2002). While
1
these stereotypical country images are considered specific for a product (or product
category) these values are generally not transferable to different categories of products,
some common effects have emerged. These common findings include a more
significant reliance on COO (and other extrinsic cues) by consumers when there is little
specific and reliable information available for consideration, when they are evaluating a
high cost/high involvement product, when they have limited personal objective
knowledge to rely on or when the COO and product category are highly congruent, for
example, French perfume or Chinese silk (Eroglu and Machleit 1988; Han 1989;
Papadopoulos and Heslop 1989; Han 1990; Papadopoulos and Heslop 1993; Eriksson
and Hadjikhani 2000; Piron 2000; Srikatanyoo and Gnoth 2002). The influence of price
as an extrinsic cue on consumer perceptions of product quality has also been studied
extensively, with research confirming that consumers often use price as an extrinsic cue
to predict quality across a wide range of products (Sullivan and Burger 1987; Pechmann
and Ratneshwar 1992; Jover, Montes et al. 2004). This price/quality relationship,
described in the literature as the price-reliance schema reflects consumers strongly
held view that you get what you pay for (Lee and Lou 1996: p24).
The focus of this research is to investigate the respective influence of COO and price as
extrinsic cues when product intrinsic cues are experienced through sensory perception.
This will be achieved through the analysis of data collected by taste testing
experiments. Previous studies have assessed the influence of price (among selected
other extrinsic cues) by numerous means of sensory evaluation including taste tests
and visual impressions (Pechmann and Ratneshwar 1992; Wansink, Park et al. 2000;
Hurling and Shepherd 2003). However, research testing the specific influence of COO
on consumer opinions using taste testing methodology is limited, representing an
opportunity to expand our understanding of their value to consumers when all intrinsic
cues are available for evaluation (Roper 1969; Pechmann and Ratneshwar 1992;
2
Aaron, Mela et al. 1994; Imram 1999; Acebron and Dopico 2000; Hoffmann 2000;
Becker 2000a; Koch and Koch 2003; Zellner and Durlach 2003).
Another important and unique aspect of the research is the opportunity to compare
findings between sensory evaluation experiments measuring respondent assessment of
experienced quality and a paper based survey measuring consumer product
preferences based on pre-use judgments of expected product quality as determined
through assessment of product descriptions incorporating intrinsic and extrinsic cues.
Hence a secondary objective of the study is to assess any differences that may exist
between respondents expectation of quality versus perceived quality, as determined
through product consumption and immediate appraisal.
1.2 Research Justification
Marketing practitioners need to understand which product attributes are used by
consumers when forming opinions regarding product quality as this determination is
often a significant precursor to a buying decision (Olson 1972; Szbillo and Jacoby 1974;
Parasuraman, Berry et al. 1993; Sweeney, Soutar et al. 1999; Northen 2000). This
intelligence reduces the risk of emphasizing product attributes that are not considered
important, may not be noticed and may not even be understood. Research has shown
that generally consumers rely more heavily on intrinsic cues than extrinsic cues to
predict quality, however, this reliance has been found to be moderated by a number of
factors (Speece, Kawahara et al. 1994; Hoffmann 2000; Skuras and Vakrou 2002). For
example, product cue usage can be situational with consumers relying on extrinsic
cues when they are under time pressure to make a decision, when they are making a
high involvement purchase or when buying gifts for others (Lee and Lou 1996; Quester
and Smart 1998; Piron 2000) Therefore, in many instances a consumers evaluation of
quality, or expectation of quality, may not even involve the evaluation of intrinsic cues
3
with extrinsic cues taking precedence. Further, market specific variables such as
consumer knowledge and self-confidence have also been found to influence the use
and understanding of both intrinsic and extrinsic cues alike, suggesting that the
potential power of cues such as price or COO may often be undervalued (Aaron, Mela
et al. 1994; Andreassen and Lindestad 1998; Alba and Hutchinson 2000; Alba 2000).
Whilst these types of consumer characteristics have been found influential, results are
often ambiguous and conflicting due to a lack of consistency in construct definitions and
measures employed (Park, Mothersbaugh et al. 1994; Flynn and Goldsmith 1999). This
research provides a vehicle to measure each distinctively and quantify their respective
influences on consumer product quality assessment. Also, as a result of this process,
new and validated measurement instruments will be confirmed for further development
and/or use in later studies.
The risk for marketers is that scarce resources may be wasted, emphasizing product
attributes misunderstood or of little importance to buyers. Rather, marketers could
exploit an important opportunity provided by this understanding to develop a hierarchy
of the product attributes (under various situations) that consumers use to determine
quality evaluations and drive their buying decisions. The study aims to address some
aspects of these gaps in the current literature and provide a background of
understanding that can be used in industry to develop effective, appropriate and specific
marketing strategies. Lastly, the study provides insight into the ability of research using
consumer assessments based on product descriptions to predict opinions based on
respondent experience, thus evaluating the efficiency and accuracy of data gathered
using paper based survey methodology alone.
4
1.3 Objectives and contribution of research
The research provides a framework illustrating consumer assessment of product quality
based on their simultaneous evaluation of intrinsic and extrinsic cues, with knowledge
(type and level) and personal self-confidence levels as variables moderating their use.
Further, the study identifies the differences between what consumers would expect in
terms of product quality, and their actual determining of product quality through the
comparison of results from a conjoint analysis survey and taste testing experiments.
The resulting framework will be a useful tool for practitioners to assist them in the
development of marketing strategies that emphasize the most powerful and influential
product cues.
The research objectives are:
1. To investigate the ability of extrinsic cues to override sensory perceptions.
2. To measure the levels and types of consumer knowledge and their moderating
influence on the use of extrinsic versus intrinsic cues.
3. To measure the levels of consumer self-confidence and its moderating effect on
cue reliance.
4. To determine the difference between expected quality and perceived quality
determinations.
5
1.4 Thesis outline
This thesis consists of 10 chapters. In each subsequent chapter a brief introduction
outlines the content to follow.
Chapter 1
Chapter 1 provides a background to the study, justifying the contribution to industry
practitioners and the literature. The purpose of this first chapter is to introduce the topic
and outline the importance and objectives of the research as well as the structure of the
thesis itself.
Chapter 2
Chapter 2 provides a comprehensive review of existing literature pertinent to the
research, including the identification of gaps in the current body of knowledge specific to
consumer use of intrinsic and extrinsic cues and the consumer characteristics of
knowledge and self-confidence. The chapter concludes with the illustration of a
conceptual framework providing the basis of the investigation into the ability of extrinsic
cues to influence consumer quality judgment, both in terms of product expectation and
experience and quantifying the degree to which specified consumer characteristics alter
the reliance on these cues.
Chapter 3
Chapter 3 is devoted to the development of a causal model and the formalization of
hypotheses to be tested. From this theoretical foundation, the rationalization for the
adopted research design and adopted stages are provided. This chapter also includes
the methodology and the results of stage 1 of the research (largely qualitative data).
6
Chapter 4
Chapter 4 details the methodology and findings of stage 1 of the research, the
exploratory component which is comprised largely of qualitative data analysis. The
purpose of this stage of the study was to confirm the choice of product attributes and
stimuli for use in the quantitative components of the research.
Chapter 5
This chapter details the conjoint analysis methodology and measures of knowledge and
self-confidence used for the 3 quantitative aspects of the study. Chapter 5 also
contains the findings from a pilot study and their contributions to the subsequent stages
of the research.
Chapter 6
An overview of the rationale underpinning the methodology used to conduct the sensory
experiments and results of data validation tests are provided in this chapter.
Chapter 7
Chapter 7 provides the results of the sensory experiment data analysis and a
discussion of their implications.
Chapter 8
Chapter 8 is comprised of an overview of the methodology used to conduct the conjoint
analysis by survey and subsequent data analysis results specific to chardonnay. Also
provided is a comparison of the survey data analysis with the sensory data analysis for
chardonnay. The chapter concludes with a presentation of hypotheses testing
outcomes.
7
Chapter 9
Chapter 9 is comprised of the respective results for brie as illustrated for chardonnay in
chapter 8.
Chapter 10
This last chapter provides a discussion of the overall conclusions reached from
conducting the research and includes suggested implications for theory in this area of
study and for marketing practitioners. Identified limitations of the research follow and
recommendations for future results close the chapter.
1.5 Summary
This chapter presents the background, objectives and contribution of the study, and an
outline of the thesis is provided. The principal contribution of the research is a
theoretical framework illustrating consumer assessment of product quality based on
their evaluation of intrinsic and extrinsic cues incorporating the impact of potentially
moderating variables: consumer knowledge and self-confidence. The research will
investigate the ability of methodology measuring consumer expectations of quality
determined using product descriptions to predict respondent quality opinions based on
sensory experience thus providing insight in the differences between expected quality
levels and actual perceptions. This information will be of critical importance to
practitioners designing new products and developing marketing strategy.
8
2 Literature Review
2.1 Consumer use of intrinsic and extrinsic cues
It is established that consumers consider both types of cues when forming their overall
evaluation of quality; consequently it must be accepted that consumers are not always
totally objective when forming their opinions and perceptions of quality levels (Bredahl
2003). Olson (1972) conceptualized the formation of quality perception as a 2 stage
process. In Olsons model the importance of a cue to a consumer depends on its
predictive value, its confidence value, and whether or not the cue is intrinsic or extrinsic
to the product. Predictive Value is defined as the extent to which the customer
perceives or believes that the cue is related to, or indicative of, product quality.
Confidence Value is the degree to which a consumer is confident in his ability to
accurately perceive and then judge the cue. Literature shows that, overall, intrinsic
cues are given more credence; however, this will not be the case when intrinsic cues
are insufficiently predictive in the minds of consumers or when they have low
confidence in their ability to evaluate and assess them (Agrawal and Kamakura 1999;
Srinivasan, Jain et al. 2004). Extrinsic cues which are also believed to be strong
predictors of value (or offering high levels of emotional appeal) are brand name, price,
retail outlet and COO (Dodds 1991; Lin and Sternquist 1994; Lee and Lou 1996;
Gluckman 2001; Kardes, Cronley et al. 2004). For example, Vranesevic and Stancec
(2003) found that respondents evaluating branded tins of pt stated that they believed
a more prominent brand to be better quality than a non-branded offering, however, in a
blind taste test the premium brand of pt was not actually preferred.
9
Consumers with low levels of knowledge relevant to the product (or product category)
assessed attribute greater credence to extrinsic cues. For these individuals, intrinsic
cues are sometimes ignored because they are confusing and poorly understood; in
contrast, a category expert has the ability to correctly interpret and evaluate them
(Maheswaran 1994; Maheswaran, Sternthal et al. 1996; Kuusela, Spence et al. 1998;
Kardes, Kim et al. 2001; Siu and Wong 2002). In his study investigating the influence of
expertise on assessment of personal computers, Mareswaran (1994) found that
consumers with sound levels of knowledge were less likely to be influenced by extrinsic
cues than others when cues were clear, unambiguous and highly diagnostic in nature.
Dependence on extrinsic cues can also be situational, such as when intrinsic cues are
unavailable or when purchasing products associated with self-image or status
(Maheswaran 1994; Quester and Smart 1998; Piron 2000; Kardes, Cronley et al. 2004).
Wilson and Brekke (1994) found that when consumers are short of time or suffering
some other form of stress they will use the cognitive shortcut that an extrinsic cue
represents as they are not able to evaluate intrinsic cues appropriately. Consumers
also vary in their need for cognition and degree of ambivalence (or non-attitude)
towards product evaluation. Individuals with lower needs for cognition and/or higher
levels of ambivalence tend to make more use of the cognitive shortcuts afforded by
extrinsic cues because they are not motivated to understand intrinsic cues and are
seeking to make the easiest and quickest decision (Zhang 1996; Olsen 1999; Bredahl
2003).
Alba (2000) provides other important insights into consumer behavior relating to the
purchase decision. For example he suggests that 1 half of any given population of a
market is below average in intelligence and therefore not as capable of assessing
intrinsic product attributes, at least not with the same rates of success. Also, there may
not be the desire, or the will, to make wholly rational or logical decisions even by
consumers with high levels of intelligence. Alba (2000) uses the term mental
10
contamination to represent all environmental factors that contribute to distracting
consumers from applying logic in their decision making (p125).
2.1.1 Extrinsic cues and sensory perceptions
As discussed, the literature demonstrates that intrinsic cues are usually relied upon
more heavily in predicting and/or evaluating product quality. However, in some
situations, inherent product attributes are discounted by consumers in favor of extrinsic
cues they believe to be more reliable (Wilson and Brekke 1994; Hamilton, Knox et al.
2000; Wansink, Park et al. 2000; Vranesevic and Stancec 2003; Zellner and Durlach
2003). Studies have also demonstrated that extrinsic cues can be powerful enough to
overcome sensory perceptions. For example, Pechmann (1992) found in her study
involving consumer assessment of orange juice quality that respondents would favor a
lower quality juice if the price were relatively high, over a juice of lower quality if the
price were correspondingly low, provided they did not have the opportunity to assess all
juice samples simultaneously. Judgments of quality can also be influenced by
characteristics implied through labeling or simply only assumed by the consumer to
exist. Wansink (2000) found in her experiment testing the influence of labeling that
some respondents reported differences in the taste and texture of breakfast bars when
they believed the product contained soy. In fact, all products tested by these
respondents were identical and did not contain any soy-based ingredients. Visual clues
are also significant in consumer quality evaluations (Imram 1999). Data from studies of
consumer preferences in beef products indicate that consumers prefer the appearance
of very lean and red steak, believing that these attributes contribute to a better tasting
product. However, under blind taste testing conditions these respondents prefer the
taste and texture of meat that is darker in color (aged longer and more tender) and
more marbled (higher fat content means the meat is juicier) (Glitsch 2000; Bredahl
2003; Hurling and Shepherd 2003). In their experiment testing the influence of visual
11
cues, Hurling and Shepherd (2003) found that respondents tasting crumbed fish
reported patties with tidy golden crumbs well adhered to the fish tasted better than
those with pale crumbs or torn crumbing. Visual cues can also affect perceptions of
odors. Blackwell (1995) found that some respondents had difficulty identifying well-
known scents when the sample color did not match expectations, e.g. a green sample
scented with orange. Research has confirmed that sensory perceptions are not always
accurate and are vulnerable to expectations and beliefs. These findings mean that
marketers cannot assume that intrinsic product attributes will be weighted and
interpreted accurately when evaluated by consumers.
2.2 COO as an extrinsic cue
2.2.1 Introduction to the COO construct
COO has been defined in many ways in the marketing literature, but it is generally
considered to be the source country of a product. Country Image (CI) can be described
as the general perceptions, or stereotypical images that consumers from one country
(or region) form about another country or region (Han 1989; Han 1990; Papadopoulos
and Heslop 1993). COO effects describe the degree to which CI (or country of brand or
country of assembly) influences consumers' evaluations of products from a specific
country. Initial research studies commencing in the 1960s concentrated on the impact
of a single cue, the 'Made in' label. However the impact of CI tended to be overstated
when the COO cue was used in isolation and other factors likely to influence consumer
perception and buying behavior were not acknowledged (Bilkey and Nes 1982; Chao
and Rajendran 1993; Al-Sulaiti and Baker 1998).
The scope of research studies has expanded and evolved over the years and COO is
now recognized as a complex multi-dimensional construct that involves consumer
12
perceptions of source countries, recognizing the influence of other considerations such
as marketing mix variables and the specific characteristics of each target audience
(Ahmed, Johnson et al. 2002). Research has established that CI perceptions form the
basis of beliefs that consumers use as part of their product evaluation process, although
its importance will vary depending on market and product specific circumstances (Han
1989; Han 1990; Agarwal and Sikri 1996; Quester, Marr et al. 1996; Ger 1999;
Srikatanyoo and Gnoth 2002). For example, CI has been found to be more critical
when consumers are evaluating high involvement, high status or highly specialized
items such as designer clothing or prestige motorcars and less important in the
evaluation of low involvement, low priced items such as toothpaste or tee shirts (Manrai,
Lascu et al. 1998; Piron 2000; Srikatanyoo and Gnoth 2002; Ahmed and d'Astous 2004;
Ahmed, Johnson et al. 2004). The research results from Piron (2000) are an example
of the product/category specific nature of COO effect: COO had a stronger influence on
perceptions of the quality/desirability of a luxury item (sports car) than of a necessity
item (toothpaste). This suggests that consumers are looking for a more extensive set of
elements to evaluate before making a decision on high involvement items and also that
CI may make a status/ego enhancement contribution that makes the product more
attractive. The more recent research of Ahmed, Johnson et. al. (2004) supports Piron
(2000) regarding the influence of CI on low involvement products. In their study of
consumer evaluations of bread and coffee, they found that while COO influenced
expectations of quality, respondents considered price and brand to be more predictive.
While in this case CI was considered a weak determinant overall, it is interesting in
terms of cue usage as reliance was placed on other extrinsic cues over intrinsic cues.
The globalization of manufacturing has made it more difficult to determine the actual
impact of CI for many products (Han and Terpstra 1988; Choi 1992; Chao 2001; He
2003). The 'source country' for a product may now be different from its country-of-
brand or country-of-assembly or country-of-design. Studies have found that COO
13
effects vary in relation to these variables according to many characteristics such as
product brand strength, product type, level of competition and stage in the product life
cycle, thus adding to the complexity of understanding the influence of CI (Chen and
Pereira 1999; Chao 2001). These hybrid products, and product combinations,
represent new challenges for international marketers as multiple countries of origin
may require distinct marketing strategies across various market segments (Okechuku
1994; Kotler and Gertner 2002).
COO effect is a multidimensional construct, but there is no consensus on the number
and type of dimensions that have been used to measure it, making direct comparisons
of research results difficult, risky and confusing (Agarwal, Teas et al. 1997; Bhuian
1997; Quester, Dzever et al. 2000; Olsen and Olsson 2002). In their study of university
students in the US and New Zealand, Ford et al. (1999) reported that one of the
significant limitations of their study was the difficulty in transferring country and product
attribute statements across countries and cultures for interpretation purposes. Different
researchers questioning the validity of rating scales and respondent biases, evident
from questionnaire design and question order, have also raised other methodological
issues (Malhotra, Agarwal et al. 1996). An example is the controversy between studies
that ask respondents to give countries an overall country rating without product
associations, or those that ask for products to be rated according to where they are
sourced (Papadopoulos and Heslop 1993). Olsen and Olsson (2002) found that
different countries were rated differently when the order of scaling attributes was
changed, with the most consistent results evident when an attribute questionnaire
design was adopted over an entity based approach. Nevertheless, attributes such as
industrialization, areas of specialization (what the country is 'famous for' in a product
sense) and technological advancement are accepted as valid examples of
measurement dimensions.
14
In spite of the body of knowledge available on this topic, and acknowledged
methodological issues, research into COO effect continues for many reasons. The
world is a global market place and few businesses are immune to the influence of
imported products into home markets. Moreover, businesses around the world actively
seek export opportunities for their products and services, in developing countries
representing new consumer markets with enhanced purchasing ability (Kaynak,
Kucukemiroglu et al. 2000; Bandyopadhyay 2001). Being able to predict the
acceptance of, or bias against, their products in new markets can be critical to success
(Badri, Davis et al. 1995; he 2003). Identifying and exploiting competitive advantages
represents significant challenges to local and international marketers alike.
2.2.2 How country images are developed
The CI is the result of the combination of history, geography, political systems, famous
citizens, economic status and culture. In general, countries enjoy a favorable CI if they
are highly industrialized, economically strong; governed by democratically elected
governments and where citizens are well educated and enjoy a high standard of living
(Nagashima 1970). CI or product country image (PCI) profiles are akin to a 'brand
image' for individual countries (Nebanzahl and Jaffe 1996; Kotler and Gertner 2002;
Papadopoulos and Heslop 2002; Srikatanyoo and Gnoth 2002). These images allow
consumers to apply a consistent standard of quality to product categories without
personal experience with them because of the reputation that has been established;
examples include Dutch Chocolate, French Perfume' and Persian Rugs.
Countries that are new to exporting products may not have an established CI, positive
or negative, in the minds of targeted consumers overseas. Marketing strategists can
take advantage of this by creating and establishing a CI that benefits their product. For
example, the CI of Columbia was literally created in the American market in the 1970s
15
in order to sell coffee. A fictional character called 'Juan Valdez" was featured in a
commercial aired in America for years. The brand values communicated in the coffee
commercials came to represent the Columbian CI to the American market. This
strategy was successful because of the lack of an existing strong CI for Columbia in
American consumers minds at that time. Whilst beneficial for similar products or
product extensions this CI could represent barriers to the acceptance of non-agricultural
goods in the United States (Eroglu and Machleit 1988; Kotler and Gertner 2002;
Srikatanyoo and Gnoth 2002). For example, it is difficult to imagine CI used effectively
to promote German perfume as numerous studies have shown Germany to be
perceived as a country famous for engineering, while France is positively associated
with fashion and perfume products. Therefore, a German perfume manufacturer would
distance their product from Germany due to CI, whereas a French perfume
manufacturer would exploit it. Some country images are so well accepted world-wide
that foreign manufacturers are borrowing origins in order to reap the benefits of these
established stereotypes (Papadopoulos and Heslop 1993).
In spite of the enduring strength of established country images, countries are not
automatically locked into producing only one product category successfully. Countries
such as Japan and the United States enjoy a positive reputation around the world for
manufacturing and technology, resulting in a positive bias across a wide range of
products (Lin and Sternquist 1994; Leonidou, Hadjimarcou et al. 1998; Mohamad,
Ahmed et al. 2000; Chao 2001). Accordingly, even though a product and CI may not be
congruent, consumers will connect them and a shift in one will result in a simultaneous
shift in the other. An example is provided by Papadopoulos and Heslop (2002): "One
may think more highly of French than of Japanese fashion - but the higher one thinks of
Japan, the higher one thinks of its fashion products" (p296). This overall reputation built
over time for producing quality at a high level is a very valuable commercial asset
described as country equity (Papadopoulos and Heslop 2002).
16
2.2.3 Building country equity through international
brands
CI can influence perceptions of brands from different countries and the reverse can also
be true, with global brands contributing to the establishment of, or change in, CI.
Expanding on the previous example of Japans CI, consider the globally established
brands of Sony, Honda and Nikon. These brands are known around the world to be
Japanese; therefore, a sound relationship is established in the minds of consumers
between these brands and Japan. Consumers appear likely to take that relationship to
the next step and assume that all Japanese goods may be similarly high in quality
(Lampert and Jaffe 1997). Likewise, Samsung and Hyundai are examples of brands
effectively 'repositioning' Korea in the minds of western consumers (Srikatanyoo and
Gnoth 2002). This interaction contributes to increased awareness and strength for both
the COO of brand (or design) and the source country (Phau and Prendergast 2000;
Kleppe, Iversen et al. 2001; Insch and McBride 2004; Srinivasan, Jain et al. 2004; Phau
and Suntornnond 2006).
International marketers have exploited the concept of country branding by using the
attributes associated with eminent products or brands from a particular country and
incorporating them into products of their own. For example, advertising developed for
the Volkswagen Fox car manufactured in Brazil promoted German Engineering in
order to take advantage of the high quality benchmarks established by famous brands
such as BMW in Germany (Thakor and Kohli 1996). However, in evaluating product
country brand equity many marketing strategists fail to consider the contribution of CI to
overall brand values until such time as it starts to cause marketing problems. Research
conducted in the late 1990s, for example, led to a significant change in communications
strategy for the manufacturers of Jaguar motorcars. The Jaguar was always
considered a classic British product and marketing communications promoted the
17
strong link between the product and its COO, believing the relationship to be beneficial.
However, research revealed that CI and the desired product brand image were not
congruent. In order to establish a high-tech, high performance image in overseas
markets, links between Jaguar and England should be avoided, not endorsed, as the
well-established CI of England embodying tradition and conservatism was in direct
conflict with the desired product image (Kaynak, Kucukemiroglu et al. 1999).
Manufacturers reap significant benefits from country equity such as the opportunity for
product differentiation and the ability to demand higher prices (Kotler and Gertner
2002). Governments too, have identified and exploited opportunities to improve CI and
positioning. This is often accomplished through global events such as the Olympics,
Grand Prix Racing or other international sporting or entertainment activities. Similarly,
country equity can be quickly eroded by negative events such as wars or political
upheaval (Papadopoulos and Heslop 2002). Consider the current American perception
of Columbia; it would now incorporate the influence of drug cartels and civil unrest as
opposed to the single reference to a genteel coffee grower.
2.2.4 Consumer bias against products from developing
countries
Generally studies have shown that products from developed, industrialized nations are
preferred over products from newly developing nations, by consumers from both
developed and less developed countries alike (Chao 1992). Over time these countries
have accumulated country equity and their perceived product quality is expected to be
superior, embodying better reliability and product performance (Nagashima 1970; Bilkey
and Nes 1982; Lin and Sternquist 1994; Kaynak, Kucukemiroglu et al. 1999; Mohamad,
Ahmed et al. 2000; Chao 2001; Hui and Zhou 2003; Jo, Nakamoto et al. 2003). For
example, Smith (1993) found that products (cloth and a wine glass) were perceived to
18
be of lower quality when sourced from developing nations in Central America versus
Asia. Support for Smiths findings is provided by Lawrence, Marr et al. (1992) in their
study in New Zealand of consumer perceptions of imported cars, and Quester, Dzever
et al. (2000) in their study of purchasing agents' perceptions of machine tools and
component parts. In his study of consumers in Canada, Germany, the U.S. and the
Netherlands measuring their perceptions of products (TV sets and car radios),
Okechuku (1994) found that while there were differences in preferences between the
consumers in different countries in terms of their first choices (home country, the U.S. or
Japan), Mexico and Korea were consistently least favorably evaluated. These empirical
results are evidence that countries such as Japan and Germany were seen to possess
superior expertise in the areas of technology and design (even over other developed
countries such as Britain), leading to an assumption of higher quality for a variety
manufactured products.
The growing purchasing power of consumers in developing nations such as India and
Mexico have motivated researchers to study the influence of CI in these important new
markets (Leonidou, Hadjimarcou et al. 1998; Bandyopadhyay 2001; Bandyopadhyay
and Banjeree 2002; Delong, Bao et al. 2004). An extensive study of consumer
perceptions in India found that products sourced from Japan and Germany were ranked
higher in expected product quality and product image than products from Korea or
Britain, with Indian products ranked lowest (Bandyopadhyay 2001; Bandyopadhyay and
Banjeree 2002). This was based on respondent perceptions of educational levels,
economic development and technological expertise in those countries, with consumers
making a strong link between those country attributes and expectations of quality and
reliability. Data also indicates that the positive bias is even more pronounced in
developing nations because of the enhanced status that owning these products often
provides. These products often represent symbols of modernity, quality, advancement
and success to consumers in these markets (Papadopoulos and Heslop 1989; Zain and
19
Y. 1997; Kaynak, Kucukemiroglu et al. 2000; Kotler and Gertner 2002). The following
quote from a qualitative study undertaken by Ger (1999), illustrates the feeling of a 26-
year-old Turkish male in relation to products from the North America when compared to
products made in Turkey.
"If the name is English it is good, what comes from abroad is good." (p66)
Research of Bangladeshi consumer product preferences conducted by Kaynak,
Kucukemiroglu et al. (1999) found that products from Japan, Sweden, Italy, England,
the U.S. and Germany were considered superior to locally made goods or products
manufactured in India; with only the rank order of these countries changing according to
product category. Similarly, Mohamad, Ahmed et al. (2000) found in their study of
Malaysian consumer product preferences that imports from Japan, the United States
and European countries were favored over local goods or goods imported from newly
industrialized Asian countries such as Singapore, Taiwan and Indonesia in all
categories except clothing. In this category, Malaysian products were only considered
superior to products from Singapore and Thailand. Zain and Yasin (1997) studied the
effect of CI on consumer perceptions in Uzbekistan, and found that products from
industrialized nations such as Japan and the USA were considered significantly better
in quality across all product categories (clothing, white goods, cars and electronics)
compared to local goods and products from Russia, Turkey, China and India. The
argument supporting the link between the level of industrialization, development and
product category specialization and perceived product quality has been conclusively
made across most, if not all, product categories including those that would not be
considered 'hi-tech such as simple cloth and wine glasses. As a consequence,
marketers in 1st world countries can use this 'halo' of positive bias to enhance product
positioning world-wide.
20
However, research also indicates that in some cases the impact of CI is product type
and market specific as well as generally less important when consumers are evaluating
low involvement products (Chao and Gupta 1995; Piron 2000). This provides newly
industrialized countries with the opportunity to produce low priced, fast moving
consumer goods without prejudice. Empirical evidence also exists that may be used as
the basis for the development of strategies to overcome negative bias towards products
from such countries. For example, Niss (1996) surveyed the managers and owners of
100 Danish export firms and found that while it is useful at the introduction stage of the
product life-cycle to have a positive link between a product and CI, the impact of
associated imagery diminishes over time. Similarly, research conducted by Chen &
Pereira (1999) indicates that an unfavorable CI is less important to consumers
evaluating mature products or enjoying an extensive choice of competitive product
offerings. This suggests that manufacturers likely to suffer a negative COO bias may
mitigate the impact by taking a market follower approach to market entry, at a time
when consumers are already familiar with products in the category and are less likely to
rely on the CO cue.
2.2.5 The challenge of marketing 'hybrid' products
The globalization of manufacturing and assembly has made it increasing difficult to
easily attribute the origin of products to any individual country. In order to take
advantage of economic benefits such as low wages and lower taxation rates,
corporations are increasingly moving their factories 'off shore' to less developed nations
(Phau and Prendergast 2000). There are risks is such strategies as well as obvious
benefits because it has been established that perceived lower levels of technological
development leads to a negative CI bias against products from these developing
countries. While Japan and Korea are proof that it is possible to reverse detrimental
21
perceptions, it should be remembered that this was accomplished over an extended
time frame (Choi 1992; Hui and Zhou 2003).
Many products manufactured globally are designed in one country and manufactured in
another, often with components sourced from many different countries around the
world. The multiple countries of origin often make it difficult for consumers to
accurately identify the actual COO of a given product (Chao 1993). Hybrid or multi-
national products represent a new challenge for marketers because of the dissonance
in perceptions of quality that consumers (particularly in developed nations) may
experience. Consumers in some markets may feel comfortable and confident with a
well-known product brand, but at the same time, feel a degree of anxiety and increased
risk because of a negative country bias toward the country of manufacture of that
product (Choi 1992; Chao and Rajendran 1993). For example, in their study surveying
Thai and American students, Amine and Shin (2002) identified that products made in
different countries were perceived differently in terms of quality depending on the COO
of consumers. Their results suggested that Thai students considered a Sony Walkman
made in China to be lower in quality than one made in Japan, illustrating that even a
strong international brand such as Sony could not overcome a negative CI bias induced
by country of manufacture. In contrast, the American students perceived little difference
in the products, showing that each product and market combination may be unique and
stressing the potential risk in generalizing across market segments. Examples of
negative CI bias in the car industry include Israeli consumers refusing to buy
Volkswagens made in Brazil and car dealers in the US tearing identification stickers off
cars made in Mexico. Nebanzahl and Jaffe (1996) go on to show in their own study that
consumer perceptions of the quality of Sony and GE products deteriorated when a
proposed country of manufacture (Poland) scored lower on CI dimensions than the
country of brand (Japan and the U.S.). A critical symptom of the brand erosion was
consumer expectation of lower prices for products made in Poland. Chao (2001)
22
supported this with a similar outcome in his experimental study of American students
and their evaluation of TV sets and stereo systems. Results suggested that when
country of production and country of assembly was a less developed country (Mexico)
than the country of brand (U.S.), respondents expected product quality, and therefore
price, to be lower. These results alert marketers to the risk of cost savings in
manufacturing being offset by lower prices and increased distribution and transport
costs.
Brand strength appears to be critical in mitigating the negative effects of CI bias. Strong
international brands, such as Sony, have been found to be more resilient to the impact
of a negative CI of manufacture than non-branded or lesser-known brands (Haubl
1996; Jo, Nakamoto et al. 2003). A solution for managers of lesser-known brands may
be found in the research conducted by Li, Murray et al. (2000) who proposed that it is
possible to overcome consumer fears of lower product quality by providing superior
warranties and guarantees. This research supports the earlier studies undertaken by
Chao (1989 a & b) who found that CI effect can also be moderated by price and choice
of retail distribution outlet. In his later study, Chao (1992) reported that consumers may
make an important distinction in product quality according to the country of design and
pricing cues. Respondents considered a TV set to be superior in quality to alternatives
when designed in Japan even when the country of assembly was Taiwan (industrially
developed, but not associated with strong, international brands). Conversely, a TV set
designed in the U.S. or Taiwan relied on higher pricing to enhance expectations of
quality. Results from these studies suggest that choosing a country of design that is
strongly congruent with the product concerned and then supporting this high quality
position with appropriate pricing and distribution strategies could be an effective
strategy to overcome a negative CI bias.
23
While some powerful global brands have weathered consumer resistance to products
manufactured outside the country of brand and some market segments seem
unconcerned, the potential harmful effects of 'hybrid' products should not be ignored in
spite of identified strategies for addressing the impact. These considerations may be
particularly important for organizations with lesser-known brands or those entering new
markets (Kleppe, Iversen et al. 2001). Target audiences may not possess enough
knowledge (correct or incorrect) to conceptualize a CI and may struggle to form positive
country image and product associations (Beverland and Lindgreen 2002). For nations
lacking a strong and positive CI, leveraging the country of brand and/or design may
initiate the building of country equity critical for small or developing countries in
international markets. Over time, as in the case of Korea, strong home country
international brands may evolve to improve the CI for these less developed nations
(Papadopoulos and Heslop 2002). Until this occurs, however, marketing
communications will need to reassure customers that country of brand product values
will be maintained no matter where a product is manufactured, and the actual country of
production should be downplayed or omitted (Thakor and Kohli 1996; Delong, Bao et al.
2004). An alternative strategy put forward by Smith (1993) involves the use of regional
labeling rather than identifying specific countries.
2.2.6 Effect of market specific attributes on use of CI
The CI can vary in different market segments because of personal characteristics of
consumers such as religion, education, gender, age and other demographics (Kleppe,
Iversen et al. 2001; Amine and Shin 2002; Balabanis, Mueller et al. 2002; Kotler and
Gertner 2002; Insch and McBride 2004). In one of the earliest studies into consumer
profiles and the influence of CI, Schooler (1971) examined U.S. consumer perceptions
of products from India and Western Europe and found that older respondents rated
foreign products lower in quality than younger respondents, females rated imported
24
products more highly than males and respondents with higher levels of education rated
foreign products more favorably than those with limited education. Beaudoin, Moore et
al. (1998) investigated the impact of CI on perceptions of foreign apparel among young
female shoppers in the U.S. Those surveyed amongst the group considered to be
fashion leaders were significantly influenced by COO, while those considered to be
followers were not, highlighting the potential for differing levels of influence amongst
segments according to social position.
A study completed by Huang and Tai (2003) also found significant differences in
product preferences between female purchasers of cosmetics from four different Asian
countries due to cultural and economic differences while Insch and McBride (2004)
reported that age impacted on acceptance of foreign products. Cultural subgroups,
historical events, political forces, family authority and ethnocentrism also contribute to
the image of a country formed in the minds of consumers (Nebanzahl and Jaffe 1996;
Knight and Calatone 2000; Balabanis, Mueller et al. 2002; Huang and Tai 2003). For
example, the literature suggests that consumers show a preference for products from a
country culturally similar to their own (Harrison-Walker 1995; Watson and Wright 1999;
Balabanis and Diamantopoulos 2004). Balabanis, Mueller et al. (2002) took an
innovative approach to assessing cultural influences by using the dimensions of the
"Schwartz Value System (SVS)" (p 588) in their study. They hypothesized that cultures
with collectivist values would view foreign products less favorably than cultures with
individualist values; however in this instance the results were inconclusive. Heslop et
al. (1998) found that English-Canadian consumers showed a clear positive bias towards
products from Ontario and a negative bias against products from Quebec, reflecting the
historic cultural conflict between the English and French-Canadian subcultures.
Interestingly, in the same study a positive bias towards products from similar cultures
was not reported when the products were sourced from less developed countries,
25
suggesting the affinity felt towards a similar culture was not strong enough to overcome
the perceived risk that products from less developed countries may be lower in quality.
Consumers in some markets can harbor feelings of animosity towards people in
different countries, and because of these strong feelings of ill will customers may
boycott products. This is a different perspective of COO effect, because the image of
the source country does not act positively or negatively on perceptions of product
quality, but instigates a very negative reaction to the country itself resulting in a refusal
by consumers to buy. In this case, the CI is established because of major, negative
events, or due to entrenched perceptions of said events. A famous example of
consumer animosity was revealed by Klein (2002) who found that the citizens of
Nanjing, China (the site of a 1937 Japanese invasion that resulted in the death of
300,000 Chinese citizens) still felt such high degrees of hostility towards Japan that they
refused to buy imported Japanese products, in spite of considering them to be high
quality. Consumer backlash against French products after France commenced nuclear
testing in the Pacific in the 1990s is a more recent example in the Australian
marketplace (Herche 1994; Klein 2002). In order to overcome these negative feelings
by consumers (Clarke, Owens et al. 2000) recommends using a 3rd neutral country as
a distribution or assembly point, and to mark products as sourced from that neutral
country. However, this strategy could be considered unethical as it sets out to mislead
consumers. Alternatively, Smith (1993) suggests diffusing the negative perception of an
individual country through the use of regional labeling.
Consumers may also exhibit negative biases towards imports due to heightened levels
of ethnocentricity. Ethnocentric consumers feel obligated to purchase local (or home
country) products in preference to imports because they feel it is morally right to do so.
Ethnocentric behavior stems from feelings of national pride and/or perceptions of threat
by foreign imports to the local economy and employment levels (Klein 2002; Balabanis
26
and Diamantopoulos 2004). For example, in their study of New Zealand respondents,
Watson and Wright (1999) found that even those with low levels of ethnocentric values
were willing to buy a local refrigerator in preference to a technically superior refrigerator
from Germany even when prices were similar. Loyalty towards domestic products also
differs among ethnic groups. For example, African-American consumers are an
important growth segment in the U.S. car market accounting for approximately 12% of
new car sales. Ede and Panigrahi (2000) investigated the attitudes of this important
buying group towards cars imported from Japan and found that in spite of believing
Japanese cars to be equal or superior in quality to locally produced cars, African-
Americans believed that purchasing an import was unpatriotic because buyers of
Japanese cars were unconcerned about the U.S. economy. In these situations,
opportunities exist for marketers to reduce the advertising emphasis on intrinsic product
attributes (where the local product may actually be inferior) and exploit an extrinsic cue
that no foreign competitor can match or counteract.
However, whether or not consumers degree of ethnocentricity will consistently
influence their purchase behavior remains uncertain. For example, surveys of
Australian consumer sentiment suggest that a positive purchasing bias exists towards
products made in Australia. This has led to big investments in 'cuing' domestic
shoppers to "buy Australian". Quester et al. (1996) investigated whether shoppers
actually buy Australian products in preference to imports by analyzing the sales of fast
moving consumer goods. In an experiment conducted in an Adelaide supermarket,
labeling 'cues' were provided to shoppers to indicate which product was made in
Australia. This study determined that it could not be proved that people were more
likely to buy an Australian made product over an imported product, even when 'cued' to
this fact. Thus, doubt was cast on the likely positive impact of a campaign based on
products being made in Australia. Again, while these results may not be generalizable
across various markets, they demonstrate the difference between what people suggest
27
they will do and how they actually behave. This is an important consideration as many
of the studies into COO effect have also provided respondents with the opportunity to
rate home country offers. Domestic products have been favored in some studies over
imports, although these results appear highly dependent on numerous variables such
as market attributes, price differences, product category and the source countries of
alternative products (Watson and Wright 1999; Chao 2005).
These market specific variables highlight the multifaceted nature of the CI and how it is
formed by consumers in specific markets. Marketers, therefore, need to be well versed
in the demographic, historical, political and economic situations of the countries
targeted for market expansion. This is necessary in order to develop strategies to
counter negative bias or take advantage of positive bias.
2.2.7 Effect of CI on commercial customers
CI has also been found to influence the purchasing behavior of industrial consumers,
either confirming or disputing beliefs that commercial decisions are less 'emotional' and
more rational in orientation (Quester, Dzever et al. 2000). Results of research
undertaken by White and Cundiff (1978) and Crawford and Lamb (1981) found that
purchasing agents appear to use CI similarly to respondents in consumer studies.
Examples of similarities include a preference for products from industrialized nations
over products from less developed countries, and a stronger transference of CI to
product image when purchasers had little knowledge or experience (use of CI as a
'halo'). However, the number of 'business to business' studies conducted is low overall
and the range of products and services tested remains very limited representing
another significant gap in our understanding of potential CI influences (Andersen and
Chao 2003).
28
2.2.8 Effect of consumer knowledge on use of CI
When consumers have limited knowledge or personal experience of products from a
country, the CI appears to serve as a halo that forms the basis of consumers opinions
of products from a specific country (Han 1989; Han 1990). In this way, the CI serves
the purpose of a cognitive 'shortcut', allowing consumers to make a quick evaluation of
a product without having to search out an extensive set of elements (intrinsic and other
extrinsic cues) to consider (Maheswaran 1994). For example, if a consumer in Australia
has little knowledge of, or experience with, products from Mexico; these consumers are
likely to form their opinions of Mexican products based solely upon their CI of Mexico.
This is understandable because these images are all they can call upon when making a
judgment (Han 1989; Han 1990; Lee and Ganesh 1999; Kotler and Gertner 2002;
Srikatanyoo and Gnoth 2002). In these circumstances, the CI serves as the basis of
knowledge to fill the gap of the unknown.
Brand familiarity is a particularly important aspect of consumer knowledge with CI and
brand image linked in the minds of consumers (Phau and Prendergast 2000; Phau and
Suntornnond 2006). In her study of consumer perceptions of European beers from
different countries Schaefer (1997) tested the use of CI in product evaluation whilst
measuring respondent knowledge and brand familiarity. Results suggested that
intrinsic cues and brand familiarity have a significant effect on the use of the CI cue by
consumers. When brand names were unfamiliar, respondents showed a higher
reliance on CI than when they were known and the provision of intrinsic cues did not
diminish the use of CI in product evaluation. This is in contrast to Han (1990), who
found in his study of consumer evaluations of TV sets from the U.S. and Japan, that CI
plays a more important role in the evaluation of familiar rather than unfamiliar brands,
suggesting that a wider set of elements may be considered when consumers evaluate a
higher involvement product from 2 industrialized nations. The data from these 2 studies
29
may reflect consumers willingness to use a cognitive shortcut (CI) more readily when
evaluating a lower cost, lower involvement product such as beer.
When consumers are familiar with products and brands from different countries, CI
alone is less significant in product evaluation because the consumer has a more robust
basis of knowledge to rely on: their own experience. In this case, CI (relative to
products used) is established in the mind of the consumer as a result of this knowledge.
Han (1989) summarizes this alternative use of CI, country image becomes a construct
or file of information, which summarizes consumers beliefs about product attributes
and directly affects their brand attitude (p 228). However, brand and CI are both
extrinsic cues and less likely to be influential if a consumer is provided with predictive
intrinsic cues and has the knowledge to properly assess them (Okechuku 1994;
Nebanzahl and Jaffe 1996; Thakor and Lavack 2003). Therefore, consumer knowledge
(type and depth) has a direct impact on the role played by CI in driving the actual
purchase decision, but it will be product and market specific (Chao and Gupta 1995). A
more detailed discussion of this significant moderating variable to consumer cue usage
is provided later in this chapter.
2.2.9 COO summary
The research examined in this review was conducted in numerous countries around the
world, examining the COO effects on a broad range of products and product categories.
However, in spite of the diverse nature of the respondents surveyed, the research
methodologies used or the types of products researched, there are strong threads of
consistency throughout the results. Research has established that COO bias does exist
(in both a positive and negative way) and that it does influence consumer perceptions of
product quality (Lawrence, Marr et al. 1992; Bhuian 1997; Piron 2000).
30
However, the application of the CI cue and the degree of its influence ultimately
depends on the product type and usage being assessed, the number and type of other
cues available for assessment (intrinsic and extrinsic), specific customer characteristics
such as demographics, attitudes, beliefs (about product category and source country)
and level and type of knowledge (Han 1990; Maheswaran 1994; Papadopoulos and
Heslop 2002). Numerous studies conducted around the world have also established
that consumers evaluate products based primarily on intrinsic cues such as
craftsmanship, styling and performance, but also make use of extrinsic cues such as
price, prestige or endorsement. Other types of extrinsic cues such as retailer reputation,
promotional messages, guarantees and warranties are also used by consumers
because they reduce the risk of making a bad purchase decision, especially when
tangible intrinsic cues are difficult to evaluate. COO is considered one such extrinsic
cue and has generally been found to be more influential when intrinsic product cues are
unavailable or difficult to obtain (Szbillo and Jacoby 1974; Han and Terpstra 1988;
Herche 1994; Lee and Lou 1996; Acebron and Dopico 2000; Bredahl 2003; Jover,
Montes et al. 2004).
Further, consumers generally profess a preference for home products, although this
may not always translate into their actual purchasing behavior. Expected product
quality and perceived degree of industrial development are positively correlated. The
application of the COO cue and the degree of its influence is ultimately dependent on
the product type and usage being assessed, the number and type of other cues
available for assessment (intrinsic and extrinsic), specific consumer characteristics such
as demographics, level of product knowledge and experience and attitudes and beliefs
(about product category and source country). Competitive pressure means that
marketers today need to exploit every opportunity for competitive advantage and must
prepare strategies to deal with potential threats.
31
Different countries have become known for their specific areas of specialization and in
doing so, have accumulated country brand value or country 'equity' that can be
transferred and applied to products and product extensions in that product category
from that country. Examples include some Asian countries for miniaturization, Germany
for high quality engineering, and the U.S. or European countries for product styling
(Smith 1993; de Wet, Pothas et al. 2001). In general, highly industrialized and
economically developed nations such as Japan and Germany benefit from a positive
COO bias in both domestic and international markets in terms of product quality
perception, especially in categories requiring advanced forms of manufacturing.
Logically, this bias has been found to be less prevalent when consumers are assessing
everyday items or fast moving consumer goods. Likewise, consumers feel less risk
when purchasing high involvement or status products from industrialized countries and
may even pay a price premium for this reassurance. The opposite is also true, with
studies indicating that while some consumers may be willing to buy products from less
developed countries, there is a feeling that they should be able to buy them for less.
Consumers in both developed and less developed countries share this view fairly
consistently across the research reviewed. Brand manufacturers have the opportunity
to exploit this positive COO effect against competition from manufacturers from less
developed nations in both existing and new markets.
Consumers in less developed countries represent significant growth opportunities for
manufacturers around the world. At the same time, marketers in less developed
nations find themselves at a distinct disadvantage in both local and global markets.
Suggested strategies to combat a negative CI include the use of regional labeling rather
than a made in cue to overcome a negative individual CI. Local manufacturers may
also be able to improve perceptions of local product quality and break into new markets
through the implementation and promotion of internationally recognized quality
standards or the provision of enhanced warranties and guarantees. Another strategy
32
could be the use of origin borrowing through the use of western style branding.
Penetration pricing is another option that could be adopted to gain entry and establish a
customer base. An opportunity may also exist for business people in developing
nations to actually exploit strong CI and product associations in overseas markets by
specializing in those products unique to their country and expertise, in spite of a
generally negative CI. For example, artisans in Iran and Iraq possess distinctive
abilities in the production of silk carpets while the Cubans are famous for their fine
cigars (Kaynak, Kucukemiroglu et al. 1999).
Korea has demonstrated that countries can 'reposition' themselves over time by
establishing one or two strong international brands or by exploiting international events
such as the Seoul Olympics. In the shorter term, this could be accelerated by the trend
of global brand manufacturers such as Sony and HP producing products in these
countries. Research indicates that by making products in developing countries, brands
may actually be damaged in terms of product quality perceptions and perceived value
(price point); but this may only have short-term impact and some international brands
may be strong enough to withstand this influence. However, it sets up an interesting
potential conflict between host country and global brand manufacturers. A strategy for
these manufacturers would be to emphasize established product brand values,
reassuring consumers that these remain intact no matter where a product is made, or
alternatively omit or downplay in their communication any mention of country of
manufacture (Phau and Prendergast 2000).
Not all country images, even those that appear well established and positive, can be
used to enhance product positioning. For instance, New Zealand could be said to have
a positive CI in terms of agricultural products, but that CI could work against
manufacturers of high-tech products attempting to break into overseas markets against
competitors from Japan and the U.S. In this instance, CI needs to be de-emphasized
33
and individual brand attributes championed. Marketers also need to understand that a
positive connection between CI and a product category can erode over time. The need
for a major strategy shift in marketing communications for 'Jaguar' motorcars illustrated
this point. Previously marketed as a prestige British car, the product is now distanced
from its COO in order to escape the stereotypical CI of English conservatism and
tradition.
Market specific variables that affect the use of the COO cue such as demographics,
ethnic background and culture are beyond the influence of marketers and changes
occur over extended periods of time. Similarly, consumer attitudes such as
ethnocentricity are difficult to influence in the short to medium term. In general, COO
bias appears to be stronger when the consumer has little other knowledge to evaluate
alternatives: the greater their level of product knowledge, the lower their reliance on
COO for product evaluation (Han 1989; Han 1990). Consumers have also been found
to put more weight on intrinsic cues (or tangible attributes) of the product being
evaluated than on extrinsic cues such as CI. However, empirical evidence also exists
to demonstrate that even experts, or individuals with relatively high levels of knowledge,
may consider COO to be an accurate predictor of product quality (Schaefer 1997).
Therefore, marketers entering new markets need to research the characteristics of
potential customers in order to develop strategies to overcome or exploit potential
biases.
In summary, while COO has been established as an influence on consumer product
evaluations the significance of the CI appears to vary according to individual marketing
situations. This means that CI should be used to provide additional marketing leverage
to enhance product acceptance, or considered as a negative bias to be countered with
sound marketing strategy.
34
2.3 Price as an extrinsic cue
2.3.1 The role of price in consumer buying decision
making
The role of price in consumer evaluation of product alternatives and their ultimate
buying decision is a multi-dimensional one (Monroe 1982; Erickson and Johansson
1985). Consumers consistently use price as a predictor of quality, particularly when
they have limited knowledge of product category offerings (Monroe 1976; Dickson and
Sawyer 1990; Manrai, Lascu et al. 1998; Glitsch 2000; Bredahl 2003; Jover, Montes et
al. 2004; Kardes, Cronley et al. 2004). Consumers have been found to believe that
there is a natural ordering of products according to a price scale, where the higher
quality products are more expensive and products of lesser quality are cheaper. This
bias in consumer beliefs can result in 2 negative outcomes for marketers. In the first
instance, consumers may deduce (often incorrectly) that a high quality product is more
expensive that it actually is, leading to a constraint on their likely purchase or a value
pricing strategy can lead to consumer assumptions of lower than actual quality
(Erickson and Johansson 1985). Therefore, this belief can also provide an opportunity
to underpin a quality proposition for their product, as a higher price is likely to lead to
assumptions of higher quality even if unwarranted. This subjective view of price leads
to its additional role in the buying decision process as specifying a conditional trade off.
If consumers believe that price and quality are tied in, then paying a lower price means
they must accept lower quality. Conversely, to gain better quality, a monetary sacrifice
must be made, perhaps beyond that which is desirable to the payer (Rao and Monroe
1988; Rao and Olson 1990; Kardes, Cronley et al. 2004). As suggested, for many
consumers with limited financial resources, price may act as a constraint reducing the
number product or service alternatives that can be considered (Snoj, Pisnik et al. 2004).
35
Price can also be considered a means for reducing risks; these may include increased
search endeavors, additional waiting time (if a cheaper product is not currently
available) or loss of social status or embarrassment (Snoj, Pisnik et al. 2004). For
example, Quester and Smart (1998) investigated how situation and usage affected
purchases of red wines and found that consumers tended to be more price sensitive
when buying wine for their own consumption than if they were buying for a special
occasion or an important gift. In these instances, a premium wine is more likely to be
chosen in order to match the standing of the occasion, mitigating the risk of buying a
product of poor quality and exploiting the personal status value of an expensive and
well-known brand. In these situations, a higher price is justified in the minds of
consumers due to the increased value of the product and the security that is felt as a
result. Finding an acceptable balance in the sacrifice that will be made and the value
received clearly represents an important challenge for many consumers.
Price is not always proven to be strongly linked to perceptions of product quality; other
considerations such as retail store image and point of purchase information have been
found to diminish or override price point differences (Erickson and Johansson 1985;
Dodds 1991; Wakefield and Inman 1993). This means that the influence of price as a
product cue is usually more powerful when little else is known about the product, and is
reduced when consumers have more extensive information on which to base a
decision. Further, Dickson and Sawyer (1990) found that consumer recall of prices was
largely over-assessed. In their research of respondents recall of prices paid for
supermarket goods, they found that immediately post-purchase almost half could not
remember what they paid for an item purchased or remember if it was purchased on
special. This suggests that price is not usually retained in long-term memory, but only
considered when making a purchase decision, then purged. This data supports early
work by Zeithaml (1988) and Conover (1986) signifying that consumers have low levels
of current and accurate knowledge regarding many consumer goods. Brand positioning
36
is powerful in this context as it will trigger an expectation of price, hence quality, in the
minds of consumers (Monroe 1976; Dodds 1991; Bredahl 2003). Prices may also be a
reflection of availability of supply (intended or otherwise), where a product may
command a significantly higher price if it is more difficult to obtain (Horowitz and
Lockshin 2002). Indeed the power of price is linked to the additional information
available for consumers to consider, making price a powerful tool to support other
provided product cues to strengthen positioning strategies.
2.3.2 Effect of consumer knowledge on use of price
Considered a reliable predictor of quality for most consumers, price has been found to
be one of the most strongly weighted extrinsic cues, particularly when intrinsic cues are
not available, or when they are not understood (Kardes, Cronley et al. 2004). Therefore
when consumers are unfamiliar with a product category and lack objective knowledge,
price will be very significant in determining their expectations of quality. In some
instances it will even overcome experienced intrinsic cues (Pechmann and Ratneshwar
1992). Consumers with low levels of knowledge find it particularly difficult to assess
quality if intrinsic cues are complex, leading consumers to become intimidated by
price, according to Jover, Montes et al. (2004) in their study measuring the impact of
extrinsic variables on consumer expectation and evaluation of wine quality. Thus, as
with CI, consumers with sound levels of knowledge will usually use price when it is
legitimate to do so (e.g. there is a strong relationship between price and intrinsic
product quality), and/or when intrinsic product information is limited.
2.3.3 Price summary
The degree to which price influences product quality perceptions is dependent upon the
product type, usage, situation and the type and number of other cues available for
37
evaluation. Consumer attributes, such as knowledge, will moderate the evaluation of
extrinsic cues and buyers with an accurate understanding of sufficient intrinsic cues
should use them as the primary basis of any decision. Conversely, if intrinsic attributes
are not available or are not understood due to ambiguity or a lack of knowledge, then
price is likely to be used as the primary surrogate indicator of quality.
2.4 Consumer expertise and self-confidence
2.4.1 Two dimensions of knowledge
Consumer expertise is comprised of two dimensions, objective knowledge and self-
assessed or subjective knowledge. Objective knowledge is accurate and current
information stored by an individual in their long-term memory. This type of knowledge is
based largely on cognitive type learning and experience with the product category via
instrumental learning (Alba and Hutchinson 1987). Subjective knowledge is the
consumers perceived level of expertise or self-assessed level of knowledge, more
accurately described as product class familiarity. This is usually based on personal
experience, exposure to marketing information and recommendations (Monroe 1976;
Brucks 1985; Park, Mothersbaugh et al. 1994; Wirtz and Mattila 2003). Consumers will
naturally have differing levels of knowledge about different categories of products, and
their levels of knowledge, both objective and subjective, may vary over time. For
instance, an individual can develop a high level of objective knowledge in relation to
cars due to a personal interest or because of research conducted prior to a purchase
decision, but if not constantly maintained with current information, this knowledge will
not remain current or reliable. Therefore, expertise developed as a result of objective
knowledge should not be confused with product familiarity or past experience alone
when assessing knowledge levels. This leads to consumers consistently over-
assessing their levels of expertise, creating a gap between their own perception of what
38
they believe to be true regarding product offerings and an accurate judgment. Empirical
evidence has established that consumers, generally, do not possess the level or quality
of objective knowledge they believe they do (Alba and Hutchinson 1987; Heimbach,
Johansson et al. 1989; Alba and Hutchinson 2000; Alba 2000).
Consumers with differing types and levels of knowledge adopt differing approaches and
intensity to information search and processing. Consumers with high levels of objective
knowledge will seek limited information regarding a specific product or category
because they already have a credible source of information readily available in their
own memory. As a result, they will only seek to update and confirm what they already
know. Consumers with low levels of objective knowledge are also likely to restrict their
search; but in their case it is because they believe they know enough already or are
comfortable using one or more sources to make a decision. Consumers with moderate
levels of objective knowledge undertake the most extensive search because they
recognize the gaps in their understanding and seek to address them in order to make a
good decision and buy with confidence (Bettman and Park 1980; Park and Lessig 1981;
Rao and Olson 1990). The resulting differences in search behavior results in an
inverted U pattern when illustrated graphically, where low levels searched are
conducted by consumers with low and high levels of objective knowledge (representing
opposite ends of the spectrum), and a more extended search behavior undertaken by
those with moderate levels (Park, Mothersbaugh et al. 1994). Considering these
differing behaviors and patterns, it is not surprising that many consumers often
misjudge product quality through limited searches and erroneous interpretation of both
intrinsic and extrinsic cues.
While there is a considerable body of knowledge in this area, research concerning how
consumers use knowledge has generated conflicting and ambiguous results. This is
due to inconsistencies in methodologies and definitions of what knowledge is. For
39
example some researchers measure only familiarity or self-assessed knowledge and
deem this objective knowledge for the purpose of the study (Spence and Brucks 1997;
Alba and Hutchinson 2000). Since consumer knowledge (type and level) is an
important dimension in understanding consumer buying behavior, there is clearly a
need to understand how different types and levels of knowledge impact on consumer
search patterns and information gathering and processing relevant to extrinsic and
intrinsic cues.
2.4.1.1 Objective knowledge
Consumers with high levels of objective knowledge have been found to distinguish
more easily and more precisely between important product and service attributes,
disregarding those product characteristics that are less critical to making a sound
buying decision. They have developed the ability to efficiently search out and
accurately filter new pieces of information, be they intrinsic or extrinsic cues, due to
enhanced diagnostic skills. These are then efficiently coded and compared against the
information about previous alternatives already stored in long-term memory. As a
result, consumers with high levels of objective knowledge use a large and detailed pool
of information when assessing products (Brucks 1985; Park, Mothersbaugh et al. 1994;
Mason and Bequette 1998; Kardes, Kim et al. 2001; Wirtz and Mattila 2003). A product
expert drawing upon this accumulated knowledge is less likely to seek advice from
others; they are also less likely to place much credence on advertised product benefits
when gathering information. Instead, they seek to understand critical attributes, making
their own judgments regarding any resulting benefits (Kuusela, Spence et al. 1998).
This allows them to match particular product brands and models to specific usages.
This was demonstrated by Maheswaran, Sternthal et al. (1996) in their study
investigating the application of knowledge to the correct choice of a sewing machine
and it was determined that respondents with high levels of objective knowledge were
40
much more successful in correctly choosing the right sewing machine for the job to be
done than those with either low levels of objective knowledge or high levels or
subjective knowledge. Further, these women could name a more extensive range of
brands and models, identifying and describing the expected extra features to be found
on a more expensive, high quality brand as opposed to a budget model. While the
literature highlights the importance of objective knowledge in the accurate assessment
of products, this discussion does not preclude the inclusion of familiarly and experience
in the development of consumer expertise. Cowley (1994) suggests that a true expert
has knowledge based on extensive personal experience, category familiarity and
objective knowledge. The critical addition to the determination of expert is a high level
of objective knowledge as it tempers and moderates the effect of the information gained
through the other 2 learning methods.
Knowledge at this level leads to superior confidence in the final purchase decision,
resulting in lower levels of brand loyalty amongst experts. The literature suggests that
this is because they are more likely to seek out up to date product information and then
more likely to use it objectively in evaluating product alternatives. As their base of
knowledge comprises external information as well as product experience, the decision
making process involves more comprehensive information and the consideration of a
broader range of alternatives (Alba and Hutchinson 2000; Wirtz and Mattila 2003). The
logical application of information relevant to product performance results in a bias
towards intrinsic cues. Brand, like price or CI, is an extrinsic cue and may thus be
discounted in favor of a better product, even if the alternative brand is new or not as
highly positioned in the market place. Extrinsic cues, however, are not discounted if
they are truly predictive of quality (Heimbach, Johansson et al. 1989; Rao and Olson
1990; Andreassen and Lindestad 1998). Alba and Hutchinson (1987) found in their
early research that even true experts can still be influenced by biases if those biases
are felt strongly enough, leading to improper weighting of both intrinsic and extrinsic
41
cues. Chiou (2003), in his recent study of the impact of CI on perceptions of product
quality, found that respondents with high levels of pre-trial objective knowledge favored
digital cameras from Japan over those from Taiwan and cloth from Italy over that from
Taiwan. However, in post-trial product evaluations, these experts favored the superior
product over the inferior product regardless of country of origin. Interestingly, when
comparing the superior product ratings between cells, when the better product was
sourced from Japan and Italy the products rated better overall than when the (identical)
superior product was believed to be sourced from Taiwan. Therefore, even though the
better product was chosen by the experts it still suffered some bias in comparative
terms. While this supports the emphasis on intrinsic cues in actual product evaluation,
it also highlights the power of a predictive extrinsic cue in influencing expectation pre
and post-trial.
2.4.1.2 Subjective knowledge
In contrast to consumers with high levels of objective knowledge, those relying on
subjective knowledge lack an extensive collection of pertinent information upon which to
draw. They can usually only recall a few brand names and models, and perhaps only 1
or 2 specific attributes about each (Mitchell and Dacin 1996). Consumers with high
levels of self-assessed knowledge have been found to use their own experiences
(however limited) as the basis for their expertise and who limit their external search for
up-to-date information, believing they already know enough (inverted U pattern
discussed previously). However, the empirical evidence strongly suggests they usually
know much less about products than they believe (Alba and Hutchinson 2000).
Some consumers acknowledge their low objective knowledge levels, but because they
have a low need for cognition and/or they are quite ambivalent towards the purchase
decision, they exhibit no desire to learn more (Zhang 1996; Olsen 1999). For them, a
42
limited information search will be restricted to seeking opinions from others or using
readily recalled marketing messages that highlight product benefits (Maheswaran,
Sternthal et al. 1996; Till and Busler 1998). For this group, therefore, the cognitive
shortcut provided by extrinsic cues, such as price and CI, is especially attractive.
Consequently, unlike experts, these types of consumers find it much more difficult to
correctly match the correct brand or model with a specific usage situation (Brucks 1985;
Park, Mothersbaugh et al. 1994; Wirtz and Mattila 2003).
A large body of evidence supports the influence of extrinsic cues when objective
knowledge is low, or not considered necessary by consumers (Monroe 1976; Brucks
1985; Sullivan and Burger 1987; Harrison-Walker 1995). For example, in his study of
consumers assessing financial institutions Devlin (2002) found that those with high
levels of objective knowledge put emphasis on intrinsic attributes such as interest rates
and fees, whereas high subjective knowledge respondents were more concerned with
testimonials and retail location. Maheswaran (1994) in his study measuring the impact
of CI on consumer preferences for personal computers, reported that while all
respondents were relatively familiar with personal computers through experience, only
those with strong objective knowledge (IT students) could properly assess the
significance of described intrinsic attributes. Respondents using subjective knowledge
placed a significantly stronger reliance on the CI cue in their assessment of expected
quality. It appears that individuals relying on subjective knowledge are not able to filter
out the attributes which are unimportant to performance. Instead, they filter out those
they do not understand, and use extrinsic cues such as brand, price or CI to fill any
gaps (Cordell 1992; Schaefer 1997). Customers in this situation tend to be more brand
loyal than experts. This is because product switching requires a renewed information
search and subsequent evaluation of alternatives, unattractive to these individuals due
to their lack of motivation and/or ability to correctly evaluate cues. Given that there are
relatively few true experts in most consumer markets, the importance and weight given
43
to extrinsic cues cannot be underestimated (Alba and Hutchinson 1987; Alba and
Hutchinson 2000; Alba 2000).
2.4.2 Consumer self-confidence
Consumer self-confidence levels have been found to affect the interpretation and use of
both intrinsic and extrinsic cues (Wilson and Brekke 1994; Wansink, Park et al. 2000;
Bearden, Hardesty et al. 2001; Jover, Montes et al. 2004). Individuals with low levels of
self-confidence may lack self-belief to the point, where if faced with a strong opposing
opinion or predictive extrinsic cues, they will allow their better judgment to be
overridden. This may even occur when an individual is a true product expert in a
specified category. Alternatively, consumers with high levels of self-confidence possess
strong product attitudes that are very difficult to change due strong self-belief (Olsen
1999). This strength of conviction leads them to hold on to their beliefs irrespective of
support by others or legitimacy (Rao and Olson 1990). Therefore, although these
individuals believe themselves to be experts, it is more likely their knowledge is only
subjective. As this variable has not been measured concurrently with objective and
subjective knowledge in previous studies, it may be another contributing factor to
inconsistent and conflicting results (Bell 1967; Fazio and Zanna 1978; Bearden,
Hardesty et al. 2001).
An early study by Bell (1967) illustrates some of these points. In his research, he
investigated consumer confidence levels in relation to new car purchases. His work
revealed that individuals with higher self-confidence levels used the opinions of friends
less, as they had a stronger belief in their own ability to decide. Those with less
confidence, on the other hand, used the opinions of friends and adopted buying teams
to make purchasing decisions. Bell (1967) also found that the ability to be persuaded
had nothing to do with intelligence, as smart people were as easily persuaded as less
44
intelligent respondents, however, self-confidence was found to be highly correlated with
high self-belief, often leading to stubbornness. Interestingly, people with low self-
confidence could become stubborn also, but it was because they became defensive
under the pressure of decision making, not because they necessarily believed they
were right (Bell 1967). People with moderate levels were the most easily influenced as
they were neither defensive nor overly confident, but adopted a more balanced
approach.
Therefore, the literature suggests that the particular combination of knowledge (type
and level) in conjunction with self-confidence levels significantly moderates credence in
extrinsic cues. If a highly confident person also has high levels of objective knowledge
and is presented with relevant intrinsic product cues, it would be expected that extrinsic
cues would be discounted in quality assessment. However, if knowledge is primarily
subjective resulting in an inability to correctly interpret the intrinsic cues, they may be
ignored or misjudged. For the low confidence person, high levels of objective
knowledge should enhance and support their judgment, but as discussed, this may not
be the case. For the low confidence consumer with low objective knowledge levels,
extrinsic cues may well represent the principal basis for many product comparisons.
2.4.3 Consumer knowledge and self-confidence
summary
Consumer knowledge and self-confidence are established as important moderating
factors in how consumers use intrinsic and extrinsic cues to form opinions about
products both pre and post-trial (Maheswaran 1994; Schaefer 1997; Chiou 2003; Jover,
Montes et al. 2004). Consumers may feel confident that they can accurately assess or
predict quality, but this assumption is often based primarily on product experience or
category familiarity. Indeed, research shows that the majority of consumers are not true
experts in terms of objective product knowledge, making accurate quality
45
determinations difficult. Given that high levels of objective knowledge are not evident
amongst the majority of consumers, and that high levels of self-confidence cannot be
assumed, the importance of understanding the power of extrinsic cues in actual product
evaluation, and on the formation of product quality expectations, cannot be
underestimated (Bell 1967; Fazio and Zanna 1978; Alba and Hutchinson 1987; Alba
2000; Bearden, Hardesty et al. 2001; Moorman 2001).
2.5 Gaps in the existing literature
It is established that consumers use both intrinsic and extrinsic cues to form opinions
about expected and experienced product quality (Lee and Lou 1996: Quester and
Smart 1998: Piron 2000). Two extrinsic cues found to be used consistently in this
process are CI and price (Han 1989: Han 1990; Zeithaml 1988). However, the ability of
CI and price to override sensory perceptions of quality has not been specifically tested.
Moreover, while consumer knowledge and self-confidence are known to moderate
reliance on extrinsic cues, results of previous studies are sometimes ambiguous and
conflicting due to inconsistencies in definitions and measurement (Park and
Mothersbaugh et al. 1994). Often studies measure subjective and/or self-assessed
knowledge, and levels of objective knowledge are not established. Or alternatively,
consumer self-confidence has not been measured in combination with these 2
variables, despite its importance as a moderating variable, given its effects on peoples
beliefs in their ability to make an independent and sound decision. Consumers lacking
self-confidence are less likely to support their own opinions if they contradict those of
others or predictive extrinsic cues (Bell 1967; Alba and Hutchinson 1987; Aaron, Mela
et al. 1994; Alba 2000). While empirical evidence exists regarding aspects of these
specific variables, these gaps in the literature remain.
46
2.5.1 Towards a conceptual framework
This discussion provides the basis for a conceptual framework articulating the possible
relationships between intrinsic and extrinsic cues and consumer judgments of product
quality (either experienced or expected). Product attributes are represented in the
model (Figure 2.1) by intrinsic and extrinsic cues. The degree of reliance placed upon
the attributes (intrinsic or extrinsic) in determining product quality is moderated by
consumer specific characteristics, such as expertise and self-confidence, although to
what degree, is unclear.
Expertise Consumer
(subjective self-
and objective confidence
knowledge)
Evaluation
of intrinsic
cues Relative
influence of
cues type on
perception of
Evaluation quality
of extrinsic
cues
Figure 2.1 Conceptual framework - effect of consumer characteristics on assessment of

product quality
The figure provides the framework of investigation into the ability of extrinsic cues to
influence consumer quality judgment, both in terms of product expectation and
experience; and further, to quantify the degree to which specified consumer
characteristics alter the reliance on these cues.
47
2.6 Summary
Chapter 2 summarized key aspects of the literature in areas relevant to the research:
consumer use of intrinsic and extrinsic cues (price and CI) and the moderating effects of
consumer expertise and self-confidence. This review illustrates that extrinsic cues can
be very powerful in consumer assessment of both expected and experienced quality.
Figure 2.1 provides a conceptual model articulating the expected relationships between
these variables and consumer determination of product quality, with identified gaps in
the literature providing support for further investigation.
48
3 Causal model and hypothesis development
3.1 Introduction
Chapter two provided a review of the literature relevant to the respective contributions
of intrinsic and extrinsic cues to consumer product quality evaluation, and discussed
how the moderating variables of knowledge and self-confidence can influence the use
of and reliance upon these cues. From this, a conceptual model articulating possible
relationships between the variables was provided. In chapter 3, the tested hypotheses
are formalized and summarized. Evolving from this theoretical foundation, the
rationalization for the research design and adopted stages are provided. The next
sections provide an overview of methodology and the results of stage 1 of the research
(qualitative data).
3.2 Research framework
Prior to finalizing hypotheses and determining methodology, it is useful to summarize
the key theoretical aspects and state, in broad terms, the overriding research questions.
3.2.1 Testing the power of extrinsic cues
It is established that consumers use both intrinsic and extrinsic cues in forming opinions
about product quality, both expected and experienced, and that COO and price are two
types of extrinsic cues influencing these consumer determinations (Brucks 1985;
Acebron and Dopico 2000; Bernues, Olaizola et al. 2003; Bredahl 2003). As an
extrinsic cue, COO has been studied intensively for its effect on consumers' product
evaluation since the 1960s. COO is generally described as the 'source country' for a
product or service provider, which may be different from the country of brand or country
49
of assembly or design. The literature also shows consistent support for the view that
consumers believe in price as an accurate predictor of quality across a wide range of
products, even when experienced product quality has been contrary to pricing level
indicators (Sullivan and Burger 1987; Pechmann and Ratneshwar 1992; Jover, Montes
et al. 2004).
3.2.2 Testing the influence of consumer
characteristics
Consumers are not always able to accurately evaluate cues before making a buying
decision. Research has also shown that consumers are not always successful in
judging quality accurately even in a post-purchase evaluation (Alba 2000; Kardes, Kim
et al. 2001). There are a number of reasons this may occur, including lack of
understanding, lack of self-confidence, information misinterpretation or inaccessibility
(Roper 1969; Olson 1972; Szbillo and Jacoby 1974; Monroe 1976; Rao and Olson
1990; Northen 2000; Teas and Agarwal 2000; Wansink, Park et al. 2000; Siu and Wong
2002; Kardes, Cronley et al. 2004). Specifically, consumer knowledge (subjective and
objective) is known to moderate the use of, and reliance upon, extrinsic cues.
Consumer self-confidence is also an important moderating variable as it will affect an
individuals belief in relation to their ability to make a sound decision independently.
Consumers lacking self-confidence are less likely to support their own opinions if they
contradict those of others or strong extrinsic cues (Bell 1967). Specifically, the study
will seek to answer the following research question:
1. What are the relative influences of the extrinsic cues of CI and price on product
quality judgments, experienced and expected?
a. Are these influences moderated by consumer knowledge
(type and level)?
50
b. Are these influences moderated by consumer self-confidence?
3.2.3 Use of sensory experiments
The objective of the research is to quantify the influence of COO and price on consumer
opinions of quality when product intrinsic cues are actually experienced, not merely
described, to determine if experienced quality can be overwhelmed. This was
accomplished by analyzing data collected through an experimental design involving
taste testing. Previous studies have been conducted to assess the influence of price
(among selected other extrinsic cues such as labeling) by means of sensory evaluations
such as taste tests and visual impressions (Pechmann and Ratneshwar 1992; Wansink,
Park et al. 2000; Hurling and Shepherd 2003). However, the influence of COO as an
extrinsic cue does not appear to have been examined previously using conjoint analysis
and taste testing methodology (Roper 1969; Pechmann and Ratneshwar 1992; Aaron,
Mela et al. 1994; Imram 1999; Acebron and Dopico 2000; Hoffmann 2000; Becker
2000a; Koch and Koch 2003; Zellner and Durlach 2003). Further, the moderating ability
of consumer knowledge (type and level) and consumer self-confidence have also not
been investigated specifically in studies testing these extrinsic cues in conjunction with
sensory perceptions. Choice of this methodology represented another significant
opportunity to add to the current knowledge in this area (Wilson and Brekke 1994;
Schaefer 1997; Wirtz and Mattila 2003).
3.2.4 Suitability of conjoint analysis experimental
design
Conjoint analysis is a well established multivariate technique, used extensively in both
psychology and marketing research, for determining the importance consumers place
on product attributes (Green and Srinivasan 1978; Green and Srinivasan 1990;
Okechuku 1994; Hair, Anderson et al. 1995; Lee and Lou 1996). As the name implies,
51
this method involves the measurement of respondents psychological judgments of
perceived similarities or differences between different product profiles by studying the
joint effects of these attributes on consumer evaluation or product preference (Hair,
Anderson et al. 1995). The technique estimates the structure of consumers
assessment of product alternatives comprised from a set of pre-specified levels of a
range of attributes (product cues). It is based on the simple premise that consumers
evaluate a product overall in a holistic manner by combining the separate amounts of
utility (value or attractiveness) provided by each attribute level (Jaeger, Hedderley et al.
2000).
A high degree of flexibility in conjoint analysis design permits respondents to consider a
wide range of attribute combinations (product profiles) thus affording a high degree of
market realism within a controlled experimental design (Hair, Anderson et al. 1995;
SPSS-Inc 1997). Respondent self-report bias and carry-over effect is minimized by
presenting respondents with a set of choices without making explicit the fundamental
attributes under study (Henderson and Reibstein 1985). Further, there are few
constraints in terms of attribute types that can be tested, allowing for metric, non metric
and categorical variables, all at various levels (eg. differing product price points, colors,
or distribution options). As long ago as the 1970s, thorough testing by McCollough and
Best (1979) concluded that conjoint estimates of consumer purchase behavior are both
structurally and temporarily reliable. Based on this foundation of research and
experience, conjoint analysis design was used in the study.
3.2.4.1 Formulating a conjoint analysis design
Conjoint analysis design requires the researcher to determine a set of product attributes
important to consumer product evaluations and then choose differing levels to test
within each attribute. The selection of credible product attributes and levels is critical to
52
market realism and the subsequent external validity of results (Hair, Anderson et al.
1995). Individual product profiles are formulated from a rotation of the attributes and
levels. That is, determining a profile that reflects every possible combination of varying
attribute levels.
The task for respondents is to assess the resulting profiles according to the specific
requirements of a given study. They may rate each profile individually, or rank them or
choose from a specified set of 2 or more. The allocated score or ranking is a reflection
of a process where evaluations are determined by trading off some product features in
order to satisfy a desire to gain (or retain) others, thus giving higher scores to those
profiles that include the most desirable attributes at the most preferred level.
Conversely, those product profiles given low scores, or rankings, will be comprised of
attribute levels considered to be of little value or undesirable (SPSS-Inc 1997; Kupiec
and Revell 2001; Dean 2004). Analysis allows each respondents preferences to be
measured, yielding the importance for each attribute, and the part-worth or utility
value for each level. Consolidation of these results reveals which attributes are
making the strongest contribution to opinions and which attribute levels are most and
least preferred (Hair, Anderson et al. 1995; SPSS-Inc 1997; Dean 2004). Conjoint
analysis assumes that any product or service can be decomposed into its component
attributes, and therefore, the value of this product to consumers is a sum of the utilities
derived for the specific combination of attributes (SPSS-Inc 1997).
In order to closely mimic the purchasing process, the hypothetical products presented
must include attributes that are important to their evaluation and decision process.
Whilst the inclusion of all potentially influential attributes would describe a product more
comprehensively, anything in excess of five or six attributes is argued to diminish the
reliability of conjoint output (Green and Srinivasan 1990). Green and Srinivasan (1990)
also claim that the relative importance of an attribute is biased upwards as the number
53
of levels on which it is defined increases. Accordingly, a maximum of four or five
attributes, over three levels each, is suggested to provide an adequate description of
the product and still maintain a manageable number of stimuli for respondents (Hair,
Anderson et al. 1995).
3.2.5 Expected quality vs perceived quality
Previous studies have investigated the influence of extrinsic cues using an experimental
design where respondents experience various product offerings and provide opinions
regarding their quality or attributes (McIlveen and Buchanan 2001). Conjoint analysis
has also been used where respondents choose their preferred product option from a
number of described product profiles or rank them according to preference (Lee and
Lou 1996). However, the influence of extrinsic cues had not been tested previously
using both types of methodology to determine if any significant differences exist
between experienced quality gained through sensory perceptions and expected quality
as determined through mere description. This enabled assessment of differences that
may exist between respondents expectation of quality and experienced quality,
providing the opportunity to answer a second research question.
2. Is conjoint analysis a reliable predictor of actual consumer opinions?
3.2.6 Development of causal model and hypotheses
3.2.6.1 Causal model and hypotheses
The dependent variable shown in the conceptual model (Figure 2.1) as Quality is
derived from assessment of intrinsic and extrinsic cues, with their use moderated by the
consumer characteristics of knowledge and self-confidence. Quality has been defined
in many different ways, and there are many different conceptual approaches to its
54
measurement. A scientific approach to an objective quality definition and measurement
relies on determining quality according to an exact technical specification, and provides
an objective assessment without the direct involvement of consumers (Ophuis and Trijp
1995; Grunert 1997). In area of food assessment, the International Organization of
Standardization (IOS) provides a definition reliant on features and characteristics of a
product or service, that affect its actual ability to satisfy stated or implied needs (Becker
2000a).
The general philosophy in social research regarding quality definitions involves the
provision of product characteristics that are perceived by consumers to meet their
needs (Ophuis and Trijp 1995; Acebron and Dopico 2000; Becker 2000a). The
conceptual framework adopts this form of consumer oriented definition of quality
assessment as one based on consumer perceptions of overall quality, or product
superiority, in comparison to alternative offers. This perceived quality approach results
in a determination that is purely the result of a subjective judgment on the part of
respondents (Northen 2000).
The objective quality definition is appropriate when describing the manipulation of
intrinsic product attributes that occurs as part of the experimental design in the
research. This is due to the need to control and measure changes in actual product
quality accurately (Gatchalian 1999). Having established that the research would
involve experimental methods incorporating taste testing and conjoint analysis of data,
the next step is to develop a causal model providing rationalization for testable
hypotheses leading to analysis addressing the research questions.
Conjoint analysis uses an evaluation of quality (as defined) to determine the relative
importance of each variable contributing to this assessment. In doing so, the variable
quality is transformed into utility values representing each product attribute (and level)
55
tested. As a result, the respective influences of individual product attributes can be
quantified and each product profile tested can obtain a comparable value (quality)
score.
Country 1
Importance
Objective knowledge COO
Country 2
Price
Importance
Subjective knowledge price
Price
Level 1
Importance
Self-confidence acid / fat
Level 2
Figure 3.1 Causal model - moderating effects of consumer characteristics on product cue
usage
Figure 3.1 illustrates the associations formalized in the stated hypotheses that will test
the associations between the independent variables of COO, price and acid/fat on
product quality assessment and the potential moderating influence of knowledge
(objective/subjective) and self-confidence.
56
3.2.6.2 Hypotheses
A broad base of objective knowledge enables experts to quickly and accurately
determine which intrinsic cues are critical to quality and review them correctly and
efficiently; only relying on an extrinsic cue as a cognitive shortcut when it is truly
predictive of quality (Eroglu and Machleit 1988; Bhuian 1997; Kuusela, Spence et al.
1998; Acebron and Dopico 2000). Therefore:
H1 Objective knowledge is positively associated with the relative contribution
of intrinsic cues to an individuals assessment of product quality.
H2 Objective knowledge is negatively associated with the relative contribution
of extrinsic cues to an individuals assessment of product quality.
Conversely, those with low levels of objective knowledge tend to rely on limited
personal knowledge or product familiarity instead. These consumers are also more
likely to misinterpret any intrinsic cues considered due to the inability to assess them
accurately, subsequently increasing the risk of misjudging their value. This behavior
leads to a heavier reliance on extrinsic cues to form the basis of product quality
expectations (Rao and Monroe 1988; Rao and Olson 1990). Therefore:
H3 Subjective knowledge is positively associated with the relative
contribution of intrinsic cues to an individuals assessment of product
quality.
H4 Subjective knowledge is negatively associated with the relative
contribution of extrinsic cues to an individuals assessment of product
quality.
57
The literature shows that self-confident people display more strength and conviction in
their attitudes towards products than those less self-secure (Fazio and Zanna 1978;
Bearden, Hardesty et al. 2001; Moorman 2001). Given that they are less likely to find
even predictive extrinsic cues intimidating, it is expected they will strongly support
sensory evaluation (intrinsic cues) when objective product quality varies. Therefore:
H5 Self-confidence is positively associated with the relative contribution of
intrinsic cues to an individuals assessment of quality.
H6 Self-confidence is negatively associated with the relative contribution of
extrinsic cues to an individuals assessment of quality.
As conjoint analysis is a reliable means for measuring both expected and experienced
product quality, the ability of this methodology to predict actual consumer reactions to a
simulated product offer can be assumed.
H7 The paper based conjoint analysis method will be predictive of the
sensory based test in terms of the relative average importance of the tested
product attributes (intrinsic and extrinsic)
3.2.7 Research paradigm
The purposes of the research are to investigate the ability of extrinsic cues to override
sensory perceptions and quantify threshold levels by testing empirically the conceptual
model shown in chapter 2 (Figure 2.1). The study clearly delineates between the two
types documented forms of consumer knowledge and self-confidence levels. As these
moderating variables are expected to be used as filters through which product cues will
be evaluated, and in some cases ignored, this delineation clarifies their contribution to
the product quality assessment process. The model will go beyond currently existing
58
literature involving CI and price studies and may also make a contribution in general to
consumer behavior research relating to consumer use of extrinsic cues.
Quantitative measurement will test the theoretical model (Figure 3.1) described,
indicating that the research is undertaken largely in the tradition of objectivism and
positivism (Mangan, Lalwani et al. 2004). However, the study design also includes a
qualitative component, recognizing the benefits derived through the inclusion of
appropriate qualitative techniques, even when research is undertaken in a
predominantly positivist vein. Incorporating aspects of both interpretist and positivist
paradigms is an accepted method of triangulating data to enhance validity (Hussey and
Hussey 1997; Naslund 2002). Therefore, while a component of the study is qualitative
and interpretist in nature, reliance is ultimately based on operationalizing concepts
through the testing of quantifiable variables. This approach is consistent with positivist
beliefs; assuming the existence of an objective physical and social world where the
researcher is an independent observer gathering empirically testable data (Mangan,
Lalwani et al. 2004). This functionalist view is employed in the hope of providing
rational explanations that can be used by practitioners in marketing strategy
development or as the basis of further research.
3.2.8 Justification for the model
When Hunt (1991) describes the normative criteria used to validate the explanatory
adequacy of a model, he specifies, (i) the phenomenon to be explained should be
expected to occur and (ii) the model should be pragmatic, intersubjectively certifiable,
and have empirical content (p51). The literature has established that consumers
consider both intrinsic and extrinsic cues when forming opinions about expected and
experienced product quality. Chapter 2 provides empirical evidence to support price
and COO as two extrinsic cues that can be expected to contribute to overall quality
59
appraisals. Further, consumer knowledge and self-confidence have been found in
previous research to moderate the use of both types of cues. The gaps in existing
knowledge relate to the specific measurement of self-confidence and differing types and
levels of knowledge and their specific influence on the use of, and reliance on, extrinsic
cues. The proposed model is pragmatic and contains empirical content tested in the
experiment design methodology employed to objectively measure these independent
and moderating variables and quantify their respective influences on judgments of
product quality.
3.2.9 Stimuli used
The model was tested using two food categories, wine and cheese. In line with the
recommendation from Hair, Anderson et. al. (1995), the conjoint analysis design
developed was based on two extrinsic cues (price and COO) and one intrinsic cue each
varied over three levels. The rationale for choosing these stimuli and stipulated levels
for testing is provided in this section.
3.2.9.1 Wine and cheese
A significant body of empirical evidence exists using food products to research the
influence of extrinsic cues on consumer assessment of product quality; examples of
types tested include beef, chicken, pork, eggs, wine, cheese and orange juice
(Pechmann and Ratneshwar 1992; Grunert 1997; Al-Sulaiti and Baker 1998; Becker,
Benner et al. 2000b; Bernues, Olaizola et al. 2003; Bredahl 2003; Jover, Montes et al.
2004; Kardes, Cronley et al. 2004). A review of these studies also revealed a wide and
varied range of methodologies including conjoint analysis, taste test experiments, pre
and post purchase surveys, and in-store surveys, each supporting the suitability of food
products for the proposed methodology. In order to use a general population sample
60
for the quantitative stages of the research, it was necessary to pick food products that
are routinely consumed by members of the adult Australian population, and wine and
cheese were identified as examples of these types of products (ScanTrack-Liquor 2005;
www.Dairy-Australia.com.au 2007).
Interviews with product experts Jim Smith (wine maker and industry consultant) and
Louise Elder (cheese maker and wholesaler) were conducted to select specific wine
and cheese varieties that are readily available, commonly consumed by Australian
shoppers and suitable for objective quality manipulation. Unwooded chardonnay was
suggested as a suitable wine due to its familiarity to consumers, and a single intrinsic
cue (acid), can easily and accurately be manipulated to produce significant differences
in objective product quality. Increasing the acid level in chardonnay wine produces sour
wines (termed green) that are sharp and unpleasant on the palate (Baldy 1993). For
cheese, a soft variety such camembert or brie was suggested as they are readily
available, commonly purchased and are available with differing levels of fat content.
With camembert and brie, as with many other cheese varieties, fat is an important
intrinsic cue affecting objective product quality. Generally, higher fat levels result in a
creamier texture (termed enhanced mouth feel) and a better taste (Aaron, Mela et al.
1994; Hamilton, Knox et al. 2000). Based on this information, unwooded chardonnay
(wine) and camembert or brie (cheese) were chosen as test products; with acid and fat
being manipulated to provide measurable and controllable changes to objective product
quality.
3.2.9.2 COO and price
COO has been found to influence consumer perceptions of both wine and cheese
products (Keown and Casey 1995; Manrai, Lascu et al. 1998; Gluckman 2001; Jover,
Montes et al. 2004). For example, France is famous for producing a wide variety of
61
high quality wines and is strongly and positively associated with this product; therefore,
France is expected to be considered by respondents to be a producer of high quality
chardonnay (Keown and Casey 1995; Jover, Montes et al. 2004). Conversely,
countries such as Chile, South Africa and Canada are far less famous for producing
high quality wines and are unlikely to be associated positively with chardonnay for
Australian consumers. To select suitable countries for the 3 levels needed for each
product in the conjoint analysis design, alternative country/product alternatives were
explored in the qualitative component of the methodology.
3.2.9.3 Hypotheses summary
As the stimuli to represent the dependent variables are established, it is now possible to
incorporate these into specific hypotheses testing the relationships indicated in Figure
3.1 and summarized in table 3.1 that follows.
62
Table 3.1 Hypotheses tested
H1 Objective knowledge is positively associated with the relative contribution of

intrinsic cues to an individuals assessment of product quality.
H1a Objective knowledge is positively associated with the relative contribution of an intrinsic cue
(acid) to an individuals assessment of product quality (wine).
H1b Objective knowledge is positively associated with the relative contribution of an intrinsic cue
(fat) to an individuals assessment of product quality (cheese).
H2 Objective knowledge is negatively associated with the relative contribution of

extrinsic cues to an individuals assessment of product quality.
H2a Objective knowledge is negatively associated with the relative contribution of an extrinsic cue
(COO) to an individuals assessment of product quality (wine / cheese).
H2b Objective knowledge is negatively associated with the relative contribution of an extrinsic cue
(price) to an individuals assessment of product quality (wine / cheese).
H3 Subjective knowledge is positively associated with the relative contribution of

intrinsic cues to an individuals assessment of product quality.
H3a Subjective knowledge is positively associated with the relative contribution of an intrinsic cue
(acid) to an individuals assessment of product quality (wine).
H3b Subjective knowledge is positively associated with the relative contribution of an intrinsic cue
(fat) to an individuals assessment of product quality (cheese).
H4 Subjective knowledge is negatively associated with the relative contribution of

extrinsic cues to an individuals assessment of product quality.
H4a Subjective knowledge is negatively associated with the relative contribution of an extrinsic
cue (COO) to an individuals assessment of product quality (wine / cheese).
H4b Subjective knowledge is negatively associated with the relative contribution of an extrinsic
cue (price) to an individuals assessment of product quality (wine / cheese).
H5 Self confidence is positively associated with the relative contribution of

intrinsic cues to an individuals assessment of quality.
H5a Self confidence is positively associated with the relative contribution of an intrinsic cue (acid)
to an individuals assessment of product quality (wine).
H5b Self confidence is positively associated with the relative contribution of an intrinsic cue (fat) to
an individuals assessment of product quality (cheese).
H6 Self confidence is negatively associated with the relative contribution of

extrinsic cues to an individuals assessment of quality.
H6a Self confidence is negatively associated with the relative contribution of an extrinsic cue
(COO) to an individuals assessment of product quality (wine / cheese).
H6b Self confidence is positively associated with the relative contribution of an extrinsic cue
(price) to an individuals assessment of product quality (wine / cheese).
H7 There will be no significant differences in attribute utilities, or average

importance, between data collected via sensory experiment and paper based
survey.
63
3.3 Overview of research design
The research is largely hypothetico-deductive, in that the underlying hypotheses come
from the literature, and refined with insight derived from qualitative data and tested
through an established experimental approach called conjoint analysis (implemented in
2 different ways to compare results).
Stage 1 Stage 2 Stage 3 Stage 4 Stage 5
Qualitative Quantitative Quantitative Quantitative Comparison of

Study Conjoint Pilot Sensory Conjoint study results from
Study Experiment stages 3 and 4
Figure 3.2 Stages of the research
The study is comprised of 5 stages (Figure 3.2). The first stage, the qualitative
component, consists of 2 focus groups. The subsequent sections of this chapter outline
the research methodology used to conduct these groups, including the sampling plan,
data collection instruments and data analysis procedures. The remaining parts of this
stage are quantitative in nature, and this methodology is outlined in Chapter 5.
Stage 2 consists of a pilot survey in the form of a self-administered questionnaire.
Respondents are required to rate described product profiles where objective product
quality is varied (via manipulated intrinsic cues), often in conflict with the extrinsic cues
provided (via manipulated price levels and COO). Objective knowledge, subjective
knowledge and self-confidence are also measured in this pilot study, using instruments
to be described in chapter 4. The primary objectives of the pilot survey are to validate
measurement instruments, products, product cue types (chosen as stimuli) and attribute
levels. It is hoped that by conducting the pilot survey, any underlying weaknesses or
errors in the planned methodology will be identified and remedied prior to conducting
64
stage 3, given its experimental nature and the associated time and resource
restrictions.
Stage 3 consists of sensory experiments (taste tests) where respondents will taste and
evaluate product samples manipulated to replicate the same product profiles described
in the final conjoint analysis study. Analysis of the resulting data will investigate
empirically the ability of extrinsic cues (price and COO) to overwhelm experienced
intrinsic cues and also to gauge the degree to which knowledge (type and level) and
self-confidence can moderate these opinions.
Using the questionnaire which will be developed and refined after conducting the pilot
study and taste test experiment, stage 4 consists of a survey where the respective
influence of described intrinsic and extrinsic cues is measured employing the
established conjoint analysis design. Identical product profiles employed in the
previous stages are to be replicated for this aspect of the research. As in the pilot,
intrinsic cues will be described rather than experienced, resulting in a determination of
quality based on expectation rather than sensory experience.
In stage 5, comparisons will be made between expected product quality as measured
by the conjoint analysis survey data and perceived quality as determined by analysis of
the sensory evaluation data. Quantifying any significant differences between quality
expectations derived by appraisal of product descriptions and actual sensory evaluation
for each tested intrinsic and extrinsic cue combination will provide insight into the ability
of conjoint analysis to predict of consumer opinions.
65
3.4 Stage 1 Qualitative study
Focus groups were conducted to confirm that Australian consumers believe that COO
and price are important and predictive extrinsic cues influencing product quality for the
selected food products tested in the following stages. It was also necessary to confirm
which countries would be positively and negatively associated with these products given
that COO effect has been found to be product, country and market specific (Hastak and
Hong 1991; Al-Sulaiti and Baker 1998; Kuusela, Spence et al. 1998; Insch and McBride
2004).
Group interviewing enables interaction amongst participants allowing the researcher to
easily and quickly gain valuable insight into consumer opinions regarding topics of
specific interest (Malhotra, Hall et al. 2002). Focus groups are also useful to triangulate
information from other sources and may also reveal new and unexpected findings for
further investigation (Hussey and Hussey 1997; Naslund 2002). A substantial body of
literature exists, indicating countries that are likely to generate applicable images and/or
associations with wine and cheese; however, these could not be assumed. As
discussed in chapter two, COO effects are often market, product and country specific
(Chao 1989 b; Cordell 1992; Chao and Rajendran 1993; Al-Sulaiti and Baker 1998;
ScanTrack-Liquor 2005). Therefore, it is necessary to identify countries that would
generate varying expectations of quality for Australian consumers specific to
assessment of wine and cheese. Furthermore, it is considered important to verify
important practical aspects of employing wine and cheese in the research.
3.4.1 Sampling
A judgment sample of part-time night students (TAFESA City Campus) was recruited to
participate in 2 groups of 4 and 5 participants each respectively. While a small
66
judgment sample has limitations, in particular that the views of such a limited number
participants will not be generalizable, this sampling method can be justified for use in
qualitative, exploratory research (Malhotra, Hall et al. 2002). Prior to final selection all
group members were screened to ensure they purchased and consumed wine and
cheese at least once per fortnight. The demographics of group members are provided
below in Table 3.2.
Table 3.2 Demographic profile of focus group participants
Demographics Group 1 Group 2

Age range 30 to 52 years 24 to 38 years
Gender 2 females, 2 males 2 females, 3 males
3.4.2 Data collection
Respondents were video-taped and responses to questions regarding country and
product preferences were recorded on individual score sheets and butchers paper. A
buying scenario was provided to group members in order to stimulate their thinking in
terms of product attributes and their respective levels of importance to the purchasing
decision. They were asked to consider purchasing a bottle of chardonnay as a gift for
someone special/significant or to be taken to an important dinner party. Respondents
were advised that no Australian wines were available to consider and they must make a
purchase before they leave the wine shop. The group members were asked to list
product attributes (extrinsic and intrinsic) they considered when buying wine, and to
score them according to overall importance when making a choice. A rating scale of 0
(not at all important) to 10 (very important) was provided. Respondents then listed
countries they thought would produce high, low and average quality chardonnay.
Identical methodology was repeated for cheese. No prompting or suggestions were
given in relation to possible countries or attributes for either product.
67
3.4.3 Results
Attributes considered important by group members when evaluating wine and cheese
are illustrated in tables 3.3 and 3.4. Data from the focus groups relating to the types of
extrinsic cues and their considered importance to the purchasing decision are largely
consistent with the literature regarding price and COO cited, thus supporting their
choice as products for stages 2 and 3 of the study (Schifferstein 1996; Siu and Wong
2002; Jover, Montes et al. 2004). Groups were alike in opinion of the importance of
price when considering both wine and cheese; however, there was strong contrast in
their views relating to the importance of COO between the two products. While
respondents in both groups believed COO is highly important when considering the
quality of a wine, most in group 1 did not think it was likely to make much difference to
the quality of cheese. The literature supports that the COO effect being diminished may
be attributed to limitations imposed from the small sample size, or that the purchase of
cheese is considered a much lower involvement purchase (Piron 2000; Ahmed,
Johnson et al. 2004). Based on these results, further testing of price and COO as
extrinsic cues affecting expected product quality is justified and cheese and wine have
been found suitable as products for testing this influence.
Table 3.3 Important product attributes for wine
Scale items scored 0 10 Where 0 is not at all important and 10 is very important
Attributes listed Ratings Group 1 Ratings Group 2
Brand 6 9
COO 8 8
Description on label 3 Not given
Label (artwork) Not given 9
Price 9 8
Purpose (situation) 8 7
Rarity Not given 3
Region 4 Not given
Taste 10 3
68
Table 3.4 important product attributes for cheese
Scale items scored 0 10 Where 0 is not at all important and 10 is very important
Attributes listed Ratings Group 1 Ratings Group 2
Purpose (situation) 5 Not given
Price 9 8
Taste 10 6
Texture 10 9
Brand 3 3
COO 2 10
Appearance 9 Not given
Packaging Not given 9
Rarity Not given 8
Participants were then asked to list the countries they believed would produce the
highest, lowest and average levels of quality respectively in terms of the products being
discussed. Again these were unprompted responses. The countries listed and their
expected influence on product quality is illustrated below in Table 3.5.
Table 3.5 Countries suggested by participants as indicators of quality for wine and cheese
Countries Wine Cheese

Highest quality France, Germany, U.S. UK, France, Holland, Denmark, NZ
Average quality Chile, Italy, Spain, South Africa NZ, U.S. Canada, UK,
Lowest quality Canada, South Africa, UK, China, Argentina Argentina, China, Greece, Chile
This segment of the group discussions were the liveliest and elicited the most varied
and polarized results. Initially it was difficult to motivate respondents to discuss foreign
products, particularly wine. Group members consistently remarked that they purchased
only Australian wines and usually Australian cheeses. Consequently, they had little
knowledge or experience to drive their expectations for either product if sourced
elsewhere. However, when pressed for a response France was cited as a likely source
country (by both groups) for the highest quality wine and cheese and was not
associated with average or low quality levels. This result is not surprising given
Frances reputation for producing fine wines and gourmet cuisine and illustrates a
strong reliance on the COO cue. There was considerable debate and disagreement
amongst respondents deliberating where average and low quality products may be
69
produced. Countries not known for producing dairy products were listed as sources of
poor quality cheese, e.g. China. Respondents found it hard to even a compose a
strong country image for South American countries such as Chile and Argentina and
used what very little knowledge they possessed form an opinion of hypothetical
products sourced from these locations (Han 1989). As many believed these countries
to be very poor (3rd world), they seemed to make a link between extensive poverty and
low quality in all things including cheese and wine (Chao 1993). However, countering
these opinions were individuals believing Chile and Argentina would produce good
wines, as they had read about them and heard they were good. This increased level
of knowledge supported a higher expectation for this wine, but not for cheese. It was
generally accepted that European countries made good cheese and any tropical or
Asian countries would make poor cheese and poor wine. The opinions relating to
Canada and the U.S. ranged from an expectation that anything they produced would be
at least average due to their industrialized status (Chao 2001), through to an
expectation that quality would be very low because everything they make is massed
produced and while that is acceptable for manufactured goods, they believed it would
have a negative impact on food and wine products. There also seemed to be a
reasonable level of concern regarding pollution and pesticide levels affecting expected
quality and product safety (Tse 1999; Siu and Wong 2002). It is useful to review
selected comments from group members as they provide an insight into their
perceptions regarding the importance of price and COO.
70
Comments from focus group participants
Price:
You only get what you pay for
When you cant take a risk, pay more
I think you can still buy some good wines that are reasonable, price is not always
that reliable
I wouldnt take something cheap to a party; it may not be good enough and people
would think Im cheap
Price is the most important if I couldnt buy Australian wine I would only look at
price
I dont even buy cheap wine for home whats the point?
I dont know a lot about wine, so Id be afraid to buy a cheap one
COO:
All Europeans make good cheese, dont they?
Asians dont make cheese do they no cows!
The Kiwis make good cheese; at least it would be safe
I dont think they would make good cheese in South America, its too hot isnt it?
Ive never drunk French wine, but youd have to expect it would be good
I think the South Americans would make good wine, but not the Canadians
Anything from France at least would make you look like you spent money
Isnt everything genetically modified in the States? That cant be good
I dont know anything about South America except theyre all poor
They make lots of wine in Chile, it must be OK. Ive been reading about it
The data indicates that France is generally expected to product high quality for both
wine and cheese. However, results are mixed and somewhat contradictory in relation
to source countries associated with average and low quality products. This problem
was resolved by conducting a short and simple survey of part-time marketing students
(n = 33) at TAFESA (City Campus) where respondents were asked to rank the 5
countries under consideration from 1 to 5, where 1 was considered to be the source
country of the highest quality wine/cheese and 5 was believed to produce the lowest
71
quality wine/cheese compared to the other countries listed for consideration. The
results are shown later in this chapter.
3.4.3.1 Taste testing
Following the discussion of important product attributes and wine/cheese producing
countries, participants were asked to sample chardonnay from four different countries.
All of the wines purchased were of almost equal value (approximately $10 AUS);
however, the actual variance in their objective quality was unknown. In the case of
cheese, four different Australian Tasty Cheese brands were purchased at random
(approximately $5 AUS), where again, objective product quality was not predetermined.
The actual objective product quality was not deemed important for this aspect of the
study, as the objectives of these informal taste tests were to explore the influence of
price and COO on sensory perceptions by seeing if respondents would be consistent in
their rating of the samples provided across both tests rather than to measure their ability
to accurately judge objective product quality.
When the initial samples were tasted, the COO and an assigned price were revealed to
the respondents, and they were asked to rate the sample from 0 to 10 (0 = poor
quality and 10 = excellent quality). Respondents then tasted the same products again,
but the samples were presented in a different order and the COO and price were not
revealed. Results for these taste tests are shown below in Table 3.6 and Table 3.7.
Table 3.6 Taste tests - wine
Mean score Mean score Variance

Countries
COO/price revealed COO/price hidden in mean scores
Chile - $7 6.333 6.111 - 0.222
France - $53 3.667 2.889 - 0.778
NZ - $12 5.111 5.889 0.778
USA - $22 3.222 4.444 1.222
72
Table 3.7 Taste tests - cheese
Mean score Mean score Variance

Countries
COO/price revealed COO/price hidden in mean scores
Poland - $3 8.556 8.222 -0.333
England - $8 5.111 5.000 -0.111
NZ - $5 5.556 4.889 -0.667
As stated, it was not the intention of the taste tests to provide empirical evidence, but
rather to explore the stated areas of this stage of the research. For this purpose the
data is very useful; participants reported a marked difference in their perceptions of
quality between the samples tested (both products) indicating provided extrinsic cues
had influenced their perceptions. The results for the cheese as compared to the wine
suggest that these cues were less important in appraising a lower value/status product.
This outcome is consistent with expectations and the literature (Quester and Smart
1998; Piron 2000; Ahmed, Johnson et al. 2004).
3.4.3.2 Country rankings (survey of students)
In order to make a final choice of source countries associated with differing product
quality expectations (chardonnay and camembert) a short survey was conducted. A
sample of 33 part-time evening students of marketing at TAFESA (City Campus)
completed a short self-administered questionnaire; lecturers from two evening
marketing classes agreed to distribute the questionnaires prior to the commencement of
classes and participation by students was voluntary.
The questionnaire listed countries cited by the members of the focus groups and
respondents were to review the respective countries given and consider each as a
possible source country for a bottle of high quality chardonnay wine. They were then
asked to rank them from 1 to 5, where 1 would be the country they expected to make
73
the best quality of chardonnay and where 5 would be expected to produce the poorest
quality. The respective list for cheese and identical instruction was repeated for
camembert cheese. Analysis of the results is seen below in Table 3.8 and Table 3.9.
Table 3.8 Country rankings for chardonnay
Chile Canada France USA NZ

Mean 4.09 3.82 1.79 3.12 2.18
Std. error of mean 0.19 0.19 0.16 0.19 0.23
Median 4.00 4.00 2.00 3.00 2.00
Mode 5.00 5.00 1.00 2.00 1.00
Std. dev. (SD) 1.07 1.07 0.93 1.11 1.33
Range 1.00 3.00 3.00 3.00 4.00
Minimum 1.00 2.00 1.00 2.00 1.00
Maximum 5.00 5.00 4.00 5.00 5.00
The data plainly confirms France is more closely associated with the highest quality
chardonnay than other countries listed with a mean score of 1.79 and Chile is the likely
source of the lowest quality with a rating of 4.09. The choice country for the average
or mid-range quality wine was less clear. The U.S. was determined as the most
appropriate option for a number of reasons. The literature indicates that the stronger
the perceived cultural and/or geographical links between a market and a source
country, the more favorable the response to products from those countries is likely to
be. Conversely, products from countries considered distant are viewed less favorably
(Heslop, Papadopoulos et al. 1998; Watson and Wright 1999). Using the notable
research of Hofestede (1991) as a basis, New Zealand would be considered close to
Australia in both geographical and cultural terms. During the group discussions, it
became obvious that these Australian consumers had very little knowledge or
experience with foreign wines or cheeses, with comments commonly reflecting their
strong loyalty to Australian wine in particular. The risk is that if provided with a product
sourced from New Zealand, respondents will use it as a proxy for an Australian offering
when evaluating the product profiles provided in later stages of the research (Hofestede
1991). Therefore, the U.S. was chosen instead of New Zealand in order to eliminate
74
this possibility. Using the same principles, Argentina, Canada and France were
selected as sources of varying quality of cheese.
Table 3.9 Country rankings for camembert
NZ Argentina U.S. Canada France

Mean 2.12 3.88 3.88 3.54 1.58
Std. error of mean 0.19 0.21 0.17 0.15 0.18
Median 2.00 4.00 4.00 4.00 1.00
Mode 2.00 5.00 5.00 4.00 1.00
Std. Dev. (SD) 1.11 1.19 1.00. 0.87 1.06
Range 4.00 3.00 3.00 3.00 4.00
Minimum 1.00 2.00 2.00 2.00 1.00
Maximum 5.00 5.00 5.00 5.00 5.00
3.4.4 Qualitative data analysis summary
Analysis of data provided by focus group participants supports the influence of COO
and price on expectation of product quality for the products proposed for testing. As a
result of the data (both qualitative and quantitative), France, the U.S. and Chile will be
employed as source countries for high, average and low quality chardonnay
respectively. For cheese, France, Canada and Argentina will represent high, average
and low quality levels respectively. While the focus group and survey respondents
represent the views of relatively few individuals the results are consistent with existing
literature and supportive of the conceptual framework and propositions. Chapter 5
provides a detailed description of the methodology used in the quantitative aspects of
the research.
75
3.5 Summary
Chapter 2 provided an illustration and justification of the conceptual model (Figure 2.1)
articulating the contribution of intrinsic and extrinsic cues to product quality evaluation
and discussing how the moderating variables of knowledge and self-confidence can
influence consumer use and reliance on these cues. Evolving from this theoretical
foundation, a causal model (Figure 3.1) articulating the relationships between the
dependent product variables (extrinsic/intrinsic cues) and independent consumer
variables (knowledge and self-confidence) was proposed and defended in Chapter 3.
Based on these proposed relationships and effects, hypotheses were developed and
summarized in Table 3.1. The next section of the chapter discussed the research
design, including a brief description of the 5 stages included in the study. This was
followed by the methodology used in stage 1 (the qualitative component) and the
findings of subsequent analysis of results, including their applications relevant to later
stages of the research. Chapter 4 is devoted to the discussion and explanation of the
adopted research methodology specific to the quantitative aspects of the research,
including the measures used and the development of data collection instruments.
76
4 Quantitative methodology
4.1 Introduction
Chapter 3 provided, firstly, an illustration and justification of the causal model and
resulting hypotheses to test the relationships suggested. The remaining sections
provided an overview of the research design, including a brief description of each of the
5 stages. This was followed by a description of the methodology used for stage 1 (the
qualitative component) and the results of subsequent data analysis. The knowledge
gained from that stage of the study was used to develop the conjoint analysis design
and data collection instruments described in chapter 4. Also included here are
descriptions of the measure used to quantify consumer knowledge (both types) and
self-confidence and a brief overview of employed data analysis tools.
4.2 Sampling methods
Non-probability, convenience sampling was employed with respondents recruited in
Adelaide (South Australia) using a variety of communications and incentive methods,
determined according to the requirements of the particular stage of the study. Samples
were comprised of members of the general public and the student (evening classes
only) and staff populations at campuses of TAFESA South.
It is acknowledged that student samples can contribute to the limitations in results as
they have been found to be unrepresentative of the general population due to the
limited nature of the demographic profiles: e.g. a very high proportion of students are
under the age of 25 years (Peterson 2001). However, research has demonstrated that
the effect of this limitation can be minimized by using mature age students enrolled in
evening classes. Due to the diverse demographics typical of these students they are
77
more likely to be representative of the general population (James and Sonner 2001).
Participation was restricted to those over the age of 18 years.
4.3 Data collection instruments
The questionnaires used in each stage of the research were based upon the initial
version using the measures described in chapters 3 and 4. Respondent demographics
such as gender, age, income, education and occupation were also captured in order to
compare the profile of each respective sample with that of the general population of
Australia (over the age of 18 years). Data analysis results and feedback derived from
respondents participating in the pilot study were used to make any required
improvements in subsequent versions of the instrument. In the relevant sections of this
chapter, each version of the questionnaire is described in detail, including identified
limitations and improvements implemented, and copies can be found in the cited
appendices.
4.3.1 Use of full profile conjoint analysis
Full profile analysis remains one of the most commonly used forms of conjoint analysis
and has an important advantage in that each profile is assessed individually allowing
respondents to focus their attention on only one profile at a time; however, the
associated risk is increased respondent fatigue if the number of profiles to be assessed
is too high (Rao and Hauser 2004). Using this approach, it is typical for a respondent to
review an array of approximately 15 product profiles, each evaluated on a metric scale
(SPSS-Inc 1997). This centers the attention of the respondents on the acceptability
levels of attributes as opposed to comparing attributes between alternative offers, with
the usual outcome being a dominance in the rating decision by a small number of
attributes (Huber, Wittink et al. 1991). However, this was not considered a limitation in
78
this study as the design is restricted to only 3 attributes in total. Further, the objectives
of the research do not include identifying which product profile, from amongst a
selection of offers, is considered the most desirable by respondents. Rather, the aim is
to investigate the power of communicated extrinsic cues to influence respondent
expectations of product quality and to measure their impact on experienced product
quality through sensory experiments. Therefore, a survey design using a full profile
conjoint approach was deemed to be most appropriate for this research (Hair, Anderson
et al. 1995; Huber 1997).
4.3.2 Orthogonal fractional factorial design
While anywhere up to 15 product profiles has been found feasible when respondents
are assessing profiles only by description, sampling 15 wines and 15 cheeses in one
tasting session would be too onerous for participants for the sensory stage of the
research. This is because it is difficult for participants to remain susceptible to the
sensory differences in each sample due to potential desensitization of their palates;
also, they may experience fatigue due to the extended time involved in the tasks
(Gatchalian 1999). To reduce their burden, an orthogonal fractional factorial design
reduces the number of profiles overall, whilst ensuring that an adequate representation
of each attribute level is maintained to estimate a parameter of main effect of each
attribute and level (SPSS-Inc 1997; Kupiec and Revell 2001; Rao and Hauser 2004).
The full factorial design based on a 3 country x 3 price level x 3 acid/fat specification
results in 27 possible product profiles for each product. By adopting an SPSS fractional
factorial design this number was reduced to a subset of 9 attribute combinations for
each product, from which it is possible to test the part worth contribution of each
respective attribute level (Hair, Anderson et al. 1995; SPSS-Inc 1997).
79
4.3.3 Training respondents
As profiles are not assessed concurrently, participants will not get an immediate feeling
for the range of possible product characteristics and their respective variances until they
have assessed several profiles. In order to overcome this limitation, 2 warm up
profiles are recommended for respondents to gain a sense of attribute numbers and
possible levels; thereby stabilizing ratings and increasing accuracy (Curry 1996). This
advice is countered to some degree by Jaeger, Hedderly and MacFie (2000) who
suggested that when product categories are familiar and regular purchases are
commonly made, a consumers market knowledge and experience are likely to provide
adequate attribute referencing. However, they caution that when product profiles are
only described (either verbally or through print) unaccompanied by samples or realistic
pictorial representation, some respondent training or attribute familiarization is
recommended (Jaeger, Hedderley et al. 2000). Based on this, two hold out or practice
profiles were incorporated into the design for respondent training and to enhance the
internal validity. While the ratings given to hold outs is not included in the
determination of attribute part worths, they are analyzed as a test of internal validity.
The part worth statistics for attributes in the hold out profiles are compared to those in
the fractional factorial design to check for consistency of respective attribute and level
influence (Hair, Anderson et al. 1995; SPSS-Inc 1997). This inclusion resulted in a final
design comprised of 11 profiles for wine and 11 for cheese.
4.4 Measures used for consum er characteristics
4.4.1 Subjective knowledge
Subjective knowledge is the consumers perceived level of expertise or self-assessed
level of product knowledge (Monroe 1976; Brucks 1985; Park, Mothersbaugh et al.
1994; Wirtz and Mattila 2003). The most common way that subjective knowledge has
80
been measured in the past is by a single self-report item; other methods include
semantic differential scales and ad hoc multi item scales developed specifically for the
pertinent study (Flynn and Goldsmith 1999) The outcome has been a wide range of
approaches with few methodologies validated through testing or use in other studies
(Flynn and Goldsmith 1999). The 8 item scale used in this research was used by
Goldsmith, dHauteville et al. (1997) and validated by testing across 8 different product
categories (including 2 wine products) in 3 separate studies by Flynn and Goldsmith
(1999). The items comprising this scale are seen in table 4.1.
Table 4.1 Subjective knowledge scale items
Response format: 9 = strongly agree; 1 = strongly disagree

1. I know pretty much about wine/cheese
2. I know how to judge the quality of wine/cheese
3. I do not feel very knowledgeable about wine/cheese (r)
4. Among my circle of friends, Im one of the experts on wine/cheese
5. Compared to most other people, I know less about wine/cheese (r)
6. I have heard of most of the wines/cheeses around.
7. When it comes to wine/cheese, I really dont know a lot (r)
8. I can tell if a wine/cheese is worth the price or not.
4.4.2 Objective knowledge
There have been inconsistencies in terms of definitions of objective knowledge type and
the instruments employed in measurement, making comparisons between studies
difficult and risky (Alba and Hutchinson 1987; Rao and Olson 1990). Objective
knowledge is product information stored in long-term memory based on experience and
cognitive learning, with consumers maintaining high levels of objective knowledge by
seeking out up to date product information (Alba and Hutchinson 1987). True expertise
is a combination of high levels of objective knowledge and consumer product familiarity
(Alba and Hutchinson 1987; Heimbach, Johansson et al. 1989; Cowley 1994; Alba
2000). It is not realistic to expect any significant segment of the consumer market to
81
possess levels of expertise consistent with a professional person working in a field
aligned with the product or product category. For example, one would not expect
consumers in the wine market to have the same knowledge as a wine maker or
vineyard owner. Nor would consumers of cheese realistically be expected to be as
knowledgeable as a cheese maker or retailer. Therefore, it was critical to develop
questions that measure objective knowledge in terms of what consumers could
reasonably be expected to know, rather than individuals in either industry.
In her study Brucks, (1985:p13) provides an insight into the appropriate measures of
consumer objective knowledge and suggests that suitable testing should include the
following types of items:
1. Terminology
a. product terms to be matched with correct definitions
b. decoys (terms not associated with the products tested)
2. Available attributes
a. critical intrinsic attributes
b. common attributes (usually present but not critical)
c. decoys (attributes not associated with the products tested)
d. criteria for evaluating attributes
3. Product usage situations
a. product examples to be matched with given situations
4. Attribute covariation
a. the relationship between an attribute and price
b. decoys (relationships between attributes that dont exist)
82
Product experts were consulted and relevant literature was examined to develop a set
of questions for each product in order to measure respondent objective product
knowledge (Baldy 1993; Goldsmith, d'Hauteville et al. 1997) It was considered critical
to encompass as many of the areas described by Brucks (1985) as achievable, whilst
ensuring that the assessment is relatively short and easy to complete. A multiple
choice format was used, where each question offered 3 possible answers and the
option of dont know if respondents were unable to choose an answer they believed to
be correct (Goldsmith, d'Hauteville et al. 1997). The scope of the questions was not
confined strictly to the specific wine and cheese variety being tested, but also included
general knowledge questions related to the broader product category. The final test
developed consists of 14 questions per product; these are shown in tables 3.3 and 3.4
in the following sections.
83
Objective knowledge test wine:
Table 4.2 illustrates the questions used to measure objective product knowledge for
wine. These questions were put together using information from the literature and input
from Jim Smith, wine maker and consultant (Baldy 1993; Smith 2004). The correct
answers are shown in italics.
Table 4.2 Objective knowledge questions wine/chardonnay
1. Letting a wine breathe: 2. Cellaring of wine is done (basically) to help:

Is to remove the cork for a time, prior to drinking 1 Wine mature quickly 1
Means aerating it so it can react with the oxygen 2 Wine mature slowly 2
Always improves the flavour of wine 3 Keep pests away from wine 3
Dont know 4 Dont know 4
Which one of the following white varietals is most Which one of the following white varietals is
3. 4.
likely to be aged in oak? most likely to improve with aging?
Riesling 1 Sauvignon Blanc 1
Chardonnay 2 Chenin Blanc 2
Sauvignon Blanc 3 Chardonnay 3
5. Tannin gives wine: 6. Chardonnay grapes are:
Bitterness 1 Among the finest grown for white wine 1
Tartness 2 Not usually used in sparkling wine 2
Astringency 3 Often used to make sweet wines 3
Terms often linked with the taste of Chardonnay
7. Chardonnay, typically has an aging potential of: 8.
are:
2 or 3 years 1 Apple, peach, citrus 1
3 or 4 years 2 Plum, spice, mint 2
5 or 6 years or longer 3 Floral, honey, lychee 3
What percentage (%) of the wine in the bottle must When thinking about matching foods with wines,
9. be made from grapes harvested and crushed in 10. trying to achieve the most complimentary
the year named, if a Vintage date is given? combinations, it is important to remember that:
85% 1 Very sweet food will counter the acid in the wine 1
95% 2 Very salty foods counter acid in the wine 2
100% 3 Very acid foods will bring out the acid in the wine 3
11. An oak aged Chardonnay will, typically, be: Champagne is an excellent choice to
12.
accompany:
Less full bodied than many other white wines 1 Smoked salmon 1
More full bodied than many other white wines 2 Chinese food 2
Comparable in body to many other white wines 3 Most foods 3
The term green is often used to describe a
13. 14. Chilling wine (even red wine) will often:
wines:
Colour 1 Improve the taste of a poor wine 1
Acidity 2 Make tannins less noticeable 2
Age 3 Make no real difference to perceived quality 3
84
Objective knowledge test cheese:
Table 4.3 illustrates the questions used to measure objective product knowledge for
cheese. These questions and answers were comprised by industry expert Louise
Elder, cheese maker and wholesaler. The correct answers are shown in italics.
Table 4.3 Objective knowledge questions cheese/camembert
1. White mould is: 2. Coagulation is:

Fresh curds aged in warm temperatures 1 The conversion of milk solids to curd and whey 1
Fresh curds ripened by introducing surface mould 2 The basis of cheese making 2
Never found in quality cheeses 3 Both of the abov. 3
3. Rennet is: 4. Camembert:
A form of yeast used to make cheese 1 Is eaten younger than brie 1
An enzyme extracted from stomach linings 2 Is eaten older than brie 2
Neither of the above 3 Matures in the same time as brie 3
5. Rind is: 6. Cheese is nicest if made from milk produced in:
An important influence on flavor development 1 The spring and autumn 1
Not found on Camembert 2 The winter and summer 2
Never washed in making quality cheeses 3 Particularly wet summers 3
7. Camembert: 8. Camembert:
Never smells of mushrooms and yeast 1 Has slightly thicker mould than brie 1
Usually smells of mushrooms and yeast 2 Has slightly thinner mould than brie 2
Can sometimes smell like old socks 3 Has identical mould to brie 3
9. Triple Cream cheese: 10. Camembert rind is usually:
Feels greasy with you eat it 1 Comprised of soft white mould 1
Is made from milk with extra cream added later 2 Comprised of grey mould 2
Doesnt exist as a product 3 Orange 3
11. When choosing wine to drink with Camembert: 12. Camembert cheese is a:
Its critical to make sure its well chilled 1 Product where price usually influences quality 1
A chardonnay will suit, but never a red 2 Product where price is no indication of quality 2
A full bodied, earthy red is a sound choice 3 Product where all brands cost pretty the same 3
13. The inside of a Camembert that is ready to eat is: 14. When pressed, the rind of a Camembert should:
Creamy, buttery and smooth 1 Feel soft, but your finger leaves no mark or indent 1
Rubbery, buttery and soft 2 Feel soft to the touch, your finger leaving an indent 2
A little chalky 3 Feel quite firm 3
85
4.4.3 Self-confidence
High levels of self-confidence are thought to empower consumers to act on their
personal beliefs, regardless of their basis or accuracy (Rao and Olson 1990).
Traditional measures of consumer self-confidence are based on instruments
investigating personal self-esteem; evidence suggests links between self-esteem and
confidence in their own personal judgment (Bearden, Hardesty et al. 2001). Other
contributing factors to self-confidence include perceived locus of control, dominance
and previous experiences (Langer 1983; Obermiller and Spangenbert 1988; Lorr 1991).
Bearden et al. (2001:p122) summarizes consumer self-confidence as two higher order
factors describing the individuals perceived ability (1) to make effective consumer
decisions, including the ability to acquire and use information; and (2) to protect himself
or herself from being misled, deceived, or treated unfairly. These factors encompass 6
dimensions represented by a 31 item scale (Bearden, Hardesty et al. 2001). a review of
the 31 scale items revealed that many of them suggest a reflection of subjective
knowledge of a specified product category, making the scale more of a combined
instrument to self-confidence and purchase confidence rather than personal self-
confidence alone. A scale determining self-image was also reviewed and considered,
but rejected on the same premise (Malhotra 1981).
An objective of this study is to clearly isolate respondent subjective knowledge,
objective knowledge and personal self-confidence and measure each independently.
Bell (1967) suggests that the early studies investigating the influence of self-confidence
on consumer buying behavior were likely to rely on versions of a 10 item set developed
by Day and Hamblin (1964), designed to gauge generalized self-confidence specifically,
rather than confidence combined with (or reliant on) category familiarity. Given the
stated objectives, the items developed by Day and Hamblin (1964) were adopted for
this research with minor modifications updating terms and phrases. In earlier studies
86
using this scale, a highly self-confident person was defined as any respondent who
agrees with statements 1, 2, 3, 8 and 10; and disagrees with statements 4, 5, 6, 7 and
9 with no validation of these items as a scale undertaken (Day and Hamblin 1964; Bell
1967). However data analysis in this study incorporates testing these items for internal
and external validity to confirm the scale as an appropriate measure of self-confidence.
Table 4.4 Self-confidence scale items
Response format: 9 = strongly agree; 1 = strongly disagree

1. I feel capable of handling myself in most social situations.
2. I seldom fear my actions will cause others to have a low opinion of me.
It doesnt bother me to have to enter a room where other people have already
3.
gathered and are talking.
4. In group discussions, I usually feel my opinions are inferior. (r)
5. I dont make a very favorable first impression on people. (r)
When confronted by a group of strangers, my first reaction is always one of shyness
6.
and inferiority. (r)
7. It is extremely uncomfortable to accidentally go to a party wearing the wrong thing. (r)
8. I dont spend much time worrying about what people think of me.
9. When in a group, I very rarely express an opinion for fear of being laughed at. (r)
10. I am never at a loss for words when I am introduced to someone I dont know.
4.5 Survey data analysis
The model will be tested by investigating the nature and existence of relationships
between consumer knowledge and self-confidence and the utility levels determined for
each product attribute level determined from the conjoint analysis procedure. In
addition to examining these relationships, any differences in knowledge and self-
confidence levels between groups will be analyzed to identify patterns of response.
87
4.5.1 Correlations and factor analysis
Correlation is a measure of the relationship between 2 or more variables, with
coefficients ranging between +1.00, representing a perfect positive relationship, and -
1.00, representing a perfect negative relationship (Malhotra, Hall et al. 2002). When
there is a significant relationship between 2 variables (<0.05) and the coefficient is close
to +1.00 or -1.00 a strong linear relationship between them is indicated. Pearsons r is
a measure that assumes normality. When required, the non-parametric measure,
Spearmans r, was used. Correlations were used to test for relationships between the
independent variables of knowledge (subjective and objective) and self-confidence and
the contribution of the dependent variables of COO, price and acid/fat. Principal factor
analysis (which is based on correlation) was used to confirm that a latent variable is
being measured from a set of items (Malhotra, Hall et al. 2002). This procedure was
completed for the scales used (at each stage) to measure subjective knowledge and
self-confidence. Following transformation of reverse coded items and validity checks, a
mean score for each respondent reflecting their subjective knowledge and self-
confidence levels was calculated.
4.5.2 Objective knowledge
Scores for objective knowledge (each product) can be determined according the
number of correct answers to the multiple choice tests (Goldsmith, d'Hauteville et al.
1997). In order to compare objective knowledge and subjective knowledge levels a
standardized score for objective knowledge was computed.
88
4.5.3 Conjoint analysis
The analysis of product preferences and levels of expected product quality from the
conjoint analysis study was completed using SPSS Conjoint Analysis Statistical
Software. The analysis revealed which product attributes (cues) made the most
important contribution to the score given to each profile. Independent analysis was
completed at each stage of the research; firstly for the sample overall and then for
segments according to knowledge levels (per type) and self-confidence levels. From
this, correlation testing and comparison of means testing were done to test the stated
hypotheses.
4.6 Determining attribute importance
The objective of conjoint analysis is to produce a set of additive part worth utilities that
use ratings given to product profiles to derive attribute utility scores (SPSS-Inc 1997).
These are basically index numbers, corresponding to regression coefficients, measuring
how valuable or desirable a particular feature is to the respondent (SPSS-Inc 1997;
Dean 2004). The Ordinary Least Squares Regression (OLS) approach to ratings based
conjoint analysis is commonly used for this analysis as it offers a straightforward, yet
robust method of deriving the different utility values (used to compute attribute part
worths) for each respondent (Hair, Anderson et al. 1995; SPSS-Inc 1997). The OLS
model computes utilities using a dummy matrix of independent variables where each
indicates the presence or absence of an attribute level.
89
The dependent variable is the respondents score representing their assessment of the
profile as described (Kupiec and Revell 2001); the model is expressed as:
Zi = f(yi1 i2im) = B1i1(x1i1) + B2i2(x2i2) + ,, + Bmim(xmim)
Where:
B = the beta weights estimated in the regression

x = the matrix of dummy values identifying the levels of the factorial design, and
y = the ranking or rating evaluations of the respondent
Part worth statistics (utility values) will be both positive and negative, expressed on a
common scale summing to zero for each attribute; whilst utility values within an attribute
may be compared, they may not be compared across attributes (SPSS-Inc 1997; Dean
2004). Therefore, the most meaningful way to interpret the part worths is to analyze the
gaps between utility levels within each attribute (Hair, Anderson et al. 1995). A high
range value (gap) between utility levels within an attribute indicates that the participants
believe that change within that particular feature has significant impact on their overall
assessment of that offer. Hence attributes with greater ranges are those used most by
respondents to differentiate between profiles and have higher levels of relative
importance in the rating (Hair, Anderson et al. 1995; Kupiec and Revell 2001). In order
to compare the relative average importance of attributes, importance scores (I) are
calculated. This is done by taking the range between the lowest and highest utility
value for an attribute (i) and dividing it by the sum of all the utility ranges (SPSS-Inc
1997).
Therefore, importance is computed as:
Ii =
(Maxu i Minu i )
(Maxu i Minu i )
90
In summary, average importance values reveal the comparative importance (in
percentage terms) of each attribute to the respondents decision and the utility values
(part worths) illustrate which attribute levels are preferred and those that are avoided
(Hair, Anderson et al. 1995; Kupiec and Revell 2001). Furthermore, a score or
perceived worth can be computed for each hypothetical product to determine which
ones comprise the most important attributes at the most attractive levels. This can be
shown as:
Total Worth for Product ij..n = Part worth of leveli for factor1 +
part worth of levelj for factor2 + +
part worth of leveln for factorm

Hair et. al. (1995:p 561)
4.6.1 Non-parametric tests
The comparison of means requires a non-categorical measure. These tests can be
used in the analysis of scores and utility values determined in this study. In this
research, comparisons of means both within groups and independent of groups was
done, and relationships between variables were investigated. Where variables failed to
meet assumptions for normality, and the equivalent parametric test is not assumed
robust to abnormality, non-parametric tests were used (Malhotra, Hall et al. 2002).
Assumptions for normality and results of normality test procedures can be seen in
appendices specified in the relevant chapters.
91
The non-parametric test used for comparison of means in paired samples is the
Wilcoxon Sign Rank test. This test analyzes ranking data rather than comparison of
means, however, it presents a significant value similar to t-tests (Malhotra, Hall et al.
2002). The non-parametric test for independent samples used is the Kruskall-Wallis H-
test and the non-parametric correlation measurement employed is Spearmans r.
4.7 Validity of instruments
Due to the complexity in methodology caused by the multi-stage research design,
generally only the summarized results of validity testing are provided in the relevant
sections for each stage, with detailed analysis being provided in appendices.
4.7.1 Normality testing
Results of normality testing of variables are found in appendix 3 (tables A 3 to A 5). As
discussed, where testing indicated that variables fail to satisfy required assumptions for
the use of parametric data analysis tools, non-parametric equivalents were used.
4.7.2 Subjective knowledge and self-confidence scale
Measurement of Cronbach Alpha coefficient scores was used to evaluate scale
reliability. Scores range in value from 0 to 1 and the higher the score the more reliable
the scale is considered to be. Generally, a score of 0.7 or higher is deemed to be
acceptable (Malhotra, Hall et al. 2002). Cronbach Alpha coefficients for each
application of scales used are illustrated in the pertinent section of this chapter along
with results of respective tests of construct validity. Divergent and convergent reliability
test results are provided in the appendices.
92
4.7.3 Objective knowledge questions
The multiple choice questions used to measure objective knowledge were compiled
according to recommendations found in the literature and confirmed with industry
experts (Baldy 1993; Park, Mothersbaugh et al. 1994). Prior to use the tests were
independently reviewed for comment by external experts in their respective fields and,
in their professional opinions, found to be appropriate for the purpose (Ross 2004;
Smith 2004).
4.7.4 Conjoint analysis
Conjoint analysis requires checks for internal and external validity (Hair, Anderson et al.
1995). Internal validation was done by testing for goodness of fit of the estimated
model with the values of r2 and Kendalls tau statistics illustrated for each data sample.
These statistics (between 0 and 1) illustrate the relationships between the observed and
estimated preferences and should always be high (the closer to 1 the better); models
with poor fit are considered suspect (Malhotra, Hall et al. 2002). A further internal
validity check is provided by the comparison of utilities derived from profiles
incorporated in the fractional factorial design with those from hold out profiles (SPSS-
Inc 1997; Malhotra, Hall et al. 2002). External validation is based upon the degree to
which respondent samples represent the populations investigated and whether or not
the attributes and levels that were selected reflect credible hypothetical products (Hair,
Anderson et al. 1995). Comparisons of respondent sample profiles with the general
Australian population are provided and discussed for each stage of the research.
Attributes and levels were selected as a result of an extensive literature review, analysis
of qualitative data, consultation with industry experts, and a review of products available
in the relevant South Australian consumer markets.
93
4.8 Summary
Chapter 4 provided details of the quantitative methodology in respect to sampling
methods and the development of the data collection instruments. This included a
pertinent discussion and justification of measures used, validation procedures and data
analysis methods. Chapter 5 is devoted to the methodology, data analysis and results
specific to stage 2 of the research: the conjoint pilot study.
94
5 Quantitative conjoint pilot study
5.1 Introduction
Chapter 4 provided an overview of the research design for all stages of the study; the
common methodology employed in the quantitative aspects and the results of the
qualitative component of the research. This chapter will detail the methodology used for
the second stage of the study, the conjoint analysis pilot study, subsequent findings of
the data analysis and their implications on subsequent stages of the research.
5.2 Conjoint analysis survey (pilot)
The first quantitative component was a pilot study where the product profiles
determined by the experimental design were assessed via a self completed
questionnaire. The pilot study was conducted in order to ensure the appropriateness of
the conjoint analysis methodology, products, attributes/levels and measures of
knowledge and self-confidence. Gathering data and subsequent analysis revealed a
number of unforeseen and significant issues relating to respondent perceptions of the
products and cues used. This resulted in the need to re-evaluate aspects of the
questionnaire design and information and instructions to respondents. This aspect of
the study provided an invaluable opportunity in terms of rectifying these potentially
disastrous problems, whilst still validating the measures used.
5.2.1 Sample (pilot)
A convenience sample of 267 respondents agreed to participate, recruited from the
population of part-time students taking evening classes at the City Campus of TAFESA.
In order to encourage participation, those completing the questionnaire were given the
opportunity to enter their name in a lottery to win $250 in cash. Students participated
on a voluntary basis and completed the questionnaire during allocated class time.
95
5.2.2 Data collection instrument (pilot questionnaire)
The initial self administered questionnaire consisted of 4 parts described below; a copy
of the questionnaire can be seen in Appendix 2. In this version, all chardonnay profiles
were illustrated in part 1, followed by the subjective knowledge items for chardonnay
and questions relating to consumption habits. This layout and order was replicated for
the camembert profiles and subjective knowledge scales etc. The wine and cheese
objective knowledge tests followed and finally the scale determining self-confidence.
The objective knowledge questions were completed after the assessment of all product
profiles and subjective knowledge questions to ensure that reported levels of self-
assessed knowledge were not diminished as a result of an inability to confidently
answer the objective knowledge questions. Self-confidence items were answered last
followed by collection of demographic details. Figure 5.1 below provides a detailed
format.
Figure 5.1 Pilot study questionnaire format
1. Front Cover Survey information and contact details
2. Part 1. a) Wine product profiles (for rating)
b) Subjective knowledge scale (wine)
c) Product consumption habits and purchasing patterns (wine)
3. Part 2. a) Cheese product profiles (for rating)
b) Subjective knowledge scale (cheese)
c) Product consumption habits and purchasing patterns
(camembert)
4. Part 3. a) Objective knowledge questions (chardonnay)
b) Objective knowledge questions (camembert)
5. Part 4. a) Self-confidence scale
b) Demographics
6. Back Page Registration form to enter the draw for a $250 cash prize
96
5.2.3 Intrinsic and extrinsic cues - wine
In order to diminish objective quality, the intrinsic description was altered to reflect
increased levels of acidity resulting in wines that would be increasingly green or sour in
taste. As chardonnay wine of good quality does not contain unusually high levels of
acidity, average, above average and high signified the 3 levels. To determine realistic
price levels for a hypothetically high, mid and low priced chardonnay, industry and wine
retail sources were consulted. Countries chosen to represent the various levels of
quality were determined from the qualitative study. Table 5.1 illustrates the selected
attributes and levels and table 5.2 illustrates the SPSS fractional factorial design.
Table 5.1 Specification of chardonnay attributes and levels
Attribute Cue type Levels

France
COO extrinsic U.S.
Chile
$39.50
Price
extrinsic $14.00
per 700ml bottle
$ 6.00
Average for chardonnay
Acid level intrinsic Above average
High
Table 5.2 Fractional factorial design chardonnay
Profile COO Price Acid level Status

1 France $14.00 Average Holdout
2 U.S. $ 6.00 Average Holdout
3 Chile $14.00 High Design
4 France $ 6.00 High Design
5 U.S. $14.00 Average Design
6 Chile $39.00 Above average Design
7 U.S. $ 6.00 Above average Design
8 France $39.00 Average Design
9 Chile $ 6.00 Average Design
10 France $14.00 Above average Design
11 U.S. $39.00 High Design
97
5.2.4 Intrinsic and extrinsic cues - cheese
Fat has been found in previous research to be an important intrinsic attribute affecting
the superiority of cheese, with high fat (creamy textured) cheeses preferred (Hamilton,
Knox et al. 2000). Therefore a hypothetical triple fat camembert is the highest quality
intrinsically, with objective quality diminishing with reduced fat content. The countries
chosen to represent different levels of quality were established from the qualitative
component of the study. Current market prices (per 500 gram piece) for camembert in
Adelaide were used to set realistic price levels for cheeses of high, mid and low prices
respectively. Table 5.3 illustrates the attributes and levels and table 5.4 illustrates the
SPSS generated fractional factorial design.
Individual product profiles were numbered from 1 to 11 for chardonnay and camembert
respectively, with 3 to 4 profiles placed on each page (A 4) in order to keep the number
of pages used to a minimum. The respondents task is to consider the information
provided for each hypothetical product and rate each according to a metric scale from 1
to 10. The literature reveals that conjoint analysis metric rating scales have often been
anchored with likelihood of purchase extremes; however, this was not considered
appropriate in this case as the objective was not to measure purchase intension (Sagar
and Scofield 1982; Hair, Anderson et al. 1995; SPSS-Inc 1997). Whilst perception of
quality and likelihood of purchase are often linked, they are not automatically
synonymous (Zeithaml 1988). Consumers may consider a product to be of high quality,
but would not consider buying. This can be for a number of reasons: the cost may be
too high, they may feel a negative bias against the COO of the product, or they may not
believe that the product provides enough value for money (Zeithaml 1988; Dodds 1991;
Lee and Lou 1996). Similarly, consumers may not consider a product to be high quality,
but may purchase it nonetheless. This may be because there are few alternatives to
98
choose from, they cannot (or do not wish to) pay more for better quality or superior
product quality is simply not required (Dodds 1991; Lee and Lou 1996).
Table 5.3 Specification of camembert attributes and levels

France
COO extrinsic Canada
Argentina
$8.00
Price
extrinsic $5.00
(per 500 gram piece)
$3.00
Triple
Fat intrinsic Full
50% reduced
Table 5.4 Fractional factorial design camembert
Profile COO Price Fat Level Status

1 France $3.00 50% Reduced fat Holdout
2 Argentina $3.00 Triple cream Holdout
3 Canada $8.00 50% Reduced fat Design
4 Canada $3.00 Full cream Design
5 France $5.00 50% Reduced fat Design
6 Argentina $8.00 Triple cream Design
7 France $3.00 Triple cream Design
8 Argentina $3.00 50% Reduced fat Design
9 France $8.00 Full cream Design
10 Canada $5.00 Triple cream Design
11 Argentina $5.00 Full cream Design
Other studies using a conjoint analysis design have utilized a wide range of alternative
descriptors to represent the lowest and highest possible rating scores. These include:
worst possible versus best possible, unlikely to pay versus very likely to pay, dislike
extremely vs. like extremely, very undesirable vs. very desirable (Tan 1999; Murphy,
Cowan et al. 2000; Katoshevski and Timmermans 2001; Moy and Lam 2003;
Pelsmacker, Driesen et al. 2005). After reviewing tested alternatives, respondents in
the pilot study were asked to rate each profile from 1 to 10, where 1 represented very
99
undesirable and 10 represented very desirable (Moy and Lam 2003). These
descriptors were chosen because it was believed that respondents could provide an
opinion of the product that was clearly positive or negative, without linking this
assessment directly to personal liking or purchase intention. To collect a simple
measure of likelihood of purchase, after rating each profile participants were asked
whether or not they would consider purchasing the product as described by choosing
either yes or no. Figures 5.1 and 5.2 show examples of how the product attributes
and levels were presented to respondents.
)
Figure 5.2 Example of wine product profile (pilot)
Chardonnay 1
Acidity Average for chardonnay
Produced in France
Retail Price $14.00
Highly Undesirable Highly Desirable
1 2 3 4 5 6 7 8 9 10

Would you consider buying this product? Yes No
Figure 5.3 Example of cheese product profile (pilot)
Camembert 1
Made with 50% Reduced Fat
Produced in France
Retail Price $3.00
Highly Undesirable Highly Desirable
1 2 3 4 5 6 7 8 9 10

100
5.2.5 Questionnaire pre-test
A pre-test of the questionnaire was undertaken using part-time evening (N=24) students
attending lectures at TAFESA City Campus. This class of students was then eliminated
from participation in any other component of the study. The class was chosen at
random, controlling only for gender balance. An important objective of the pre-test was
to determine the time taken to complete the questionnaire, in order to inform lecturers
volunteering class time. The pre-test also provided an insight into the likely validity of
the scales and objective knowledge tests used to measure independent variables.
Also, respondent reactions to the profile assessment tasks and an examination of the
preliminary data from the conjoint experimental design were possible.
Table 5.5 Total variance explained (pre-test)
Cum % eigenvalues > 1 Factors

Subjective knowledge wine 65.726 2
Subjective knowledge cheese 74.922 2
Self-confidence 74.383 3
N of Cases = 24
Table 5.6 Reliability coefficients of scales (pre-test)
Reliability coefficients
Subjective knowledge wine N of items = 8 Alpha = 0.836
Subjective knowledge cheese N of items = 8 Alpha = 0.874
Self-confidence N of items = 10 Alpha = 0.861
N of Cases = 24
Cronbach Alpha scores for each scale are in excess of 0.7, with a high percentage of
variance explained for factors with eigenvalues greater than 1.00, indicating that the
instruments are suitable for the application (Malhotra, Hall et al. 2002) (Tables 5.5 and
5.6). The pre-test results for attribute utilities (chardonnay and camembert) are
consistent with the literature, indicating that both price and COO make contributions to
101
consumer opinions, albeit price was far less influential for camembert than for wine
according to this group (Tables 5.7 and 5.8).
Given the small sample size, these results could not be considered necessarily
indicative of future outcomes. The purpose was to ascertain whether respondents
perceived clear differentiation between the profiles resulting in wide score variances
across the set. The disparity between average importance values, for each product,
confirms that this is the case. Statistics for Kendalls tau and Pearsons r statistics are
moderate to high indicating that the model fit is within acceptable parameters (SPSS-Inc
1997) (Table 5.9). While the sample used in the pre-test was too small to allow for
more in depth analysis, these results were very encouraging in terms of participant
response to the tasks and the suitability of chosen measures. Based on this, the pilot
study was undertaken without alteration to the questionnaire format or content.
Table 5.7 Summary of part worths for chardonnay (pre-test)
Ave imp Utility Level Attribute

-0.287 Chile
29.082 -0.218 U.S. COO
0.505 France
-0.690 $ 6.00
43.878 0.505 $14.00 Price
0.185 $39.00
0.338 Average
27.041 0.060 Above average Acid
-0.398 High
100% N = 24
102
Table 5.8 Summary of part worths for camembert (pre-test)

-0.269 Argentina
33.333 -0.046 Canada COO
0.315 France
-0.074 $ 3.00
7.143 0.023 $ 5.00 Price
0.051 $ 8.00
-0.505 50% Reduced fat
59.524 0.537 Full Cream Fat
-0.032 Triple Cream
100% N = 24
Table 5.9 Conjoint analysis internal validity tests (pre-test)
Chardonnay Sig. Camembert Sig.

Kendalls tau (design) 1.000 .000 1.000 .000
Kendalls tau (hold outs) .667 .006 1.000 .000
Pearsons r 0.935 .0000 0.991 .000
2
R 0.874 .000 0.982 .000
N of cases = 217
5.3 Validation of research instruments (pilot)
5.3.1 Subjective knowledge and self-confidence scales
While the items measuring subjective knowledge had been validated as a reliable scale
in previous research, the items used to measure self-confidence had not undergone this
procedure before (Day and Hamblin 1964; Bell 1967; Flynn and Goldsmith 1999).
Previously, the items were used simply as a set of statements where agreement with
some items and disagreement with others resulted in the determination of whether a
respondent was deemed confident. The initial step in the validation process is to test
for internal reliability through determination of the Cronbach Alpha coefficient for each
103
measure; these are illustrated in table 5.10 for both applications of the scale with
coefficients ranging from 0.728 to 0.869 indicating sound reliability.
Table 5.10 Reliability coefficients of scales (pilot)
Subjective knowledge chardonnay N of items = 8 Alpha = 0.866
Subjective knowledge camembert N of items = 8 Alpha = 0.869
N of cases = 238
*See appendix 4 tables A 6 through A 31 for results of construct, convergent and
divergent validity testing all treatments.
Construct validity is related to generalizing and testing is undertaken to ensure that
scale items measure the intended construct in a consistent manner and that they
account for a high percentage of variance explained in a small number of solutions or
factors (Malhotra, Hall et al. 2002). Therefore, outcomes from testing should confirm
the meaning represented by the resulting variable matches the commonly held meaning
of the construct used in the area of study being undertaken.
Table 5.11 Construct reliability for subjective knowledge and self-confidence
Cum % of variance No of factors with

Scale
explained eigenvalues >1
Subjective knowledge chardonnay 69.6 2
Subjective knowledge camembert 70.4 2
N of cases = 238
The percentages of variances explained (factors with eigenvalues > 1) are 47.3% (self-
confidence), 69.6% (subjective knowledge chardonnay) and 70.4% (subjective
knowledge camembert) as seen in Table 5.11. In their studies validating this scale,
104
Flynn and Goldsmith (1999) ultimately removed 3 items in order to produce a 1 factor
solution that accounted for approximately 60% of the cumulative variance explained.
However, scrutiny of the factors generated for both chardonnay and camembert
subjective knowledge in this research, showed the second factor consisted of the 3
reverse coded items, not those removed by Flynn and Goldsmith (1999), suggesting a
coding effect rather than these items reflecting a different factor. Their removal did not
increase the Cronbach Alpha coefficient or result in a 1 factor solution. Further, their
deletion represented a significant loss of data. On balance, the decision was made to
accept a 2 factor solution for validation purposes (at this stage), given the high level of
cumulative variance explained. This choice is supported by Malhotra et al. (1996) who
suggest that if 60% or more of cumulative variance explained is accomplished in a small
number of interpretable factors, there is little benefit in losing data for what may be little
gain and a tidier solution. The rotated component matrices for both scales are shown
in table 5.12 and 5.13. However, the items constituting self-confidence required further
analysis to determine their suitability or otherwise as a valid scale measure of self-
confidence.
Table 5.12 Subjective knowledge chardonnay (pilot)
Component
Rotated component matrix chardonnay (pilot)
1 2
1. I know pretty much about chardonnay .769
2. I know how to judge the quality of chardonnay .746
4. Among my circle of friends, Im one of the experts on chardonnay .767
6. I have heard of most of the chardonnays around .722
8. I can tell if a chardonnay is worth the price or not .776
3. I do not feel very knowledgeable about chardonnay .806
5. Compared to most other people, I know less about chardonnay .848
7. When it comes to chardonnay, I really dont know a lot .840
Principal Component Analysis. Varimax rotations with Kaiser normalization (rotation in 3 iterations)
105
Table 5.13 Subjective knowledge camembert (pilot)
Component
Rotated component matrix camembert (pilot)
1 2
1. I know pretty much about camembert .702
2. I know how to judge the quality of camembert .756
4. Among my circle of friends, Im one of the experts on camembert .770
6. I have heard of most of the camemberts around .787
8. I can tell if a camembert is worth the price or not .791
3. I do not feel very knowledgeable about camembert .817
5. Compared to most other people, I know less about camembert .847
7. When it comes to camembert, I really dont know a lot .847
The self-confidence scale is comprised of 5 statements related to an individuals
concern regarding how others view them (or their actions) in various social situations,
and 5 statements that are similar but negatively phrased as a check against responses
to the positive items. The 2 factor solution (rotated) shows that the first factor includes
only the positively worded statements and the second factor the negatively worded
ones (table 5.14). Again as with the subjective knowledge scales a coding effect is
indicated. Testing was done to determine if the 5 positively worded items alone or the 5
negatively worded items would result in a higher total variance explained and
acceptable Cronbach Alpha coefficients and Bartletts and KMO statistics. Testing
found this was not the case, although, as expected, each set of 5 items produced a 1
factor solution with high and significant correlations between items. The next step was
to investigate removing items that were cross loading between both factors in order to
strengthen results. This was tested by removing (both singly and in combinations),
items 1, 2, 4 and 7; however the total variance explained was not significantly improved
and Cronbach Alpha coefficients were lowered.
106
Table 5.14 Self confidence factors (pilot)
Component
Rotated component matrix self confidence (pilot)
1 2
4. In group discussions, I usually feel my opinions are inferior .670
5. I dont make a very favourable first impression on people .715
6. When confronted by strangers, my first reaction is shyness and inferiority. .769
7. It is extremely uncomfortable to go to a party wearing the wrong thing. .550
9. When in a group I rarely express my opinion for fear of being laughed at. .655
1. I feel capable of handling myself in most social situations. .748
2. I seldom fear my actions will cause others to have a low opinion of me .592
3. It doesnt bother to enter a room where others already talking. .714
8. I dont spend much time worrying about what people think of me. .599
10. I am never at a loss for words when I am introduced to someone I dont know. .658
Self-confidence was measured in previous studies using these items by deeming those
respondents that agreed with statements 1, 2, 3, 8 and 10 while disagreeing with
statements 4, 5, 6, 7 and 9 to be self-confident. These responses indicate the
individual has little concern with managing their image as perceived by others, and
therefore feel no anxiety concerning self-presentation in social situations (Day and
Hamblin 1964). For example, for most items responses were quite polarized indicating
a relatively healthy level of ego amongst the sample; however, the items regarding
going to a party wearing the wrong thing, and being never at a loss for words when
meeting someone they dont know averaged the lower scores compared to the other
items. This suggests that, indeed, many people would be concerned about how others
perceive them in those situations but not others considered less daunting. Therefore,
these items may serve to discriminate between individuals with differing confidence
levels.
107
Accordingly, the next check was to determine if the participants responded in the same
way to both groups of items; if so, then a consistency in response would be evident.
Reverse coded items were transformed and a paired comparison of means test was
performed. The results of the Wilcoxon Signed Ranks Test (table 5.16), shows that
there is no significant difference between means scores for the 2 groups of items tested
demonstrating that respondents were consistent in their answers to all items supporting
the applications used previously.
Table 5.15 Wilcoxon Signed Ranks Test Self-confidence
Test Statistics - Wilcoxon Signed Ranks Test (pilot)b
Self Confidence - Disagree with 4, 5, 6, 7, 9

Self Confidence - Agree with 1, 2, 3, 8, 10
Z -1.589a
Asymp. Sig. (2-tailed) .112
a. Based on positive ranks.
b. Wilcoxon Signed Ranks Test
An important consideration when using factor analysis is the application of common
sense and logic in interpreting results and considering useful and pragmatic
applications. While the total variance explained by these items is only 47%, the
Cronbach Alpha coefficient is higher than 0.7 and the results of Bartletts and the KMO
tests are more than acceptable (See Appendix 4 table A15). The objective of the
measurement is to discriminate between respondents that possess higher levels of self-
confidence and those that have lower levels. This can be achieved by computing an
average score across all items (transforming those that are reverse coded) and
segmenting samples accordingly. Therefore, in conjunction with evidence of acceptable
results for tests of discriminate and convergent validity, the 10 items, as tested, were
maintained as a measure of self-confidence for subsequent stages of the study.
Further, due to satisfactory test results for the scales items measuring subjective
knowledge, these were also maintained unchanged. However, the decision to maintain
all items in both scales was taken on balance. This was only a pilot, and the results
108
would not be used in hypothesis testing. The scales consisted of a low number of items
and didnt contribute significantly to the length of the questionnaire and the opportunity
would exist to remove any items, if appropriate, to improve clarity, validity and reliability
in the analysis of scale data in subsequent stages.
In order to compute new variables representing these constructs for each respondent,
the following computations were made. First, negatively coded items were transformed,
and then an average was calculated from the sum of scores given to each item in the
respective scales. Therefore, the score could range from 1 to 9 for each variable with a
higher value corresponding to higher levels of subjective knowledge or self-confidence.
5.3.2 Objective knowledge tests (pilot)
Respondent feedback was positive in relation to these tests. The comments received
by the researcher indicated that those providing usable questionnaires found the tests
quite interesting and easy to complete; with some respondents asking for an answer
sheet to determine how well they had scored. In line with the literature, participants
generally found that their objective knowledge was lower than expected. However,
there were no suggestions that any of the questions were unreasonable or unrealistic
(Park, Mothersbaugh et al. 1994; Alba and Hutchinson 2000; Alba 2000). The feedback
provided no indication that respondents had not answered honestly and scrutiny of the
usable questionnaires did not signify that the questions had been misunderstood or
ignored; therefore, there was no evidence to believe that the internal or external validity
had been compromised.
109
5.3.3 Conjoint analysis fractional factorial design
(pilot)
Table 5.16 illustrates the results of tests of internal validity for computed part worths and
attributes average importance for the chardonnay and camembert profiles assessed.
High values for Pearsons r (and resulting r2) and Kendalls tau statistics indicate sound
internal validity. Significant changes were made to the cheese attribute, and to price
levels for both products to improve external validity as a result of the pilot survey results
and respondent feedback (Hair, Anderson et al. 1995). These are discussed in the
relevant section of this chapter.
5.4 Results and discussion (pilot)
5.4.1 Sample (pilot)
Table 5.16 Conjoint analysis internal validity tests (pilot)
Chardonnay Sig. Camembert Sig.

Kendalls tau (hold outs) 1.000 .0001 1.000 .0000
Pearsons r 0.996 .0000 0.990 .0000
R2 0.992 .0000 0.9801 .0000
N of cases = 217
Initially 267 questionnaires were returned for analysis. Of these, 238 could be used to
measure knowledge and self-confidence and validate these instruments in subsequent
stages of the study. In all, 29 questionnaires were eliminated from scale and
knowledge analysis because only the first or second pages were completed or entire
sections of the questionnaire were ignored. From this group, a further 21
questionnaires were eliminated prior to testing the conjoint designs, leaving 217 that
could be used for test of the conjoint analysis designs for wine and cheese.
110
Questionnaires were rejected as a result of the second screening for 1 or more of the
following reasons.
1. All product profiles (chardonnay and/or camembert) were rated 1 and the box
for No was ticked, signifying they would not consider purchase of the product
as described. This was a clear message that the profiles had not been
assessed according to the cues provided, with all profiles being rejected by way
of the lowest possible rating and purchase not being considered.
2. Only the first few profiles for one or both products were completed, indicating
fatigue, annoyance or lack of interest.
3. Strong notations made on the questionnaire signifying the respondents
vehement disapproval of the absence of Australian wine/cheese profiles to
consider. Comments such as I only buy Aussie!!, and Why arent there any
Australian wines?? were typical amongst this group. This was supported by
similar comments made by some respondents returning questionnaires to the
researcher personally.
A significant proportion of the original sample (18.7%) failed to return usable
questionnaires due to a lack of interest and/or a hostile response to the design
components being tested. These comments were, to some degree, foreshadowed in
the focus groups wherein 1 or 2 respondents made comments during the taste tests
such as I feel disloyal even trying foreign wine and I have no experience with any
wines that arent Australian. Clearly, these negative and/or ambivalent reactions
required strategies to engage the respondent in the research methodology and
overcome their ethnocentric responses.
111
Table 5.17 shows the demographic composition of respondents and equivalent
demographics in the general population as sourced from the Australian Bureau of
Statistics (2001 Census).
Table 5.17 Sample demographic profile
Gender % ABS % Age % ABS %

Male 45 49.8 18 yrs to 25 yrs 21.0 11.5
Female 55 50.2 26 yrs to 35 yrs 32.0 14.2
36 yrs to 45 yrs 27.9 15.0
46 yrs to 55 yrs 12.8 13.5
Over 55 yrs 6.4 21.9
Household income % ABS % Level of Education % ABS %

Less than $25, 000 17.8 27.7 High school certificate 35.6 30.0
$25,000 to $45,000 24.7 26.5 Diploma / trade 42.9 21.8
$46,000 to $65,000 25.6 16.8 Bachelor's degree 16.4 9.7
Over $65,000 31.5 29.0 Post graduate degree 5.0 3.2
Source: (ABS 2003 - 04) Household Characteristics (No. 65230)
In comparison with demographics describing the general population of Australia, the
sample is found to exhibit a female skew of 5% and significant variations from the
general population in terms of age. However, given the convenience nature of the
methodology employed, such biases are expected and rather less severe than in other
types of convenience samples (James and Sonner 2001; Peterson 2001). Further, in
consideration of the products tested, the age and demographic skew is likely to provide
a more representative sample of consumers in these markets. Also, whilst a student
sample was employed only students studying evening classes were eligible for
participation, with previous research demonstrating that samples determined using this
methodology (due to the diverse demographics typical of these students) they are more
likely to constitute a group representative of the general population than traditional
student samples (James and Sonner 2001; Peterson 2001).
112
5.4.2 Conjoint analysis results (pilot)
Tables 5.18 and 5.19 show the individual utilities of each attribute at the specified levels
with an averaged importance for the attribute overall, illustrating its contribution towards
the final expectation of quality. The results illustrate that for chardonnay, COO was
found to be more influential in affecting quality expectations than described acidity.
While correctly assessing which levels of acidity are less desirable, the intrinsic cue was
not believed by respondents to be as important as the extrinsic cue of COO in
determining expected quality. Further, while price was considered less important than
acidity the difference in the degree of influence between them is only 2% illustrating a
similar level of influence in the rating decision. France was clearly believed to provide
the most desirable chardonnay, but interestingly, little difference in opinion exists
between wine from the U.S. and Chile. This result is surprising given that the U.S., as
an industrialized nation, should have been believed to produce higher quality (Chao
1992). This outcome may be a reflection of concern voiced in the focus groups
regarding perception of high pesticide levels and genetic modifications associated with
food products from the U.S. The low esteem placed on the Chilean product, on the
other hand, is in line with expectations given the responses recorded in the focus
groups towards products from South America. The results relating to the wine price
levels are in line with the literature, in that a particularly low price is likely to be
associated with correspondingly low quality (Zeithaml 1988; Jover, Montes et al. 2004).
The relatively low score for the highest price given may be an indication of
unwillingness to pay this amount for a bottle of chardonnay (particularly from the U.S. or
Chile), irrespective of expected quality, resulting in the mid-priced option being deemed
the most attractive by respondents. The other potential reason for this result is the use
of Highly Undesirable and Highly Desirable as anchor points. These were used as
expected surrogates for the term quality; however, it is possible to believe something to
113
be of high quality yet not desirable particularly if there is an ethnocentric influence on
participants as was apparent with this group (Herche 1994).
Table 5.18 Summary of part worths for chardonnay (pilot)

-0.3404 Chile
38.42 -0.2354 U.S. COO
0.5758 France
-0.4744 $ 6.00
30.31 0.2486 $14.00 Price
0.2258 $39.00
0.3521 Average
-0.3937 High
100% N = 217
Table 5.19 Summary of part worths for camembert (pilot)

-0.3564 Argentina
60.55 -0.0615 Canada COO
0.4179 France
-0.1101 $ 3.00
13.96 0.0418 $ 5.00 Price
0.0683 $ 8.00
0.0418 50% Reduced fat
25.49 0.1412 Full Cream Fat
-0.1839 Triple Cream
100% N = 217
As with wine, COO was considered the most important attribute when assessing
camembert quality; however, respondents acknowledged fat as more important than
price in forming their opinions (Table 5.19). In line with previous research relating to
desirable food attributes, respondents considered the highest fat content to be the least
desired. This is despite the fact that higher fat levels result in cheeses that are creamier
in texture (enhanced mouth feel) and generally better flavored than low fat cheeses.
This suggests a social desirability bias where high levels of fat in any food may be
considered undesirable regardless of its actual positive association with greater overall
114
quality in terms of taste (Hamilton, Knox et al. 2000). In this test, the importance given
to price is relatively low; perhaps because respondents perceived little difference in
financial sacrifice between the levels described thus diminishing the influence of price
overall (Hair, Anderson et al. 1995).
5.4.3 Results knowledge and self-confidence (pilot)
The reliance on extrinsic cues is not surprising given that the general level of objective
knowledge amongst the group was very low. Our data shows that respondents
achieved a mean score of only 3.05 correct answers of the 14 asked in the objective
knowledge test for wine, with 98% of respondents scoring 7 correct answers or less.
Results from the objective knowledge test for cheese were similar, with an average of
only 3.42 correct answers in that test and 91% of respondents scoring 7 correct
answers or less.
Table 5.20 Equivalent mean scores for knowledge and self-confidence
Where 0 equals the lowest score and 9 equals the highest score attained.
Mean SD
Subjective knowledge chardonnay 4.17 1.69
Objective knowledge chardonnay (standardized) 1.96 1.36
Subjective knowledge camembert 4.16 1.76
Objective knowledge camembert (standardized) 2.20 1.82
Self-confidence 6.10 1.14
N = 238
The scores for subjective knowledge (Table 5.20) suggest that while respondents
clearly did not see themselves as product experts in most cases (for either product),
the corresponding scores for objective product knowledge are considerably lower than
even the rather modest self-assessed levels. These low levels of knowledge (both
objective and subjective) are likely to be important contributing factors to the
demonstrated reliance on extrinsic cues and subsequent discounting of intrinsic cues,
115
irrespective of their impact on objective product quality. In contrast, self-confidence
scores indicate that, generally, respondents exhibited a reasonably high level of self-
confidence. Hence, the extrinsic cues provided are less likely to have been found
intimidating or lead to a diminished belief in personal opinions in subsequent testing.
5.5 Pilot study results summary
Analysis shows that the focus group findings are remarkably consistent with conjoint
analysis results and that that the direction and magnitude of the influence of the main
attributes are predicted by the literature cited. The results of the conjoint analysis,
therefore, confirm the qualitative research findings. The results also show that the
objective knowledge of respondents, in particular, is much lower than anticipated. This
suggests that respondents may not be able to evaluate intrinsic cues based on high
levels of objective knowledge and therefore must resort to extrinsic cues requiring less
specific knowledge, such as price and COO as indicators of quality. Moreover,
comparatively high levels of self-confidence would indicate that opinions of quality,
however they may be derived, are likely to be defended. However, the information or
knowledge used to form this opinion or expectation may be flawed and lead to an
inaccurate assessment.
5.5.1 Needed changes to questionnaire (pilot)
All measures used in the questionnaire satisfied tests for internal and external validity,
confirming their application in subsequent stages of the research. Instruments
measuring knowledge and self-confidence remained unaltered, as did the conjoint
analysis fractional factorial design comprised of 9 design and 2 hold outs profiles (per
product) testing a 3 attribute (COO, price and fat/acid) by 3 level configurations.
However, significant changes were made to the next version of the questionnaire in
terms of layout and the order of the various parts. Enhancements were also made to
116
the appearance and presentation of the questionnaire in order to address respondent
boredom and communicate a sense or importance to the tasks. The redesign of the
questionnaire was supported by a comprehensive respondent briefing explaining the
absence of local products in the design. Future participants needed to be reassured
that assessing foreign sourced goods does not suggest that the researcher or
respondent support them over Australian wines and cheeses. This was scripted for
inclusion in the methodology used in the sensory experiments. Further, it was obvious
that a comprehensive briefing be held with each group participating in the taste test
experiments to overcome potential resistance and hostility towards foreign sourced
products. These changes and improvements are discussed in the next section.
117
6 Sensory experiment methodology
6.1 Introduction
Chapter 5 provided an overview of methodological aspects common to all quantitative
data components of the study as well as the results of the conjoint analysis pilot survey.
An overview of the rationale underpinning the methodology used to conduct the sensory
experiments is provided in this chapter.
6.2 Sample (sensory)
The tasting experiments were conducted at the sensory laboratory (Waite Campus) at
the University of Adelaide during the months of April and May 2006. Participants were
recruited from students undertaking evening classes at TAFESA City Campus,
excluding those that had participated in the pilot study (Appendix 5 Information pack
and registration form). In addition, members of the South Australian Canine Association
and members of the general public were also invited to take part during a 3 day dog
show held at the Wayville Showgrounds, Adelaide, South Australia in March 2006. The
third strategy used to increase participation was a broadcast email invitation and
electronic registration form (identical to the printed version), sent to all staff members of
the TAFESA Southern Region.
The info pack provided contained some background information about the study and a
registration form. The registration form showed a schedule of tasting times and people
were asked to submit two different times to attend tasting sessions, nominating first
and second preferences. This allowed for some flexibility in scheduling if a first choice
was not available. The next step was to return the registration form (by mail via the
reply paid envelope supplied or electronically via email). During the month of March
2006, registrations were recorded and respondents allocated and confirmed to a
118
specific tasting time (Appendix 6 Tasting registration spreadsheet and respondent
confirmation notice). Due to the need for respondents to travel to the sensory
laboratory and the extended time that would be required to take part in the experiment,
a cash incentive of $30 per person was offered. The payment was made to each
individual upon completion of the self-administration of a questionnaire measuring
knowledge and self-confidence and the taste testing experiment. In total 263
respondents participated across 43 taste testing sessions conducted in April and May of
2006. The sensory laboratory consisted of 9 individual tasting booths, with the capacity
to accommodate 1 or 2 additional respondents in the preparation area if required. Each
session lasted approximately 2 hours and included between 4 and 11 respondents.
6.3 Data collection instrument (sensory)
Upon review of the data and respondent feedback obtained from the pilot study, further
meetings were held with industry experts to review and improve the questionnaire
content and design. The fractional factorial orthogonal design developed for the pilot
study was used again in this stage of the research, but changes were made to some
attributes and levels. For example, the price levels for both products were altered to
better reflect current consumer market characteristics. Brie was used in the taste
testing experiments rather than camembert to take advantage of products that were
more cost effective, easily available and of 3 distinct and progressive levels of objective
quality. Also, a number of changes were made to the questionnaire layout and
appearance to address issues identified as problematic in the pilot study. These
included potential respondent ill will, boredom and fatigue. Amendments were also
made to facilitate the experimental procedures. These modifications are discussed in
detail in this chapter.
119
6.3.1 Intrinsic and extrinsic cues wine
Wine price levels were revisited following discussions with Mr. Jim Smith (wine expert)
and a perusal of current wine prices at retail outlets. Consequently, the decision was
made to maintain the lowest price of $6.00 per bottle, to increase the mid-priced option
slightly to $16.00 and to increase the premium price above the threshold of $50.00 to
$53.00 (all for 750ml). The countries chosen to represent the various levels of quality
were unchanged from the pilot study questionnaire. In order to diminish objective
quality, the Chapel Hill unwooded chardonnay was treated with tartaric acid. To
represent above average acidity, 0.5 grams of tartaric acid was added to each liter of
chardonnay and 1.0 grams was added for high acid (Baldy 1993). Table 6.1 illustrates
the selected attributes and levels and Table 6.2 illustrates the modified SPSS fractional
factorial design. In order to confirm that consumers could, first, discriminate between
the unaltered, acidic and very acidic wine and second, rank the quality levels
appropriately, further validation tests were done.
6.3.2 Triangle and paired sample tests
Research shows that consumers differ in their ability to accurately and reliably
discriminate between product samples that differ in quality (Roper 1969; Buchanan,
Givon et al. 1987). As a result, testing methodologies such as the triangle test and
paired sample test have been developed to differentiate between participants
according to their taste sensitivity, product assessment reliability and to establish
consumer sensitivity thresholds (Roper 1969; Puisais and Chabanon 1974; Buchanan,
Givon et al. 1987; Seaman, Hughes et al. 1993; Gatchalian 1999). Taste panel training
has also been used to enhance consumer objective product knowledge and increase
respondent accuracy in evaluation of intrinsic product quality (Seaman, Hughes et al.
1993).
120
However, in this research, respondents have not been coached, or trained, to enhance
their knowledge of wine or cheese. Nor have they been tested to establish their natural
abilities to differentiate between product samples. These potential respondent
limitations are not relevant in this study as the researcher is not attempting to quantify
consumer sensitivity levels or identify, specifically, which product profile is considered
superior in direct comparison to any other. Rather, the purpose of the study is to test
the relationships identified between the variables shown in Figure 3.1. Furthermore,
typical consumers in the market place do not receive formal training or undergo
sensitivity tests prior to passing judgments on the products they buy. This experiment
seeks to investigate the opinions of ordinary shoppers and to quantify the respective
influence of extrinsic cues on assessed product quality as determined through their
sensory evaluation. Therefore, product training, sensitivity and reliability tests are not
formally incorporated into the methodology for the taste test experiments. However, it
was critical to ensure that consumers could differentiate between the acid levels and
could rank the treated and untreated wines in the correct order of objective quality;
therefore, triangle and paired sample tests were conducted for these purposes.
In the first of these tests, a convenience sample of 6 volunteers participated in triangle
taste testing (Buchanan, Givon et al. 1987). This was used to confirm that the different
manipulations in taste were above usual sensitivity thresholds. Participants were given
3 samples of wine whereby 2 treatments were the same and 1 was different.
Respondents were not asked for judgments regarding the quality of the samples, but
merely to identify the odd one that didnt taste like the other 2. A rotation of samples
containing all 3 levels of acid was done to ensure that all combinations were tested, with
participants cleansing their palates with water and dry biscuits between tasting. In each
case, respondents had little difficulty identifying the one sample in 3 that was different.
121
In the second test, a sample of 9 volunteers was recruited by convenience to undergo
the paired sample testing procedure. The 3 levels of wine were in plain glass
receptacles marked Wine 127 (untreated good quality), Wine 263 (0.5 grams acid
average quality) and Wine 438 (1.0 grams acid poor quality); random coding was
used to label the wine so that no order or rank was implied for any particular sample.
Respondents were each given 2 samples of 30ml of wine, and asked to report which
was their preferred sample of the pair (Roper 1969; Buchanan, Givon et al. 1987).
Respondents were permitted to taste across the samples if they wished, but were not
permitted to discuss their opinions with others. Again, the tasting panel was instructed
to cleanse their palates with water and a dry biscuit before repeating the test a second
and a third time with different pairs of samples. A rotation of combinations amongst the
group meant that each person tasted each combination. Analysis of results showed
that respondents considered the unaltered wine to be most preferred as compared to
the other 2, with the wine containing 0.5 grams of acid preferred over the higher acid
treatment. Feedback from the panel showed that only the unaltered wine was found
consistently to taste good, although some respondents didnt mind the wine where 0.5
grams of acid had been added, indicating either a higher sensitivity threshold to acid or
a liking for sharper and sourer tasting products. However, descriptors given for the
samples containing 1.0 grams of acid included Sour! Vinegar, Salad Dressing and
Foul!. These results confirmed the acid levels chosen to be relatively easy for
consumers to distinguish from each other and to effectively represent 3 differing and
progressive levels of objective wine quality. Tables 6.1and 6.2 illustrate updated
attribute levels and fractional factorial design. In order to eliminate any inference of
rank or order, random numbers were also assigned to all sample profiles.
122
Table 6.1 Specification of chardonnay attributes and levels

France
COO extrinsic U.S.
Chile
$53.00
Price
extrinsic $16.00
per 700ml bottle
$ 6.00
Untreated
Acid level intrinsic + 0.5 gram tartaric acid per liter
+ 1.0 gram tartaric acid per liter
Table 6.2 Fractional factorial design chardonnay
Profile COO Price Acid level Status

253 France $16.00 Untreated Holdout
582 U.S. $ 6.00 Untreated Holdout
481 Chile $16.00 + 1.0 gram Design
696 France $ 6.00 + 1.0 gram Design
595 U.S. $16.00 Untreated Design
924 Chile $53.00 + 0.5 gram Design
152 U.S. $ 6.00 + 0.5 gram Design
823 France $53.00 Untreated Design
494 Chile $ 6.00 Untreated Design
950 France $16.00 + 0.5 gram Design
279 U.S. $53.00 + 1.0 gram Design
6.3.3 Intrinsic and extrinsic cues - cheese
When sourcing cheese products for the experiment, it was discovered that brie cheese
(a soft cheese with a mould rind very similar to camembert) could be purchased in 3
equally discriminate fat levels; full cream, double cream and triple cream. By using
these products, a repeat of the halo effect seen in the pilot data towards the 50%
reduced fat cheese could be avoided (Hamilton, Knox et al. 2000). Advice from the
supplier confirmed the full cream brie as a very economical no brand product of the
lowest quality available in bulk. The double cream brie put forward was reasonable
123
quality product imported from France and the triple cream brie was of excellent quality
and produced in the Margaret River region of Western Australia. Further to discussions
with the cheese wholesaler, it seemed logical to change the price levels accordingly and
use the actual recommended retail prices per kilo charged for these particular cheeses
as external validity is improved when market conditions are replicated in the
experimental design (Hair, Anderson et al. 1995). The countries used in the pilot to
represent varying quality levels were maintained unchanged. Table 6.3 illustrates the
attributes and levels and Table 6.4 illustrates the fractional factorial design; as with the
chardonnay product profiles random numbers were assigned to all brie samples.
Table 6.3 Specification of brie attributes and levels

France
COO extrinsic Canada
Argentina
$69.95
Price
extrinsic $49.95
(per kilo)
$28.95
Triple cream
Fat intrinsic Double cream
Full cream
Table 6.4 Fractional factorial design brie
Profile COO Price Fat level Status

810 France $28.95 Full cream Holdout
139 Argentina $28.95 Triple cream Holdout
367 Canada $69.95 Full cream Design
266 Canada $28.95 Double cream Design
709 France $49.95 Full cream Design
380 Argentina $69.95 Triple cream Design
735 France $28.95 Triple cream Design
393 Argentina $28.95 Full cream Design
621 France $69.95 Double cream Design
178 Canada $49.95 Triple cream Design
507 Argentina $49.95 Double cream Design
124
6.3.4 Change of rating scale anchors
Review of the questionnaire design also included re-examination of the anchor points
for the metric scale used by respondents to rate quality. This included an exploration of
possible alternative descriptors of quality, as defined for this research. In the pilot
study, highly desirable versus highly undesirable were used and while consumers
may desire high quality and thereby establish a link between the two terms, pilot data
results indicated this could not be assumed. In fact, many respondents seemed to
experience conflict when considering the term desirable in association with foreign
products. For them, finding a foreign product desirable is akin to showing disloyalty to
Australian producers. As cited in chapter 5, this reaction was foreshadowed in the
focus groups where some individuals voiced feeling guilty even tasting foreign wines
and others indicated they had never before sampled non-Australian wines or cheeses
and had no intention of doing so. In light of this, it is understandable that some
Australian consumers would not find foreign wines or cheeses of any quality desirable,
hence the required link between quality and desirability cannot be assumed. For
these consumers, the rating given with these anchor points is a reflection of personal
values rather than an assessment of objective quality. In this research, it was critical
that respondents rate profiles according to objective quality criteria: as an impersonal
judgment; where the term only describes product superiority or excellence, devoid of
moral considerations or behavioral assumptions, rather than a direct indication of
buying intention (Zeithaml 1988). Therefore it was deemed necessary to change the
anchor points to low quality versus high quality in the sensory experiment
questionnaire. Figures 6.1 and 6.2 show examples of chardonnay and brie sample
profiles respectively.
125
Figure 6.1 Example of wine product profile (sensory)
Chardonnay 253
Produced in France
Retail Price $16.00
Low Quality High Quality
1 2 3 4 5 6 7 8 9 10

Figure 6.2 Example of cheese product profile (sensory)
Brie 810
Produced in France
Retail Price 28.95 per kilo
1 2 3 4 5 6 7 8 9 10

6.3.5 Knowledge and self-confidence
The scales used to measure subjective knowledge and self-confidence remained
unchanged. The objective knowledge test for wine was also unaltered; however, with
the change from camembert to brie, slight changes to the cheese test were made to
reflect the different product. As these soft mould cheese varieties share many common
attributes and characteristics, the alterations were minimal involving some modifications
in terminology and swapping the variety name camembert to brie where necessary.
126
6.3.6 Changes to questionnaire layout and format
It was considered important to communicate to participants that their opinions had real
value and that by taking part in the experiment they were making an important
contribution to the study. This strategy was used to encourage honest responses and
prolonged, focused attention to the tasks. It was apparent from participant feedback
that, by receiving payment, many felt they had taken on a job and, as a result, were
taking the experiment quite seriously.
The questionnaire used for the sensory experiment incorporated extensive instructions
for respondents, supported by a detailed and scripted briefing to each group.
Completion of various parts of the questionnaire was separated by changes in location
providing needed breaks in concentration. A booklet on A5 paper (landscape) was
printed in color, spiral bound and decorated with wine, grape and cheese graphics to
make it more attractive and interesting. The layout allowed each profile to be presented
on its own page, consistent with the provision of individual samples for tasting. Using
this format reduced the risk of respondent exposure to descriptions of samples not yet
assessed, reducing carryover effects. In order to minimize fatigue and desensitization
of the palate, wine and cheese samples were presented in a mixed order (Schifferstein
1996). This methodology, in addition to random numbering of profiles, makes it more
difficult for respondents to anticipate the attributes of the next sample or accurately
assess how many tasks they have already completed. Participants were also provided
with fresh water and dry biscuits and instructed to cleanse their palate between
samples.
6.3.7 Group briefings
Prior to commencing the experiment, a strict protocol of briefing and discussion took
place. Each group was assembled in a meeting room where the researcher provided
127
introductions and background information on the study. The group was told about the
format of the questionnaire booklet and the type and number of products they would
taste. As each section of the questionnaire was described, information regarding when
respondents were to stop and wait for further instructions was provided. During this
briefing, a scripted outline describing the sensory laboratory environment was
communicated and questions from the group were encouraged. While the setting was
somewhat formal, due to the small number of participants in each group the
atmosphere was kept friendly and a relaxed tone was encouraged.
A critical aspect of the briefing involved communicating the rationale behind the
absence of Australian products in the experiment. It was confirmed with respondents
that Australian wine (in particular) is certainly what most Australians consumers prefer
and that is was for this reason that local wines couldnt be used. Australian consumers
will naturally have specific local wine and cheese brands they buy and prefer and this
extensive product knowledge results in a bias detrimental to achieving objective
responses. Therefore, it was desirable to use foreign products unfamiliar to Australian
consumers in order to prevent this from happening. Respondents were assured that
the research was not being conducted to gain market intelligence that would assist in
bringing foreign products to Australia and was not, in any way, a reflection of support for
foreign products over Australian goods.
Following this discussion, questions were taken and participants were given their
booklets and required to sign the release/consent form. They then completed the self-
confidence and subjective knowledge scales while still in the meeting room. When
everyone had answered these sections, they moved through to the individual sensory
booths to complete the tasting component. As individuals finished tasting their samples
(each person working at their own pace), they returned to the meeting room and
completed the objective knowledge tests and provided their demographic details. At
128
this time the researcher met with respondents again, asked for feedback, distributed the
cash incentive, answered questions and thanked each person individually for taking
part. The sections of the amended questionnaire are described in Figure 6.3 below and
a copy of the questionnaire is provided in Appendix 7.
Figure 6.3 Amended questionnaire structure:
1. Front Cover Survey information, date, time and session number
2. Research information and participant release/consent form
3. Part 1. a) Self-confidence scale
b) Subjective knowledge chardonnay and purchasing patterns
c) Subjective knowledge brie and purchasing patterns
d) Instruction to close their booklet and wait to go into the sensory
lab
4. Part 2. a) Information and instructions regarding tasting samples
b) Chardonnay and brie samples
5. Part 3. a) Objective knowledge questions (chardonnay)
b) Objective knowledge questions (cheese)
c) Demographics
6.3.8 Conducting the experiment
In a typical testing day (comprised of 3 sessions each, scheduled 3 hours apart) on
average over 275 samples each of chardonnay and brie were prepared. In order to
ensure that product quality was consistent, trays of chardonnay samples were poured
less than 1 hour before tasting and each glass was covered with a plastic lid and
stored in a refrigerated room. In order to maintain consistency in cheese samples
cheeses were picked up daily from the supplier, with each cheese examined for
consistent ripeness each time. Brie samples were cut approximately 2 cm square, with
the mould crust only on one side. Samples were stored in sealed containers in the
129
refrigerator until a few minutes before each tasting session. Once on the individual
sample trays, the brie was covered loosely by moisture proof paper to ensure the
cheese did not become dry. Each respondent sat in an individual tasting booth and
signaled their readiness for a sample by using a switch inside the booth. The switch
illuminated a light in the kitchen preparation area specific to that tasting booth and by
sliding a small door open at counter height, a new sample was swapped for the empty
plate or glass that held the previous sample. A tasting schedule posted over each
booth ensured that samples were presented in the correct order. The following
selection of photos provides illustrations of product sample preparation and
experimental procedures. Figures 6.4 through 6.11 provide illustrations of sample
preparation and procedures.
Figure 6.4 Numbering wine glasses Figure 6.5 Pouring wine samples
Figure 6.6 Treated and untreated chardonnay Figure 6.7 Preparing brie samples
130
Figure 6.8 Preparing trays for a tasting session
Figure 6.9 Sample order Figure 6.10 Tray with samples Figure 6.11 Experiment in progress
131
6.4 Validation of research instruments (sensory)
The initial step in the validation process was to test for internal reliability through
determination of the Cronbach Alpha coefficients for each measure. These are shown in
Table 6.5 with coefficients ranging from 0.718 to 0.909 indicating sound reliability.
Table 6.5 Reliability coefficients of scales (sensory)
Subjective knowledge brie N of items = 8 Alpha = 0.899
N of cases = 263
The respective percentages of variances explained (for all factors with eigenvalues > 1)
are 55.9% (self-confidence), 61.7% (subjective knowledge chardonnay) and 63.4%
(subjective knowledge brie). For subjective knowledge (both products) a 1 factor
solution was achieved with a relatively high percentage of total variance explained,
consistent with results from previous studies (Flynn and Goldsmith 1999; Malhotra, Hall
et al. 2002). A higher overall percentage of variance explained for the self-confidence
scale items was found for this sample of respondents compared with the pilot study
data, after a 3 factor rotated solution (Table 6.6).
Table 6.6 Construct reliability for scales (sensory)
Cum % No of factors
Scale of variance with
Subjective knowledge brie 63.4 1
N of cases = 263
132
Table 6.7 Self-confidence factors (sensory)
Component
% of total variance explained
Rotated component matrix self-confidence
1 2 3
29.94% 15.86% 10.12%
In group discussions, I usually feel my opinions are
4. .771
inferior
5. I dont make a very favorable first impression on people .712
When confronted by strangers, my first reaction is
6. .652
shyness and inferiority
It is extremely uncomfortable to go to a party wearing
7. .514
the wrong thing
When in a group I rarely express my opinion for fear of
9. .712
being laughed at
I feel capable of handling myself in most social
1. .653
situations
I seldom fear my actions will cause others to have a low
2. .802
opinion of me
It doesnt bother to enter a room where others already
3. .716
talking
I dont spend much time worrying about what people
8. .702
think of me
I am never at a loss for words when I am introduced to
10. .737
someone I dont know
As with the pilot group, the factors in the self-confidence scale appear to be linked to
coding in that the first factor comprised all the reverse coded items. The second and
third factors comprised the remaining 5 statements. Successive testing was undertaken,
whereby items in the first factor were tested in isolation, followed by only items in factor
2, and lastly, with the combined items from factors 2 and 3. In each case, a 1 factor
solution was achieved; with varying levels of total variance explained and reliability (see
Table 6.8). The solution comprised of all positively coded items accounted only for 42%
of total variance explained and while the scale comprised of items 1 3 provided a
stronger result with 58.6%, the Cronbach Alpha coefficients in both cases were below
the acceptable reliability threshold.
133
Table 6.8 Reliability tests of self-confidence items (sensory)
Items % of variance explained Cronbach Alpha

4, 5, 6, 7, 9 47.8 0.716
1, 2, 3 58.6 0.620
1, 2, 3, 8, 10 42.3 0.620
N of cases = 263
As discussed in the section detailing the pilot study results, these scale items explore
an individuals reaction to potentially stressful social situations and investigate their
degree of anxiety in terms of managing their personal impressions on others. Once
again, the lowest scoring items (on average) in the scale were item 7 (related to
wearing the wrong thing to a party) and number 10 (regarding never being at a loss for
words when meeting a stranger). Standard deviations for items 7 and 10 were also
high in comparison to the statistics for other statements, indicating a wider spread in the
data. It appears that, typically, individuals are quite confident in most of the social
circumstances described; however, these two particular situations are perceived as
intimidating for some (Table 6.9). If so, scores for these statements may contribute
significantly to discriminating between respondents with high and low levels of self-
confidence. Ideally then, the scores for one or both items should be retained in the final
determination of items used to measure self-confidence. The next sequence of test
involved combining each of the remaining 8 items in conjunction with items 7 and 10 to
determine if a robust, interpretable and reliable outcome could be achieved. Resulting
solutions commonly showed low percentages of total variance explained (under 47%)
and correspondingly low (< 0.7) Cronbach Alpha coefficients.
134
Table 6.9 Descriptive statistics self-confidence
items (sensory)
Item Mean SD
1 7.69 1.149
2 6.56 1.174
3 6.89 1.742
4 6.97 1.927
5 6.78 1.936
6 6.41 2.158
7 4.94 2.273
8 6.10 2.017
9 6.61 2.257
10 5.77 2.231
N = 263
The purpose of determining self-confidence levels was to investigate the influence of
self-confidence on product cue usage, and this required respondents to be segmented
according to high and low self-confidence levels. Therefore, a comparison was made
between mean scores of positively coded items, negatively coded items and all items
together to investigate differences in mean scores between these groupings (Table
6.11). The average score for negatively coded items was slightly lower than for the
other item groupings; however, further testing showed the difference not to be
significant (Table 6.10). However, looking at the minimum and maximum scores,
responses to the negatively coded items clearly represented a wider range in the data.
Table 6.10 Comparison of item groupings (sensory)
Items Mean SD Min Max

Negative 6.341 1.450 2.60 9
Positive 6.602 1.141 3.40 9
All 6.472 1.047 3.30 9
N = 263
135
Table 6.11 Significance testing of average item scores
Test Statistics Self confidence items (grouped) a
N 263
Chi-Square 1.357
df 2
Asymp. Sig. .507
a. Friedman Test
Interpreting factor analysis results requires the application of commonsense and a
pragmatic approach (Malhotra et.al. 2002). The scales employed must be interpretable
and meet or exceed validity testing requirements. The solution provided by retaining all
10 items meets required thresholds for Cronbach Alpha coefficients and KMO and
Bartletts tests, and also accounted for a relatively high cumulative percentage of
variance explained. However, overall the intercorrelation coefficients between all items
were relatively low and sometimes insignificant (See Appendix 4 tables A 22, A 23).
Moreover, the 3 factors produced were not easily interpreted as separate dimensions of
the self-confidence construct. Conversely, using only the negatively coded items
produced a 1 factor solution (47.2% of variance explained) with a stronger
intercorrelation matrix where all coefficients were significant. Further, there was no
significant cross loading across items with the subjective knowledge scales,
demonstrating divergent reliability. Additionally, results of KMO and Barletts tests met
or exceeded acceptable reliability thresholds (See Appendix 4). Based on these
results, the decision was made to delete all positively coded items and to compute the
mean of the negatively coded items as the single measure of respondent self-
confidence. This allowed the retention of item 7 and represented a reasonable
compromise in terms of overall loss of data.
136
6.4.2 Objective knowledge tests
As with the pilot, respondent feedback was again positive in relation to these tests, with
no evidence to suggest that internal or external validity had been compromised. The
completion of these tests was done under supervision, with the researcher available if
help or clarification was needed. Due to the opportunity to debrief respondents at the
completion of the questionnaires, the researcher found that many participants had
become quite involved and were very interested to know how they had scored.
Consistent with the pilot study figures, levels of objective knowledge were generally
lower than anticipated.
attribute average importance for the chardonnay and brie profiles assessed. High
values for Pearsons r (and resulting r2) and Kendalls tau statistics indicate that the
data is a good fit for the application.
Table 6.12 Conjoint analysis internal validity tests (sensory)
Chardonnay Sig. Brie Sig.

Pearsons r 0.998 .0000 0.956 .0000
2
R 0.996 .0000 0.913 .0000
N = 263
137
6.5 Summary
This chapter has provided a detailed account of the methodology employed in
conducting the sensory experiments, including results of relevant validation tests
conducted on measurement and analytical instruments. Next, chapter 7 provides the
results of the analysis specific to the data collected from these experimental
procedures.
138
7 Sensory data analysis results
7.1 Introduction
The previous chapter provided a detailed description of the methodology used to
conduct the sensory experiment and results of validation tests pertinent to the
measurement instruments used. This chapter provides the results of subsequent data
analysis and a discussion of their implications.
7.2 Sample profile
A sample of 263 respondents took part in the sensory experiment, each providing a
usable questionnaire. The sample profile, as compared to demographics describing the
general population of Australia, is shown in Table 7.1. The sample is found to
correspond with the population in terms of gender balance, but to exhibit some
variations in comparison to corresponding age classifications. As with the pilot sample,
the age skew is not expected to limit results.
Table 7.1 Profile demographic profile (sensory)

Male 49.0 49.8 18 yrs to 25 yrs 13.7% 11.5
Female 51.0 50.2 26 yrs to 35 yrs 19.0% 14.2
36 yrs to 45 yrs 18.3% 15.0
46 yrs to 55 yrs 22.4% 13.5
Over 55 yrs 26.6% 21.9

Less than $25, 000 16.3% 27.7 High school certificate 27.4% 30.0
$25,000 to $45,000 16.3% 26.5 Diploma / trade 27.0% 21.8
$46,000 to $65,000 22.8% 16.8 Bachelor's degree 24.3% 9.7
Over $65,000 44.5% 29.0 Post graduate degree 21.3% 3.2
139
7.3 Conjoint analysis results (sensory)
7.3.1 Chardonnay
Table 7.2 shows the average importance of chardonnay attributes tested and the
individual utilities for each attribute level. As found in the pilot, respondents were able
to discriminate between levels of objective quality and to rank the intrinsic cue levels
appropriately. While both price and COO were found to be more influential in affecting
quality perceptions than experienced acidity, the average importance of price to
respondents renders other product cues somewhat extraneous comparatively. France
was believed to provide the highest quality chardonnay, and again, little difference was
found between opinions of the U.S. and Chile. The results relating to wine price levels
are consistent with the literature, showing that a particularly low price is likely to be
associated with correspondingly low quality, and conversely, a high price with higher
quality (Zeithaml 1988; Jover, Montes et al. 2004). The positive utility given the highest
price differs from the pilot survey analysis where the highest price was deemed
undesirable. It appears respondents in this stage of the study have assessed the
profiles according to perceived quality rather than a personal preference or desire,
validating the change of anchor points used in the questionnaire.
Table 7.2 Summary of part worths and utilities chardonnay (sensory)

-0.1383 Chile
15.08 -0.1014 U.S. COO
0.2396 France
-0.8831 $ 6.00
71.81 -0.0365 $16.00 Price
0.9177 $53.00
0.1187 Average
-0.2095 High
100% N = 263
140
Table 7.3 illustrates the minimum, maximum and average total utility values of each
profile tested, sorted from the highest total average value to the lowest. Those profiles
combining the most favored attribute levels achieve higher average utility scores. The
powerful influence of price on perceptions of taste is clear with profiles ranked first by
price, then by the combination of price and COO/taste.
Table 7.3 Average values per profile chardonnay (sensory)
Profile Attribute levels Min Max Mean SD

823 France $53.00 Average 1.22 10.89 7.09 1.98
924 Chile $53.00 Above average 0.89 10.89 6.69 1.93
279 U.S. $53.00 High 0.00 10.56 6.49 1.92
253 France $16.00 Average 0.89 10.33 6.07 1.76
950 France $16.00 Above average 0.22 10.44 5.96 1.91
595 U.S. $16.00 Average 0.89 9.67 5.79 1.77
481 Chile $16.00 High 0.33 9.11 5.43 1.71
582 U.S. $ 6.00 Average -1.00 11.00 5.02 1.97
494 Chile $ 6.00 Average 0.00 11.11 4.98 1.97
696 France $ 6.00 High -0.22 10.11 4.96 1.90
152 U.S. $ 6.00 Above average -0.11 10.78 4.92 2.08
N = 263
7.3.1.1 Chardonnay profiles and likelihood of purchase
Table 7.4 shows the average utility value for each profile segmented by those that
would consider buying the product tasted and those that would not. This analysis was
done to investigate the connection between opinions of quality and likelihood of
purchase. With the exception of the first profile, there are significant differences
between the mean scores according to purchase intention, suggesting that the
assessment of quality is positively linked to likelihood of purchase. Those subjects who
considered the samples to be of higher quality were more likely to consider buying the
product than those who did not. These finding are consistent with the literature and
expectations (Zeithaml 1988). Given that the extrinsic cues described largely
overpowered taste, this exploratory analysis has significant potential implications for the
marketing of wine products.
141
Table 7.4 Average chardonnay profile values by purchase intentions
Yes No
Profile Mann Whitney U Z Sig.
Mean SD Mean SD
253 6.08 1.17 6.03 1.11 7866.00 -0.06 .947
582 5.36 1.75 4.80 2.08 6893.00 -2.14 .032
481 6.52 1.32 4.69 1.55 3103.00 -8.89 .000
696 6.20 1.47 3.79 1.47 2116.00 -10.53 .000
595 6.88 1.19 4.74 1.55 2301.00 -10.19 .000
924 7.85 1.07 6.15 1.99 3596.00 -6.78 .000
152 6.42 1.42 3.54 1.56 1533.50 -11.48 .000
823 8.12 1.29 6.49 2.02 4139.50 -6.67 .000
494 6.16 1.63 3.81 1.54 2470.00 -9.91 .000
950 7.06 1.39 4.70 1.61 2180.00 -10.26 .000
279 7.50 1.27 6.09 1.94 4143.50 -5.60 .000
N = 263
7.3.1.2 Knowledge and self-confidence levels
Consistent with our pilot study results, the level of objective knowledge amongst this
group was generally found to be low. The data shows respondents achieving a mean
score of only 4.71 correct answers out of the 14 multiple choice questions asked in the
objective knowledge test, with 91% of respondents scoring 7 correct answers or less.
Scores for subjective knowledge imply that these respondents do not see themselves
as chardonnay experts but have still overestimated their expertise considerably (Table
7.5).
Table 7.5 Mean scores for knowledge and self-confidence (sensory)
Mean SD
Subjective knowledge chardonnay 4.55 1.72
Objective knowledge chardonnay (standardized) 2.93 1.46
Self-confidence (not product specific) 6.34 1.45
N = 263
142
7.3.1.3 Influence of knowledge and self-confidence
The respective influence of knowledge and self-confidence on cue usage was
investigated further using a staged approach. First, correlation coefficients were
calculated to quantify any significant relationships between the dependent variables of
objective knowledge, subjective knowledge and self-confidence and the independent
variables of average importance placed on COO, price and acid. Following this,
correlation coefficients were computed for the independent variables and attribute level
utility values to reveal any significant relationships specific to each. Tables 7.6 and 7.7
provide the results of the analysis using the sample as a whole. As many variables
failed tests for normality, non parametric methods were used (See Appendix 3 Table
A 2).
Table 7.6 Spearmans rho Average importance chardonnay
Objective Subjective Ave Ave

Self
knowledge knowledge Importance Importance
confidence
Chardonnay Chardonnay COO Price
Objective knowledge Chardonnay Correlation Coefficient 1.000
Sig. (2-tailed) .
Subjective knowledge Chardonnay Correlation Coefficient .516 1.000
Sig. (2-tailed) .000 .
Self confidence Correlation Coefficient .077 .138 1.000
Sig. (2-tailed) .213 .025 .
Ave Importance COO Correlation Coefficient .004 .065 -.052 1.000
Sig. (2-tailed) .948 .291 .404 .
Ave Importance Price Correlation Coefficient .071 .051 -.033 -.573 1.000
Sig. (2-tailed) .250 .413 .593 .000 .
Ave Importance Acid Correlation Coefficient -.050 -.103 .082 -.228 -.599
Sig. (2-tailed) .423 .096 .184 .000 .000
N = 263
There are no significant relationships between knowledge (either type) or self-
confidence and the average importance placed on COO, price or acid. A significant
relationship does, however, exist between objective and subjective knowledge (0.516).
This is not surprising as individuals would have some cognition of their own levels of
knowledge, even if they are not entirely accurate. Further, product experts are
individuals that possess both types of knowledge at high levels, therefore the
143
development of objective knowledge and subjective knowledge would occur
simultaneously. There is also a weak (0.138) but significant relationship between
subjective knowledge and self-confidence. This is also expected, as subjective
knowledge is self-assessed, and those individuals who are highly self-confident would
also be more likely to consider themselves knowledgeable. The relationships between
price, COO and acid are a reflection of the relative attribute and level trade-offs made
by consumers when scoring the product profiles. Hence, the effect of favoring one
aspect (e.g. price) is to diminish the importance of another.
Table 7.7 Spearman's rho - Utilities chardonnay
United Above
Chile France $6.00 $16.00 $53.00 Average High
States Average
Objective knowledge Chardonnay Correlation Coefficient -.030 .043 -.028 .032 -.069 .036 -.012 .064 -.028
Sig. (2-tailed) .630 .489 .646 .604 .261 .563 .851 .300 .652
Subjective knowledge Chardonnay Correlation Coefficient -.101 .128 -.006 .012 .077 -.028 -.032 .066 .037
Sig. (2-tailed) .102 .038 .923 .849 .211 .655 .610 .289 .551
Self confidence Correlation Coefficient .038 .091 -.120 .107 -.024 -.104 .013 .017 -.012
Sig. (2-tailed) .542 .141 .051 .084 .696 .094 .832 .781 .850
Chile Correlation Coefficient 1.000
Sig. (2-tailed) .
United States Correlation Coefficient -.519 1.000
France Correlation Coefficient -.562 -.337 1.000
Sig. (2-tailed) .000 .000 .
$6.00 Correlation Coefficient -.160 .070 .136 1.000
Sig. (2-tailed) .009 .257 .027 .
$16.00 Correlation Coefficient .116 -.112 -.030 -.383 1.000
Sig. (2-tailed) .060 .071 .633 .000 .
$53.00 Correlation Coefficient .031 .018 -.069 -.688 -.344 1.000
Sig. (2-tailed) .618 .777 .264 .000 .000 .
Average Correlation Coefficient .136 -.075 -.117 -.030 -.093 .052 1.000
Sig. (2-tailed) .028 .225 .058 .627 .132 .405 .
Above Average Correlation Coefficient -.214 .166 .110 -.014 .072 -.020 -.436 1.000
Sig. (2-tailed) .000 .007 .075 .818 .246 .748 .000 .
High Correlation Coefficient .025 -.029 .016 .043 -.008 -.010 -.523 -.458 1.000
Sig. (2-tailed) .690 .635 .793 .487 .891 .868 .000 .000 .
N = 263
Table 7.7 shows the correlation matrix for the independent variables and the utility
values for each attribute level. Only 1 significant correlation coefficient (0.128) was
revealed, between subjective knowledge and wine from the U.S. This correlation must
be considered somewhat cautiously given its isolated occurrence and its marginal
significance. The significant inter-correlations found between the various levels of each
144
attribute are a reflection of the utility scores derived for each. For example, the $6.00
wine was not considered high quality whereas the $53.00 wine was; therefore, a
relatively strong negative correlation (-0.7) existed between these two variables.
Similarly, the untreated wine was considered better quality than the high acid wine and
the moderate (-0.5) correlation coefficient is the outcome of this opinion.
Knowledge (both subjective and objective) may not exercise a strong influence on cue
usage because the levels of subjective and objective knowledge are quite low overall.
While low levels of knowledge can be expected to induce greater reliance on extrinsic
cues when products are merely described, the analysis of the test results shows that
the influence of extrinsic cues was not diminished even when low quality products were
actually experienced. The apparently neutral effect of self-confidence is also
unexpected given the healthy levels amongst the group and the greatly reduced
objective product quality experienced when tasting some samples (e.g. high acid
chardonnay). However, this may well illustrate consumer stubbornness driven by ego
as discussed in the literature, where a belief regarding an extrinsic cue (however
misleading) will be supported regardless of actual experience (Bell 1967; Owens 1993).
This stubbornness is not only confined to those with high self-confidence. Indeed, Bell
(1967) described low self-confidence individuals who relied on their beliefs regarding
extrinsic cues as a form of security, when put under pressure for a decision.
To investigate this issue further, the sample was segmented between those with high
and low levels of knowledge (objective/subjective) and self-confidence. Contrasting
levels of reliance on cue types may be more obvious amongst those with more extreme
scores specific to these characteristics. Discrimination between respondents was
calculated according to quartile range values for each variable (Table 7.8).
Respondents achieving scores in the lower 25% of the data distribution were deemed to
be low in knowledge (objective/subjective) and self-confidence while respondents
145
scoring in the highest 25% were considered to possess high levels. Those scoring in
the mid range of the data array were eliminated from further testing. High and low
segments per variable are obviously mutually exclusive; a respondent cannot possess
high and low levels of the same characteristic. However, a respondent may be high in
subjective knowledge and objective knowledge and even self-confidence, or low in all,
or possess any combination of levels possible according to the rotation of
characteristics and levels. Hence, the segmentation process did not exclude a
respondent from being represented in both segments for the purposes of correlation
testing. Preliminary cluster analysis was completed to determine if sufficiently
homogenous segments existed within the sample where groups possessing similar
levels across all 3 independent variables could be identified. However, testing with
these clusters was not useful, as the influence of the respective individual
characteristics was obscured in the results. Therefore, to gauge the degree of
commonality across consumer characteristics, correlations for all 3 independent
variables are included in each matrix (by segment) in order to quantify the degree of
variable intercorrelation without losing the opportunity to measure the specific influence
of each.
The average importance for each product attribute, and attribute level utilities, were
calculated by high and low segment to allow comparison of results (correlation
matrices for all segments can be seen in Appendix 8 Tables A 34 through A 45).
Summaries of the comparison of part worths for each segment are illustrated in this
chapter along with results of independent sample testing (comparing average utility
values by segment) to determine if significant differences existed between the groups.
146
Table 7.8 Quartile values for knowledge and self-confidence (chardonnay)
Subjective Self-
Percentiles Objective knowledge
knowledge confidence
25th 3.13 3.00 5.20
th
50 4.50 5.00 6.60
th
75 5.88 6.00 7.40
N = 263
7.3.1.4 Objective knowledge segments (chardonnay)
Table 7.9 Comparison of part worths and utilities objective knowledge
Attribute and level High objective knowledge Low objective knowledge

Ave Imp COO 9.79 8.86
Chile -0.0151 -0.0274
U.S. -0.0072 -0.0850
France 0.1123 0.1125
Ave Imp Price 75.20 76.75
$ 6.00 -0.8043 -0.8999
$16.00 -0.6162 0.0878
$53.00 0.8659 0.8121
Ave Imp Acid 15.01 14.39
Average 0.1268 0.1536
Above average 0.0797 0.0137
High -0.2065 0.1674
N = 70 N = 67
Pearsons r 0.994 sig. .000 0.997 sig. .000
Kendalls tau 0.944 sig. .000 1.000 sig. .001
Table 7.10 Utility comparison between high and low objective knowledge groups
a
Test Statistics - Respondents with High and Low Objective Knowledge Chardonnay (lab)
United Above
States Average
Mann-Whitney U 3529.500 3542.000 3677.500 3602.000 3340.000 3547.000 3681.000 3508.500 3575.500
Wilcoxon W 7807.500 6863.000 7955.500 6923.000 7618.000 6868.000 7002.000 6829.500 7853.500
Z -.599 -.561 -.148 -.378 -1.176 -.545 -.137 -.663 -.458
Asymp. Sig. (2-tailed) .549 .575 .883 .706 .240 .586 .891 .507 .647
a. Grouping Variable: Grouped Objective Knowledge Chardonnay (sensory)
147
Results of correlation testing between the average importance of COO, price and acid
and the independent variables according to high and low objective knowledge clusters
revealed no significant relationships. Correlation analysis between attribute level utility
values and knowledge and self-confidence for the low objective knowledge cluster
revealed a significant and low (0.224) relationship between objective knowledge and the
utility value for $53.00 and a low and negative (-0.228) between subjective knowledge
and the utility value for U.S. For the high objective knowledge group, only 1 significant
correlation was identified; a low and negative (-0.274) relationship between self-
confidence and France (See Appendix 8 Tables A 32 A 35). While these weak
relationships were revealed, the analysis failed to uncover a pattern of significant
correlations with knowledge or self-confidence, only a repetition of the inter-knowledge
and inter-attribute/level correlations found previously in testing the sample as a whole.
Table 7.9 illustrates a comparison of summary of part worths for each group, showing
slight variations in outcomes. However, a comparison of average utility values between
the low and high knowledge groups shown in Table 7.10 demonstrates that there are no
significant differences between values placed on the chardonnay product attributes, or
their respective levels, between these groups.
7.3.1.5 Subjective knowledge segments (chardonnay)
Correlation matrices found illustrating the resulting coefficients between product
attributes and consumer characteristics segmented according to high and low
subjective knowledge show no significant relationships for either group. Again, results
of testing with attribute level utility values found no pattern of significant relationships.
However, for the high subjective knowledge group, 2 significant correlations were found
between self-confidence and specific level utilities. The first is a moderate (0.363) and
significant relationship with the U.S. and a low (0.241) and significant coefficient for the
$6.00 utility. This indicates that higher subjective knowledge, supported by self-
148
confidence, can result in willingness to support a lower price and a CI which is less
congruent with quality wine. However, the absence of any significant relationships with
acid levels indicates that high subjective knowledge levels (alone or in conjunction with
objective knowledge and self-confidence) do not significantly contribute to consumers
assessment of objective quality. The correlation matrix for the low subjective
knowledge group also revealed 2 significant relationships. These are between
subjective knowledge and the utility values for Chile (-0.334) and France (0.302)
indicating a typical halo COO effect in favor of France (CI highly congruent with quality
wine) and against the South American country, as indeed would be expected from a
group with little category knowledge (Han 1989) (See Appendix 8 Tables A 36 A
39).
Table 7.11 Comparison of part worths and utilities subjective knowledge
Attribute and level High subjective knowledge Low subjective knowledge

Ave imp COO 11.11 17.67
Chile -0.1776 -0.0365
U.S. 0.0698 -0.2454
France 0.1079 0.2819
Ave imp price 73.52 63.17
$ 6.00 -0.9873 -0.8823
$16.00 0.0841 -0.1211
$53.00 0.9032 1.0033
Ave imp acid 15.37 19.17
Average 0.1558 0.2819
High -0.2397 -0.2902
N = 70 N = 67
Table 7.11 shows the summaries of part worths for the high and low subjective
knowledge groups. Whilst there is some variation between average utility values,
subsequent testing showed these not to be significant (Table 7.12). These variations
do, however, reveal that for those with low subjective knowledge, there is a wider range
of opinions across attribute levels. Thus, this segment has been somewhat more
149
reactive to taste differences and more influenced by differing country images than the
higher subjective knowledge group.
Table 7.12 Utility comparison between high and low subjective knowledge groups
a
Test Statistics - Respondents with High and Low Subjective Knowledge Chardonnay (lab)
United Above
States Average
Mann-Whitney U 2108.000 1786.000 2108.500 2307.000 2001.000 2265.500 2172.000 2009.000 2184.500
Wilcoxon W 4593.000 4064.000 4593.500 4792.000 4279.000 4750.500 4657.000 4287.000 4462.500
Z -1.022 -2.411 -1.020 -.164 -1.483 -.343 -.746 -1.450 -.692
Asymp. Sig. (2-tailed) .307 .016 .308 .870 .138 .732 .456 .147 .489
a. Grouping Variable: Grouped Subjective Knowledge Chardonnay (lab)
7.3.1.6 Self-confidence segments (chardonnay)
Correlation matrices illustrating the coefficients between the average importance for
each product attribute and self-confidence levels showed no significant relationships
between the dependent and independent variables for either the high or low self-
confidence groups (See Appendix 8 Tables A 40 A 43).
For the high self-confidence group, two significant correlations emerged between
subjective knowledge and the country utilities for Chile (-0.246) and the U.S. (0.371),
similar to the analysis of the subjective knowledge segments. These correlations are
consistent with the focus group results and the literature, whereby consumers are less
likely to believe products from Chile to be high quality as compared to products from the
U.S. or France. In the low self-confidence group, 2 isolated but significant correlations
emerged between self-confidence and the COO utilities for Chile (0.285) and France (-
0.354); representing an opposite opinion to those correlations discussed above. Whilst
this cluster possesses the lowest levels of self-confidence, it appears that as levels
increase amongst this group the probability of supporting Chilean (less congruent CI)
wine increased, and support for French wine diminished. This result is more in line with
expectations for the high self-confidence segment as opposed to this cluster. However,
150
it must be remembered that all the significant correlations reported here are relatively
low, isolated, and that no readily interpretable pattern is apparent. For example, while
correlation testing suggests that this group may fail to support French wine, the average
utility value for France for this segment suggests the opposite (0.3397) (See Tables A
42 A 45). Whereas the average utility value for France for the high self-confidence
segment suggests a propensity towards a negative prejudice. Table 7.13 shows the
utility comparison between segments and, indeed, the average utility for France
between groups is the only significant result. The comparison of part worths and
utilities generated results more in line with our expectations. The wide range of
opinions produces divergent results between the groups for COO and acid. Those with
high self-confidence correctly rated the high acid wine harshly, and paid the least
attention to COO. Whereas the low self-confidence group relied much more heavily on
COO, rated French wines particularly well and backed this up with strong reliance on
the highest price when forming their opinions. Further, this was the only group to
incorrectly believe the untreated wine was the poorest quality. However, irrespective of
these varying opinions regarding acid and COO, for both groups price was the major
contributing factor tquality as assment. These conjoint findingsU.S. or empered by
the results of the Mann Whitney test for comparison of averages seen in Table 7.14,
where only one utility comparison between groups was found to be significant and
samples sizes are relatively small.
151
Table 7.13 Comparison of part worths and utilities self-confidence
Attribute and level High self-confidence Low self-confidence

Ave imp COO 4.34 19.44
Chile -0.0422 -0.1984
U.S. 0.0578 -0.1413
France -0.0156 0.3397
$ 6.00 -0.8222 -0.9698
$16.00 -0.0689 -0.0984
$53.00 0.8911 1.0683
Average 0.2378 -0.2222
High -0.2556 -0.1289
N = 78 N = 58
Table 7.14 Utility comparison between high and low self-confidence groups
a
Test Statistics - Respondents with high and low self confidence (sensory)
United Above
States Average
Mann-Whitney U 2099.500 1838.500 1792.000 1849.500 2108.500 1956.500 2125.000 2224.000 2174.500
Wilcoxon W 3810.500 3549.500 4873.000 3560.500 5189.500 5037.500 3836.000 5305.000 5255.500
Z -.716 -1.867 -2.071 -1.817 -.676 -1.346 -.604 -.167 -.386
Asymp. Sig. (2-tailed) .474 .062 .038 .069 .499 .178 .546 .867 .700
a. Grouping Variable: Grouped Self Confidence
7.3.1.7 Chardonnay summary
Results from the sensory experiment demonstrate the ability of COO and price to
influence perceptions of quality, even in some instances, overwhelming sensory
experiences. Surprisingly, whilst isolated significant correlation coefficients were found
between the independent and dependent variables tested, no reliable pattern of strong
relationships was found to exist. Comparison of attribute average importance and utility
values between groups shows general differences in the range of opinions due to
segmentation criteria. However, further testing found only 1 average utility value to be
significantly different between high and low clusters. These outcomes are particularly
152
unexpected given the controlled nature of the experiment, where the usual distractions
accompanying consumption were absent, minimizing respondents distraction from the
sensory experience.
7.3.2 Brie
As with wine, price was clearly considered the most important attribute contributing to
perception of brie quality; however, these respondents acknowledged fat as more
important than COO in forming their opinions (Table 7.15). Overall for the group the
lowest quality brie (full cream) was found better than the triple cream brie; a result
inconsistent with the literature or expectations (Hamilton, Knox et al. 2000). Where the
wine sampled differed in taste only, the cheese samples also varied in texture due to
differing fat levels, and consequently, afforded a more complex and powerful sensory
experience. Feedback from participants during debriefing indicated the triple cream brie
was found by some to be extremely rich and somewhat overpowering. For them, the
sensory experience would be negative; hence, the triple cream brie samples would be
considered lower quality and rated accordingly. The double cream brie achieved a high
and positive utility, indicating that this product seems to strike the right balance in terms
of taste and texture for many people. As with wine, regardless of sensory perceptions
price was found to be the most influential product cue when determining quality. The
increased influence of price is attributable in some part to the higher prices used and
the wider variance in price levels (Rao and Hauser 2004). Nonetheless, considering the
complexity of the sensory experience, once again, the overwhelming influence of price
is surprising.
153
Table 7.15 Summary of part worths and utilities brie (sensory)

-0.1841 Argentina
25.07 0.0208 Canada COO
0.1633 France
-0.3405 $28.95
43.99 0.2689 $49.00 Price
0.0717 $69.95
-0.0148 Full Cream
30.95 0.2218 Double Cream Fat
-0.2070 Triple Cream
100% N = 263
Table 7.16 Average values per profile brie (sensory)
Profile Attribute levels Min Max Mean SD

810 France $69.95 Double cream 0.78 10.89 7.29 1.73
709 France $49.95 Full cream 1.11 11.22 7.24 1.88
507 Argentina $49.95 Double cream 1.11 11.67 7.13 1.87
367 Canada $69.95 Full cream 1.11 11.11 6.91 1.91
178 Canada $49.95 Triple cream 0.44 11.22 6.90 2.33
266 Canada $28.95 Double cream 1.44 11.22 6.73 1.84
810 France $28.95 Full cream 0.11 11.67 6.64 1.97
380 Argentina $69.95 Triple cream 0.33 10.22 6.51 2.36
735 France $28.95 Triple cream 0.56 11.44 6.45 2.23
393 Argentina $28.95 Full cream 1.11 10.33 6.29 1.85
139 Argentina $28.95 Triple cream 0.11 10.78 6.10 2.12
N = 263
Table 7.16 Illustrates the average value attributed to each profile as derived from the
specific combination of utilities involved. The comparison for brie demonstrates the
strong conviction held by respondents that a lower price is linked to lower quality and
even negative perceptions of taste can be overcome by a combination of higher price
supported by congruent country of origin.
154
7.3.2.1 Brie profiles and the likelihood of purchase
Table 7.17 Average brie profile values by purchase intentions
Yes No
Mean SD Mean SD
810 6.87 1.90 5.84 1.99 3898.00 -3.05 .002
139 6.72 1.93 5.34 2.10 5294.50 -5.17 .000
367 7.73 1.56 6.24 1.89 4586.00 -6.40 .000
266 7.44 1.34 4.65 1.41 2867.50 -10.13 .000
709 7.76 1.31 5.99 1.98 3307.50 -7.47 .000
380 8.23 1.10 5.47 2.28 2461.00 -9.49 .000
735 7.83 1.35 4.47 1.63 1000.50 -12.00 .000
393 7.13 1.45 4.79 1.42 1937.50 -9.89 .000
621 7.90 1.30 6.72 1.81 5171.50 -5.46 .000
178 8.26 1.39 5.42 2.20 2322.50 -10.04 .000
507 7.86 1.40 6.05 1.82 3497.00 -7.48 .000
N = 263
Table 7.17 shows the average total utility value for each brie sample grouped by
consideration of purchase. As with chardonnay, the mean values are significantly
higher for those that would consider purchasing as compared to those that would not,
linking the perception of quality with purchase intention.
7.3.2.2 Knowledge levels brie
In relation to cheese, the general level of objective knowledge amongst this group was
found to be quite low. Results from the objective knowledge test found that
respondents achieved an average of 4.59 correct answers and that 90% of them scored
7 correct answers or less. Scores for subjective knowledge imply that these
respondents did not see themselves as product experts for cheese either but as with
wine, have overestimated their expertise (Table 7.18). As self-confidence is not product
specific, the score determined previously will be used for analysis respective of brie
also.
155
Table 7.18 Mean scores for knowledge (brie)
Mean SD
Subjective knowledge brie 4.59 0.80
Objective knowledge brie (standardized) 2.95 0.51
Self-confidence (not product specific) 6.34 1.45
N = 263
7.3.2.3 Influence of knowledge and self-confidence (brie)
Tables 7.19 and 7.20 illustrate the correlation coefficients calculated to investigate
significant relationships between knowledge and self-confidence and product cues
contributing to perceptions of quality specific to brie for the sample as a whole. As
expected, there was significant inter-correlation (0.443) between subjective and
objective knowledge. As previously discussed, it is not possible to develop objective
knowledge and be unaware of it and expertise is also strongly linked to category
familiarity (Alba and Hutchinson 1987; Park, Mothersbaugh et al. 1994; Alba 2000).
Also, consistent with the outcomes determined for chardonnay, significant correlations
were found between levels within an attribute as well as between the attributes
themselves. These represent the degree of relative trade off occurring between the
attributes/levels as part of the conjoint analysis utility calculations.
156
Table 7.19 Spearmans rho Average importance brie
Objective Subjective Average Average

Self
Knowledge Knowledge Importance Importance
confidence
Brie Brie COO Price
Objective Knowledge Brie Correlation Coefficient 1.000
Sig. (2-tailed) .
Subjective Knowledge Brie Correlation Coefficient .443 1.000
Sig. (2-tailed) .614 .514 .
Average Importance COO Correlation Coefficient .080 .053 .039 1.000
Sig. (2-tailed) .198 .388 .530 .
Average Importance Price Correlation Coefficient -.023 -.087 -.068 -.058 1.000
Sig. (2-tailed) .709 .160 .275 .351 .
Average Importance Fat Correlation Coefficient -.027 .042 .006 -.588 -.722
Sig. (2-tailed) .664 .502 .919 .000 .000
N = 263
Table 7.20 Spearmans rho Utilities brie

$28.95 $49.00 $69.95 Full Double Triple
Argentina Canada France
per kilo per kilo per kilo Cream Cream Cream
Objective Knowledge Brie Correlation Coefficient .038 -.040 .036 .054 -.056 -.019 -.143 -.091 .173
Sig. (2-tailed) .539 .515 .560 .384 .364 .761 .020 .142 .005
Subjective Knowledge Brie Correlation Coefficient -.022 -.038 .095 .050 -.018 -.041 -.220 -.028 .188
Sig. (2-tailed) .720 .537 .123 .416 .774 .505 .000 .654 .002
Self confidence Correlation Coefficient -.009 -.034 .070 .179 -.108 -.042 .006 -.046 .016
Sig. (2-tailed) .880 .580 .258 .004 .081 .501 .921 .460 .790
Argentina Correlation Coefficient 1.000
Sig. (2-tailed) .
Canada Correlation Coefficient -.485 1.000
Sig. (2-tailed) .000 .000 .
$28.95 per kilo Correlation Coefficient -.206 -.070 .272 1.000
Sig. (2-tailed) .001 .256 .000 .
$49.00 per kilo Correlation Coefficient -.028 .076 .004 -.491 1.000
Sig. (2-tailed) .656 .221 .946 .000 .
$69.95 per kilo Correlation Coefficient .225 .015 -.280 -.493 -.445 1.000
Sig. (2-tailed) .000 .806 .000 .000 .000 .
Full Cream Correlation Coefficient .017 -.090 .038 .054 -.021 .007 1.000
Sig. (2-tailed) .780 .145 .539 .386 .731 .912 .
Double Cream Correlation Coefficient .027 -.016 .025 .011 -.041 -.013 .092 1.000
Sig. (2-tailed) .664 .802 .688 .856 .503 .837 .136 .
Triple Cream Correlation Coefficient -.020 .073 -.043 -.036 .028 .002 -.759 -.666 1.000
Sig. (2-tailed) .746 .237 .484 .561 .653 .975 .000 .000 .
N = 263
The correlation matrix for utilities shows 2 weak but significant relationships between
objective knowledge and fat levels. A negative relationship (-0.142) for full cream and a
positive coefficient (0.173) for triple cream indicated that those that know more about
cheese favored the higher fat content brie, an outcome consistent with the literature and
157
expectations (Hamilton, Knox et al. 2000). However, the relationships were weak, and
represented little overall influence in the quality assessments made. A similarly weak
(0.179) but significant relationship existed between self-confidence and the $28.95
utility, suggesting that higher confidence levels allow for the lowest priced product to be
seen more favorably. Once again, however, the correlation coefficients were very low
and, therefore, represented weak associations.
The next step was to segment the sample and to cluster those individuals possessing
the most extreme levels of knowledge and self-confidence in order to see if more robust
relationships between the dependent and independent variables could be found.
Identical methodology to that used in the analysis of chardonnay data was used to
cluster respondents with high and low levels of knowledge and self-confidence specific
to brie. Again, those respondents with scores in the lowest 25% were considered to be
low in objective/subjective knowledge and those in the highest 25% were deemed to
possess high levels, with all other respondents ignored. Table 7.21 illustrates the
quartile values used to segment the sample and, as self-confidence is not product
specific, the existing segments were used. (See Appendix 9 Tables A 56 to A 67 for
correlation matrices)
Table 7.21 Quartile values for knowledge and self-confidence (brie)
Percentiles Subjective knowledge Objective knowledge

th
25 3.13 3.00
th
50 4.38 4.00
th
75 5.63 6.00
N = 263
158
7.3.2.4 Objective knowledge segments (brie)
There were no significant relationships between levels of objective knowledge and the
average importance of the attributes tested. For those in the high objective knowledge
segment, there is a low (-0.284) and significant relationship between subjective
knowledge and price, suggesting that those more modest in their self-assessment of
knowledge considered price more indicative of quality than others (Lee and Lou 1996;
Acebron and Dopico 2000; Teas and Agarwal 2000). A similar association (-0.266)
existed for the low objective knowledge group between self-confidence and price. This
is consistent with the literature, whereby those who do not possess high levels of
objective knowledge, and also lack self-confidence, use extrinsic cues as welcome
cognitive shortcuts (Bell 1967). Once again, however, the coefficients are low indicating
that reliable predictions regarding the degree of influence on quality perceptions exerted
by the dependent variables are not possible (See Appendix 9 Tables A 56 A 59).
The correlation matrix specific to utility values and the low objective knowledge cluster
revealed no significant correlations. For the high objective knowledge segment, 2
significant associations were found. The first (-0.370) is between subjective knowledge
and the full cream level, consistent with that determined for the sample as a whole. As
levels of subjective knowledge increased, utility value for the lowest quality product
diminished (-0.370) and value for the triple cream product increased (0.302). Since
higher levels of subjective knowledge, in conjunction with sound objective knowledge,
contributed to true expertise, this outcome is consistent with the literature and
expectations (Park, Mothersbaugh et al. 1994). The other (0.274) is between self-
confidence and the utility value for $28.95 reflecting the willingness of higher self-
confidence individuals to support a lower price.
159
Table 7.22 provides a comparison of part worths and utility values between the high and
low objective knowledge groups. The average importance of price is comparable, but
opinions regarding the average importance of COO and fat varied greatly for each
segment. Analysis of data from the high objective knowledge group revealed that these
respondents reported a wide range of variance in opinions across different country
levels, whereas the low objective knowledge group perceived all countries as similar.
This provides an average importance of COO for one group (high objective knowledge)
of 35.86% but only 9.69% for the other (low objective knowledge). In the case of fat
levels, both groups perceived the specific levels to be different; however, in comparison,
the range of opinions was extreme for the low objective knowledge group. With an
average importance of 53.27% for fat, the low objective knowledge segment strongly
favored the double fat cheese, returning a positive utility for the full fat brie and
punishing the triple fat with a strongly negative utility. The high group also agreed that
the double fat was the best level, but it returned a positive utility for the triple cream and
disfavored the full fat level. The differences in opinion regarding the full fat and triple
fat brie are significant, as shown in Table 7.23. This outcome suggests that those with
high levels of objective knowledge may possess a more sophisticated palate and
appreciate the creamier texture and richer flavor of the triple cream cheese while those
with less experience find it somewhat unpleasant.
The credence given to the COO cues by the high knowledge group is likely to be
another indication of their knowledge and experience in relation to the product category.
Experts are known to make use of extrinsic cues when these are legitimate indications
of likely quality; hence the high utility value for France and the respectively low utility
value for Argentina (Alba and Hutchinson 1987; Alba 2000). In this case, the group with
the lowest level of knowledge applied higher credence to the intrinsic cues than those
with better product understanding, suggesting that knowledge may blind some
consumers to objective product quality because they believe in what should be the case
160
rather than assess the product more objectively. Conversely, the low knowledge group
had no expectations in relation to products or cheese attributes and hence relied more
heavily on what their own senses were conveying to them. While using price in their
assessments, taste for them, was more important and COO relatively meaningless.
Table 7.22 Comparison of part worths and utilities brie objective knowledge
Attribute and level High objective knowledge Low objective knowledge

Ave imp COO 35.63 9.69
Argentina -0.2495 -0.1250
Canada -0.0448 0.0119
France 0.2943 0.1131
$28.95 -0.3372 -0.5179
$49.00 0.2651 0.3929
$69.95 0.0721 0.1250
Ave imp fat 24.90 53.27
Full cream -0.2144 0.3452
Double cream 0.1657 0.4821
Triple cream 0.0487 -0.8274
N = 57 N = 56
a
Test Statistics - Respondents with High and Low Objective Knowledge Brie (sensory)
Mann-Whitney U 1580.500 1490.000 1373.000 1427.000 1410.500 1549.000 1087.00 1328.500 1042.500
Wilcoxon W 3176.500 3143.000 2969.000 3023.000 3063.500 3202.000 2740.00 2981.500 2638.500
Z -.089 -.610 -1.283 -.972 -1.067 -.270 -2.926 -1.538 -3.180
Asymp. Sig. (2-tailed) .929 .542 .200 .331 .286 .787 .003 .124 .001
a. Grouping Variable: Grouped Objective Knowledge Brie
161
7.3.2.5 Brie subjective knowledge segments
There are no significant correlations for any of the independent variables and attribute
average importance for the high and low subjective knowledge segments. There was 1
significant correlation (0.248) found for the high subjective knowledge group between
self-confidence and the utility value for $28.95, reflecting previous findings. For the low
subjective knowledge group, 2 significant relationships emerged for subjective
knowledge and the COO utilities for Canada (0.260) and France (-0.246). These
correlations are likely to be a reflection of their (self confessed) lack of category
knowledge where the congruent CI such as that provided by France over a country
such as Canada is ignored. Also, this group believed COO to be comparatively
unimportant to their quality assessments and may be reflecting opinions of CI alone
(favoring Canada over France) rather than making a country and product association.
In addition, 3 significant relationships were found between objective knowledge and
$28.95 (0.310), double cream (-0.295) and triple cream (0.262) (See Appendix 9
Tables A60 A 63). The influence of objective knowledge is a reflection of results
found earlier. Also, whilst these relationships are significant, they are weak, for
example a correlation of 0.262 represents an influence of only 7% on the dependent
variable. This is borne out when considering comparison of the conjoint analysis
results.
Table 7.24 illustrates the comparison of part worths and utility values for the high and
low subjective knowledge clusters. While the average importance of fat was similar for
both groups, 2 significant and opposite opinions were observed. Those with low
subjective knowledge supported the full cream brie and rejected the triple cream,
whereas the reverse was true for the other group. As with the objective knowledge
cluster, the appreciation of a cheese with a higher fat content is likely to indicate a more
experienced palate. Also, since the objective and subjective segments were not
mutually exclusive and there is a moderate correlation between these 2 groups, some
162
respondents were represented in both the high objective and high subjective knowledge
segments. The higher range of values for price for the low subjective knowledge group
indicates that they used price as a surrogate indicator of quality as they believed they
knew little about the product or category (Monroe 1976; Liefeld, Heslop et al. 1996;
Kardes, Cronley et al. 2004).
Table 7.24 Comparison of part worths and utilities brie subjective knowledge
Attribute and level High subjective knowledge Low subjective knowledge

Ave imp COO 28.37 11.38
Argentina -0.1732 -0.0415
Canada -0.0458 -0.0614
France 0.2190 0.1028
$28.95 -0.2222 -0.6847
$49.00 0.1846 0.2570
$69.95 0.0376 0.1277
Ave imp fat 42.20 44.14
Full cream -0.3350 0.2322
Double cream 0.0866 0.1725
Triple cream 0.2484 -0.4046
N = 68 N = 67
a
Test Statistics - Respondents with High and Low Subjective Knowledge Brie (sensory)
Mann-Whitney U 2187.500 2231.500 1980.000 2176.000 2270.500 2192.500 1588.000 2192.500 1664.500
Wilcoxon W 4533.500 4577.500 4258.000 4454.000 4616.500 4538.500 3934.000 4538.500 3942.500
Z -.399 -.205 -1.314 -.449 -.033 -.377 -3.039 -.377 -2.701
Asymp. Sig. (2-tailed) .690 .838 .189 .653 .974 .706 .002 .707 .007
a. Grouping Variable: Grouped Subjective Knowledge Brie
163
7.3.2.6 Brie self-confidence segments (sensory)
The correlations between knowledge and self-confidence and average attribute
importance for the high and low self-confidence groups confirm that there are no
significant relationships between these variables regardless of self-confidence levels.
Further, the correlations between the independent variables and attribute utility values
presented similarities with those previously found to exist between objective/subjective
knowledge and fat levels (high self-confidence). One correlation was found in analysis
between $69.95 (0.265) and objective knowledge for the low self-confidence group and
another between objective knowledge and double cream (-0.423) for the high self-
confidence cluster (See Appendix 9 Tables A64 A 67). Analysis reveals no
significant correlations between self-confidence and utility values for either group,
confirming that self-confidence has no direct bearing on respondents opinions of brie
quality.
Comparison of average importance and average utility values between the 2 groups
showed 1 significant difference in opinions regarding attribute levels; this is specific to
views of brie at $28.95 per kilo (Table 7.27). Those with low self-confidence judged this
pricing point unfavorable and took it into account in their product ratings. Heavy
reliance on price to indicate quality (43.33%) was expected from this group. More
surprisingly, however, they showed wider ranges of opinions in relation to actual
sensory feedback than the high self-confidence group (Table 7.26). While both groups
reported a negative utility value for the triple cream brie, the low confidence group
reported a wider range of values, although the difference in average utilities was not
significant. The high self-confidence group, on the other hand, considered COO almost
comparable to taste in influence.
164
Table 7.26 Comparison of part worths and utilities brie - self-confidence
Attribute and level High self-confidence Low self-confidence

Ave imp COO 34.21 15.28
Argentina -0.2151 -0.1871
Canada -0.0142 0.1345
France 0.2293 0.0526
$28.95 -0.1980 -0.5673
$49.00 0.1610 0.3450
$69.95 0.3700 0.2222
Ave imp fat 38.16 41.39
Full Cream 0.0883 0.1345
Double Cream 0.2037 0.3684
Triple Cream -0.2920 -0.5029
N = 78 N = 58
a
Test Statistics - Respondents with High and Low Self Confidence Brie (sensory)
Mann-Whitney U 2241.500 1969.000 1925.500 1731.500 2044.500 2036.000 2238.500 2043.000 2094.000
Wilcoxon W 3952.500 5050.000 3636.500 3442.500 5125.500 5117.000 5319.500 5124.000 3805.000
Z -.090 -1.291 -1.483 -2.337 -.958 -.996 -.103 -.965 -.740
Asymp. Sig. (2-tailed) .928 .197 .138 .019 .338 .319 .918 .335 .460
a. Grouping Variable: Grouped Self Confidence
7.3.2.7 Summary of test results Brie
As with chardonnay, the results of the sensory experiment specific to brie also
demonstrated the ability of COO and price to influence perceptions of quality over taste.
In the analysis, instances of significant correlation coefficients between the independent
and dependent variables tested were found, but no strong and reliable pattern of
relationships were revealed. Comparison of attribute average importance and utility
values according to high and low segments showed general variances in the range of
opinions between segments but only 4 significant differences were found between
165
average utility values. Given the rich sensory experience afforded by the brie products
tasted and the controlled nature of the experiment, these results are inconsistent with
the general body of literature, suggesting that the power of extrinsic cues to influence
product quality may be even stronger than previously believed.
7.4 Sensory analysis summary
Data analysis for both products yields similar results. While isolated instances of
significant relationships were found to exist between knowledge (objective and
subjective) and self-confidence and the product attributes tested, there does not appear
to be any evidence that these consumer characteristics exert a strong, clear and
significant influence on the determination of objective product quality. Generally,
objective knowledge levels were found to be much lower than expected, and if
knowledge is not there, it cannot exert any influence. Moreover, testing between those
who do possess sound levels of knowledge and those who do not, revealed few
significant differences between the 2 groups in their ability to accurately assess quality.
Indeed, often the extrinsic cues of price and COO were more influential amongst the
higher level groups than those possessing prospectively lower levels of knowledge and
self-confidence. The segment possessing high subjective knowledge (for brie) reported
the highest utility value to the triple cream brie and the lowest to the full cream brie, with
an overall average importance for fat of 42.20%, 29.30% for price and 28.37% for COO.
This group, with high self-assessed product category familiarity, was the most accurate
in their product assessments. Yet, the relationships found between their personal
consumer characteristics and respective product utility values were not significant.
Therefore, an empirically proven link has not been established. Investigation of the
influence of self-confidence found little to suggest that high levels of this characteristic
provides an effective shield against the influence of potentially misleading and irrelevant
extrinsic cues. In fact, a strong belief held in relation to price and COO was upheld
166
even when product quality diminished. The analysis overwhelmingly validated price as
the most consistently influential product cue amongst those tested in these
experiments, with COO also found to be extremely influential with many respondents.
167
8 Conjoint analysis survey and results
for Chardonnay
8.1 Introduction
Chapter 7 provided results of conjoint analysis using data gathered from participants in
the sensory experiment and quantified the influence of knowledge (objective and
subjective) and self-confidence on respondents use of intrinsic and extrinsic cues. An
overview of the methodology used to conduct the conjoint analysis by survey and
subsequent data analysis results specific to chardonnay are provided in this chapter. In
addition, a comparison of the survey data analysis with the sensory data analysis for
chardonnay is provided. The chapter concludes with a presentation of hypotheses
testing outcomes. The respective results for brie follow in chapter 9.
8.2 Methodology
8.2.1 Sample (conjoint survey)
A convenience sample of 274 students enrolled in evening classes at TAFESA City
Campus was recruited to participate in the final stage of research. Only students
recently enrolled were considered for inclusion to ensure that respondents had not been
sensitized by previous participation in other aspects of this study. Classes of new
students in marketing, finance, small business and justice studies were selected for
sampling. Experience gained from the completion of previous stages of the research
indicated that participants would find the questionnaire time consuming, complex and
challenging, and that it required at least 15 minutes to complete accurately. Hence, it
was not feasible to expect lecturers or students to sacrifice class time in order to take
part. Therefore, to encourage participation and minimize respondent fatigue and error it
was considered necessary to offer an attractive incentive and allow participants to take
168
the questionnaire home to complete (See Appendix 10 Invitations to participate in the
survey).
A 2-step strategy for questionnaire distribution and return was implemented. First,
booklets were distributed in October and November 2006 by a number of selected
lecturers to their students during their evening classes. Students were advised that
they would receive a free bottle of Scarpantoni School Block red wine for each
completed questionnaire returned. As this was a general population sample (over the
age of 18 years) those volunteering to take a questionnaire were also permitted to take
booklets for partners, friends, work associates et cetera, but each person was clearly
instructed to complete their questionnaires in isolation and not to discuss or share
answers with others. This directive was reinforced in the written instructions provided in
the booklet. A wine pick-up and questionnaire return point was established and
attended over four evenings during the first and third weeks of November 2006 at the
TAFESA City Campus, allowing respondents to drop off completed booklets and collect
their wine.
As experienced in the original pilot conjoint study, respondent fatigue and lack of
attention to instructions proved to be a limitation here also. This is in spite of the
attractive incentive offered and the unlimited time provided to complete the task. From
over 375 survey booklets distributed to apparently willing respondents, only 274 usable
questionnaires were returned. Of those returned, in many cases entire sections of the
booklet were not completed and, again, the suggestions of an ethnocentric reaction to
the hypothetic products were apparent. Examples include where respondents rated
each wine profile as 1 or 5, with no consistently answered in response to the
question regarding consideration of purchase. In order to directly compare results from
the sensory experiment with this survey, it was necessary to replicate the data
collection instrument. However, this meant that the briefing in relation to Australian
169
wines was not possible and this affected the result of the paper-based survey. Without
the benefit of personal briefings, supervision and support, respondent error and fatigue
appear to have impacted on participation rates and data accuracy.
8.2.2 Data collection instrument (conjoint survey)
As there was no opportunity to brief participants personally, a written summary of the
information presented to groups involved in the sensory experiment was incorporated
into the beginning of the questionnaire booklet. With this exception, the layout, format
and content of this questionnaire were identical to that used in the sensory experiment
(See Appendix 11 Conjoint survey questionnaire). As with the pilot study
questionnaire, the intrinsic cues of acid and fat were described in order to communicate
changing levels; figures 8.1 and 8.2 illustrate examples of chardonnay and brie profiles.
Figure 8.1 Example of wine product profile (survey)
Chardonnay 253
Produced in France
Retail Price $16.00
Acid Average
1 2 3 4 5 6 7 8 9 10
170
Figure 8.2 Example of cheese product profile (survey)
Brie 810
Produced in France
Retail Price $28.95 per kilo
Fat Full cream
1 2 3 4 5 6 7 8 9 10
8.3 Validation of research instruments (survey)
As the decision had been taken previously to only use the negatively phrased items in
the self-confidence scale, the same methodology was repeated in this phase of the
study, prior to the validation of the self-confidence scale and computation of a measure
of self-confidence. Therefore, reported statistics for self-confidence in this chapter
reflect the use of the previously retained scale items. For measures of subjective
knowledge (chardonnay and brie) all 8 items were retained as per the analysis of
sensory data. All scales were tested for internal reliability through determination of the
Cronbach Alpha coefficient for each measure; these are illustrated in Table 8.1 with
coefficients ranging from 0.753 to 0.903, indicating sound reliability.
Table 8.1 Reliability coefficients of scales (survey)
Subjective knowledge brie N of items = 8 Alpha = 0.932
N of Cases = 274
171
Table 8.2 Construct reliability for scales (survey)
Cum % No of factors
Scale of variance with
Subjective knowledge brie 68.6 1
N = 274
The respective percentages of cumulative variances explained (factors with eigenvalues
> 1) are 51.4% for self-confidence, 59.2% for subjective knowledge chardonnay and
68.6% for subjective knowledge brie (Table 8.2). For subjective knowledge (both
products) and self-confidence, a 1-factor solution was achieved with a relatively high
percentage of total variance explained thus confirming construct reliability (Flynn and
Goldsmith 1999; Malhotra, Hall et al. 2002). Results of convergent and divergent
reliability testing can be seen in Appendix 4.
8.3.2 Objective knowledge tests
As found with the previous stages of the study, respondent feedback was positive in
relation to these measures with no evidence to suggest that internal or external validity
had been compromised. Respondents continued to show interest in the tests with
some querying answers when returning their booklets. Review of usable questionnaires
prior to data entry did not reveal instances of missing data or evidence of respondent
fatigue, intentional error or omission for this aspect of the questionnaire booklet.
Scoring showed results consistent with the pilot study and data from the sensory
experiment: levels of objective knowledge were generally found to be low.
172
the average importance of attributes for the chardonnay and brie profiles assessed.
The high values for Pearsons r (and subsequent r2) and Kendalls tau statistics indicate
sound validity.
Table 8.3 Conjoint analysis internal validity tests (survey)
Chardonnay Sig. Brie Sig.

Pearsons r 0.998 .0000 0.999 .0000
2
R 0.996 .0000 0.998 .0000
N = 274
8.4 Conjoint survey data analysis results
8.4.1 Sample profile
A sample of 274 usable questionnaires was used for analysis. The sample profile, as
compared to the general population of Australia is shown in Table 8.4.
Table 8.4 Profile demographic profile (survey)

Male 51.1 49.8 18 yrs to 25 yrs 19.0 11.5
Female 48.9 50.2 26 yrs to 35 yrs 20.1 14.2
36 yrs to 45 yrs 21.9 15.0
46 yrs to 55 yrs 20.1 13.5
Over 55 yrs 18.6 21.9

Less than $25, 000 11.9 27.7 High school certificate 21.2 30.0
$25,000 to $45,000 22.7 26.5 Diploma / trade 36.1 21.8
$46,000 to $65,000 16.4 16.8 Bachelor's degree 20.8 9.7
Over $65,000 48.9 29.0 Post graduate degree 21.9 3.2
173
As with the sample used for the sensory experiment, gender balance essentially
matches the Australian general population but there is variance in terms of age
classifications. However, the age skew is not expected to critically limit results.
8.4.2 Chardonnay conjoint analysis results
Table 8.5 shows the average importance of chardonnay attributes tested and the
individual utilities for each attribute level. For comparison purposes, the table also
shows the corresponding results from the sensory experiment data. The utility values
specific to the levels of acid described confirm that respondents were once again able
to differentiate between the levels and to rank them correctly in terms of better versus
diminishing objective quality. However, the acid levels are overcome by price and
virtually matched by the power of CI in terms of influencing expectations of quality.
France was believed to provide the highest quality chardonnay and, surprisingly, Chile
was deemed more attractive than the U.S. Although for this group it is a matter of one
country being perceived to be somewhat worse than the other, as neither were
favored. The results relating to wine price levels are, again, consistent with the
literature and show that a particularly low price is likely to be associated with
correspondingly low quality, and conversely, a high price with higher quality (Zeithaml
1988; Jover, Montes et al. 2004).
As previously explained, average importance is calculated using extreme utility range
values. Whilst the order of preference for attribute levels is found to be the same for
both the sensory data and the survey data, the quality evaluations based on
expectations (survey) show wider ranges amongst attribute levels for price and acid,
and hence, some variance in relation to the determined average importance of these
attributes. Surprisingly, the influence of acid was more profound in the survey data
174
results (reflecting expectations) than in the sensory evaluation data (reflecting actual
experience).
Table 8.5 Summary of part worths and utilities chardonnay
Attribute Research stage

and level Sensory Survey
Ave imp COO 15.08 12.27
Chile -0.1383 -0.1604
U.S. -0.1014 -0.2019
France 0.2396 0.3622
$ 6.00 -0.8831 -0.7379
$16.00 -0.0365 0.0887
$53.00 0.9177 1.6492
Average 0.1187 0.3061
High -0.2095 -0.3398
N = 263 N = 274
Table 8.6 illustrates the minimum, maximum and average total utility value for each
profile tested ranked from the highest total average value to the lowest. Those profiles
combining the most favored attribute levels achieve higher average utility scores.
Profile scores from the sensory experiment are included for comparative purposes.
The dominant influence of price on quality expectations appears clearly, with the 3 wine
profiles priced at $53.00 achieving the highest utility values in spite of differences in
acid levels. The prevalence of price over acid is demonstrated by the ranking of
remaining wine profiles, since other wines with average acid levels were downgraded
against lower quality products when associated with lower prices. The parity found
between the sensory and survey results demonstrate the consistent ability, under vastly
different circumstances, of the survey method to predict that extrinsic cues influence
quality evaluations, in support of Hypothesis 7. Indeed, the rankings for all 11 profiles
175
are identical with the exception of a reversal of rankings for wine profiles 494 and 582
due to the difference in opinion regarding the U.S. and Chile between the samples.
Table 8.6 Comparative average values per profile chardonnay
Min Max Mean SD

Profile Attribute levels
sens sur. sens sur. sens sur. sens sur.
823 France $53.00 Average 1.22 2.78 10.89 11.33 7.09 7.76 1.98 1.51
924 Chile $53.00 Above average 0.89 -0.89 10.89 11.11 6.68 6.97 1.93 2.02
279 U.S. $53.00 High 0.00 1.44 10.56 10.78 6.49 6.55 1.92 1.96
253 France $16.00 Average 0.89 1.67 10.33 10.22 6.07 6.20 1.76 1.50
950 France $16.00 Above average 0.22 1.33 10.44 10.00 5.96 5.93 1.91 1.56
595 U.S. $16.00 Average 0.89 1.11 0.67 10.00 5.79 5.64 1.77 1.49
481 Chile $16.00 High 0.33 0.22 9.11 11.44 5.43 5.03 1.71 1.96
494 Chile $6.00 Average 0.00 -1.56 11.11 8.11 4.98 3.85 1.97 1.67
582 U.S. $6.00 Average -1.00 -1.22 11.00 10.00 5.02 3.81 1.97 1.76
696 France $6.00 High -0.22 -0.44 10.11 10.00 4.96 3.73 1.90 1.86
152 U.S. $6.00 Average -0.11 0.00 10.78 10.00 4.91 3.54 2.08 1.80
N = 263 (sens.) N = 274 (sur.)
8.4.3 Chardonnay profiles and likelihood of purchase
Table 8.7 shows the average utility value for each profile segmented by those that
would consider buying the product described, against those that would not. Comparing
these results with those from the sensory experiment illustrates consistency in
response; those profiles rating more highly were also more likely to be considered for
purchase.
Table 8.7 illustrates that only one profile (823) was not found to be significantly different
between these 2 groups. This wine was believed to be the best quality (on average) by
both groups and was considered to be comprised of the most attractive attribute levels.
However, it still may not be considered for purchase in spite of quality expectations due
to its high price and foreign source of production. For the remaining profiles, there are
significant differences between the mean scores according to purchase intention,
suggesting that the assessment of quality is positively linked to likelihood of purchase.
176
Consistent with results from the sensory experiment, those believing the product to be
of higher quality are generally more likely to consider buying it. Price has now been
shown to influence likelihood of purchase both for expectations of quality and
perception of quality, suggesting that for wine at least, it would be very difficult to
convince consumers that a good quality wine is available at a low price. Conversely, it
appears possible to position a wine significantly above its intrinsic quality by simply
pricing it more expensively and leveraging an attractive CI or other attractive extrinsic
cues.
Table 8.7 Average chardonnay profile values by purchase intentions
Yes No
Mean SD Mean SD
253 6.54 1.50 5.73 1.40 6442.50 -0.08 .000
582 4.85 1.92 3.61 1.66 3198.00 -3.99 .000
481 6.52 1.64 4.30 1.66 2924.00 -8.71 .000
696 5.34 1.93 3.19 1.48 2565.50 -7.89 .000
595 6.41 1.29 5.04 1.34 4279.50 -7.55 .000
924 7.63 1.38 6.84 2.10 4155.00 -2.02 .044
152 4.81 1.89 3.30 1.70 2668.50 -4.79 .000
823 7.94 1.38 7.68 1.56 7671.00 -0.611 .541
494 4.91 1.50 3.39 1.53 3709.50 -6.96 .000
950 6.50 1.44 5.26 1.43 4967.00 -6.58 .000
279 7.80 1.47 6.35 1.95 2663.50 -3.99 .000
N = 274
8.4.4 Knowledge and self-confidence levels
Consistent with results from the pilot and sensory experiment, the level of objective
knowledge amongst this group was also low. Data shows respondents achieving a
mean score of only 4.07 correct answers out of the 14 multiple choice questions asked
in the objective knowledge test, with 89% of respondents scoring 7 correct answers or
less. Again, respondents did not believe themselves to be wine experts but still
believe their knowledge to be significantly better than suggested by their test scores
(Table 8.8). On average, these respondents also exhibited reasonably robust levels of
177
self-confidence. Comparison of means tests showed the difference in average
subjective knowledge to be small but significant; the variance in standard deviations
highlights a greater comparative range of objective and subjective knowledge levels for
the survey respondents (Table 8.9).
Table 8.8 Mean scores for knowledge and self-confidence (sensory and survey)
Mean SD
sens. surv. sens. surv.
Subjective knowledge chardonnay 4.55 4.07 1.72 1.72
Objective knowledge chardonnay (standardized) 2.93 2.92 1.46 1.51
Self-confidence (not product specific) 6.34 6.21 1.45 1.51
N = 263 (sens.) N = 274 (surv.)
Table 8.9 Mann-Whitney U test for comparison of knowledge and self-confidence
Objective Subjective
Self
knowledge knowledge
confidence
chardonnay chardonnay
Mann-Whitney U 34594.000 28920.000 33714.000
Wilcoxon W 71995.000 66321.000 71115.000
Z -.734 -3.897 -1.220
Asymp. Sig. (2-tailed) .463 .000 .222
The respective influence of knowledge and self-confidence quality determinations was
tested by the same methods used in the analysis of sensory data. First, correlation
coefficients were calculated to quantify any significant relationships between the
dependent variables of objective knowledge, subjective knowledge and self-confidence
and the independent variables of average importance COO, price and acid. Following
this, correlation coefficients were computed for the independent variables and attribute
level utility values to reveal any significant relationships specific to each. Tables 8.9
and 8.10 provide the results of the analysis, testing the sample as a whole. As the
majority of variables failed the tests for normality, non-parametric alternative methods
were used (See Appendix 3 Table A 3)
178
Table 8.10 Spearman's rho - Average importance chardonnay

Self
Knowledge knowledge Importance Importance
confidence
chardonnay chardonnay COO Price
Objective knowledge chardonnay Correlation Coefficient 1.000
Sig. (2-tailed) .
Subjective knowledge chardonnay Correlation Coefficient .478 1.000
Sig. (2-tailed) .010 .005 .
Average Importance COO Correlation Coefficient -.003 .020 -.066 1.000
Sig. (2-tailed) .963 .738 .277 .
Average Importance Price Correlation Coefficient -.005 -.035 .026 -.736 1.000
Sig. (2-tailed) .929 .567 .667 .000 .
Average Importance Acid Correlation Coefficient -.005 .012 -.020 .159 -.738
Sig. (2-tailed) .929 .848 .747 .009 .000
N = 274
As found in the analysis of the sensory data, there were no significant relationships for
the sample between knowledge (either type) or self-confidence and the average
importance placed on COO, price or acid. Reflecting previous results, a significant
relationship was found between objective and subjective knowledge (0.478) providing
further evidence of the positive link between real and perceived knowledge. Similarly,
there are significant, weak relationships between self-confidence and both objective
knowledge and subjective knowledge (0.155 and 0.170 respectively). These findings
are within expectations given previous results and the existing body of literature.
However, the relationships between knowledge (both types) and self-confidence are
very weak and therefore we cannot assume that self-confidence exerts a strong
influence on perceived knowledge or that high levels of knowledge underpin self-
confidence. The pattern of significant inter-correlations found between attributes and
utility values is similar to that found in the sensory experiment data analysis, and are
again, a reflection of the relative attribute and utility level trade-offs made by
consumers when scoring the product profiles. The favoring of one attribute involves
rating another more negatively (in comparison); hence a number of significant negative
correlations were found between utility values. However, only 1 significant relationship
179
was found between knowledge and self-confidence and product attribute utility values
(A correlation coefficient of 0.178 was found between subjective knowledge and the
utility value for $16.00). However, no noteworthy pattern signifying an effect of
knowledge or self-confidence on utility values was revealed. These results are
consistent with the results of sensory data analysis where, similarly, only 1 significant
relationship (0.128) was found between subjective knowledge and the utility value for
the U.S.
Table 8.11 Spearman's rho - Utilities chardonnay
United Above
Chile France $6.00 $16.00 $53.00 Average
States average
Objective knowledge chardonnay Correlation Coefficient -.022 .035 .001 -.060 .101 .005 .102 -.152
Sig. (2-tailed) .714 .564 .988 .324 .097 .940 .094 .012
Subjective knowledge chardonnay Correlation Coefficient .001 .012 .026 -.056 .178 -.044 .022 -.081
Sig. (2-tailed) .990 .848 .667 .353 .003 .473 .720 .181
Self confidence Correlation Coefficient .079 -.009 -.065 -.014 -.054 -.001 -.005 -.035
Sig. (2-tailed) .194 .887 .283 .818 .370 .980 .934 .570
Sig. (2-tailed) .
Sig. (2-tailed) .000 .000 .
Sig. (2-tailed) .002 .679 .005 .
$16.00 Correlation Coefficient .071 -.113 .047 -.229 1.000
Sig. (2-tailed) .245 .062 .438 .000 .
Sig. (2-tailed) .019 .643 .001 .000 .000 .
Average Correlation Coefficient -.206 .077 .128 .103 -.061 -.082 1.000
Sig. (2-tailed) .001 .207 .034 .089 .315 .176 .
Above average Correlation Coefficient .092 -.054 -.050 -.011 -.083 .037 -.317 1.000
Sig. (2-tailed) .128 .371 .411 .857 .169 .543 .000 .
High Correlation Coefficient .139 -.015 -.120 -.122 .102 .100 -.679 -.406
Sig. (2-tailed) .021 .808 .047 .043 .092 .098 .000 .000
N = 274
The inability of knowledge to influence cue usage may again be due to the relatively low
scores achieved by respondents on this characteristic. Additionally, the lack of
influence exerted by self-confidence amongst this group suggests this variable may be
less influential than indicated by previous studies (for this product at least). The next
step undertaken in the analysis was to segment the sample, in order to cluster
respondents with extreme levels of knowledge and self-confidence. Determination of
segments was achieved by replicating the methodology used in the analysis of data for
180
the sensory experiment: using quartile scores. Those achieving scores in the lowest
25% the data distribution were deemed to be low in knowledge (objective/subjective)
and self-confidence while respondents scoring in the highest 25% were considered to
possess high levels. Those scoring in the mid-range of the data array were eliminated
from further testing (Table 8.12).
The average importance for each product attribute, and attribute level utilities, were
calculated by high and low segment to allow comparison of results. Summaries of the
comparison of part worths for each segment are illustrated in this chapter along with
results of independent sample testing (comparing average utility values by segment) to
determine if significant differences existed between the groups.
Subjective Objective Self-

Percentiles
knowledge knowledge confidence
25th 2.62 3.00 5.20
50th 4.13 5.00 6.20
th
75 5.25 6.00 7.20
N = 274
8.4.5 Objective knowledge segments (chardonnay)
Table 8.13 provides an illustration of attribute average importance and utility values for
the high and low objective knowledge segments. For comparative purposes, the results
of the sensory data analysis are also included. Findings indicate little difference in
attribute average importance between the high and low objective knowledge segments
for COO. However, the more knowledgeable respondents showed a wider variance in
opinion regarding acid, strongly favoring the average level and punishing the high,
whereas, those with lower levels of objective knowledge reflected more polarized
opinions regarding price. However, as found in the analysis of the sensory data, in
181
spite of these variances in utility ranges, no significant differences in average utility
values were apparent (Table 8.14).
Table 8.13 Comparison of part worths and utilities objective knowledge chardonnay
Attribute High objective knowledge Low objective knowledge

and level Sensory Survey Sensory Survey
Ave imp COO 9.79 12.85 8.86 12.25
Chile -0.0151 -0.1695 -0.0274 -0.0913
U.S. -0.0072 -0.2147 -0.0850 -0.2103
France 0.1123 0.3842 0.1125 0.3016
Ave imp price 75.20 71.88 76.75 76.50
$ 6.00 -0.8043 -1.7514 -0.8999 -1.5853
$16.00 -0.6162 0.1525 0.0878 -0.0258
$53.00 0.8659 1.5989 0.8121 1.6111
Ave imp acid 15.01 15.27 14.39 11.25
Average 0.1268 0.4068 0.1536 0.1944
Above average 0.0797 -0.1017 0.0137 0.0813
High -0.2065 -0.3051 0.1674 -0.2758
N = 70 N = 59 N = 67 N = 56
Pearsons r 0.994 sig. .000 0.998 sig. .000 0.997 sig. .000 1.000 sig. .000
Kendalls tau 0.944 sig. .000 1.000 sig. .000 1.000 sig. .001 1.000 sig. .001
a
Test Statistics - Respondents with High and Low Objective Knowledge Chardonnay (survey)
Above
Chile United States France $6.00 $16.00 $53.00 Average High
average
Mann-Whitney U 4002.500 4088.500 4093.000 3971.500 3902.500 4333.000 3959.500 4383.500 4047.000
Wilcoxon W 16563.500 5684.500 5689.000 16532.500 5498.500 5929.000 5555.500 5979.500 16608.000
Z -1.060 -.845 -.833 -1.137 -1.312 -.229 -1.169 -.102 -.949
Asymp. Sig. (2-tailed) .289 .398 .405 .255 .189 .819 .242 .919 .343
a. Grouping Variable: Grouped objective knowledge chardonnay
As found in the sensory data analysis, results of correlation testing between the
average importance of COO, price and acid and the independent variables according to
high and low objective knowledge clusters revealed no significant relationships. For the
high objective knowledge segment, 4 significant and moderate correlations were
revealed between specific utility levels and the consumer characteristics tested. For
objective knowledge, a moderate, negative (-0.341) correlation was found with the
$6.00 utility and a moderate, positive (0.422) relationship was apparent with the $53.00
utility. While a moderate and positive (0.329) relationship was found between the utility
for $16.00 and subjective knowledge. Lastly, a moderate and positive (0.322)
182
correlation was seen between self-confidence and the utility for the U.S. No other
significant relationships were revealed in the matrix Therefore, whilst it may appear
that a high level of objective knowledge has some minimal and sporadic influence in
cue usage, it must be noted that these relationships are with the extrinsic cue of price,
with no effect of objective knowledge evident in consideration of acid levels; the cue
effecting objective product quality. In comparison, only 1 significant relationship was
found (-0.274) in the analysis of sensory data: that between self-confidence and the
utility value for France. No correlations were found between utility values and objective
knowledge. (See Appendix 8 Tables A 44 A 47). For the low objective knowledge
cluster in the survey, a single significant relationship was found between self-confidence
and the utility value for Chile (0.303), whereas in the sensory analysis 1 significant
correlation was found between objective knowledge and the utility for $53.00 and
subjective knowledge and U.S. Table 8.15 shows a comparative summary of significant
correlations between objective knowledge and attribute utilities for the high and low
knowledge segments (sensory experiment and conjoint survey) and demonstrates an
absence of any pattern of significant correlations for either stage of the research.
Table 8.15 Objective knowledge summary comparison of significant correlations
Above
Survey Chile U.S. France $6.00 $16.00 $53.00 Ave
ave
High
High objective knowledge -0.341 0.422

Subjective knowledge 0.329
Self-confidence 0.322
Low objective knowledge
Subjective knowledge
Sensory
High objective knowledge
Self-confidence -0.274
Low objective knowledge 0.224
Self-confidence
183
8.4.6 Subjective knowledge segments (chardonnay)
Table 8.16 illustrates the results of conjoint analysis for the high and low subjective
knowledge segments. The impact of COO is consistent across the segments and
reliance on price is similarly high. Whilst both groups recognized the average acid level
as the most favorable and the high acid level as the worst option, those with higher
levels of category knowledge were more favorable and respectively intolerant in their
assessments in relation to this variable. The result is the range is wider range between
the highest and lowest average utility value, and subsequently, a higher average
importance for acid than in the lower subjective knowledge group. Regardless of this
variance, price was the most influential attribute in their assessment of quality
expectations.
Attribute High subjective knowledge Low subjective knowledge

Ave imp COO 11.11 12.37 17.67 10.89
Chile -0.1776 -0.1730 -0.0365 -0.1127
U.S. -0.0698 -0.1968 -0.2454 -0.1925
France 0.1079 0.3698 0.2819 0.3052
Ave imp price 73.52 72.14 63.17 76.88
$ 6.00 -0.9873 -1.7921 -0.8823 -1.7230
$16.00 -0.0841 0.2794 0.1211 -0.0657
$53.00 0.9032 1.5127 1.0033 1.7887
Ave imp acid 15.37 15.49 19.17 12.23
Average 0.1558 0.3698 0.2819 0.2441
Above average 0.0841 -0.0302 0.0083 0.0704
High -0.2397 -0.3397 -0.2902 -0.3146
N = 70 N = 70 N = 67 N = 71
Pearsons r 0.996 sig .000 0.993 sig. .000 0.995 sig. .000 1.000 sig. .000
Kendalls tau 0.889 sig .004 0.889 sig. .004 1.000 sig. .002 1.000 sig. .001
184
a
Test Statistics - Respondents with High and Low Subjective Knowledge Chardonnay (survey)
Above
average
Mann-Whitney U 2392.500 2461.500 2258.500 2479.000 1759.000 2114.000 2318.000 2203.500 2394.500
Wilcoxon W 4877.500 4946.500 4814.500 4964.000 4315.000 4599.000 4874.000 4688.500 4879.500
Z -.382 -.097 -.936 -.025 -3.000 -1.531 -.690 -1.163 -.374
Asymp. Sig. (2-tailed) .702 .923 .349 .980 .003 .126 .490 .245 .709
a. Grouping Variable: Grouped subjective knowledge chardonnay
Results of testing for the high and low subjective knowledge groups are comparable
with the findings from the sensory data with a noteworthy exception. In the experiment,
the low subjective knowledge group was more sensitive to the variance in acid levels
than the high subjective knowledge group. As discussed in chapter 7, their greater
responsiveness to changing acid levels in comparison to the high subjective knowledge
cluster may be due to a lack of bias based on pre-conceived beliefs regarding price
overcoming their attendance to taste.
The correlation matrix illustrating relationships between attribute average importance
and subjective knowledge for the low subjective knowledge segment revealed 2
significant relationships (See Appendix 8 Tables A 48 A 51). A weak and positive
correlation (0.339) was found between subjective knowledge and price and a low and
negative correlation (-0.273) was found between subjective knowledge and acid. The
correlation between the average importance of price and subjective knowledge for this
group was supported by the 2 significant relationships found in the utilities correlation
matrix. A negative correlation (-0.363) was found between subjective knowledge and
the utility for $6.00 and a positive (0.320) for the utility for France. These results
indicate that increases in subjective knowledge (amongst an overall low knowledge
group) led to an even greater reliance on the extrinsic cues, rather than intrinsic ones,
to determine quality.
185
Similar to the results found for the objective knowledge segments, for the high
subjective knowledge groups, low but significant relationships existed between objective
knowledge and the utility values for $6.00 (-0.235) and $53.00 (0.291). These were
supported by a weak and positive relationship (0.262) between subjective knowledge
and the utility for $53.00. Importantly, once again, no significant relationships were
found between subjective knowledge and acid utilities, indicating that increased
category familiarity leads to higher reliance on a belief in the price/value schema.
In comparison, results from analysis of the sensory data (Table 8.16) showed more
reliance on the other communicated extrinsic cue, COO, with no relationships evident
with acid levels. As these relationships with COO utilities were amongst the low
subjective knowledge group, a lower level of category familiarity (and retail wine prices)
in their case may have led to an emphasis on CI rather than on expectations regarding
a price influence.
Table 8.18 Subjective knowledge summary comparison of significant correlations
Above
ave
High
Objective knowledge -0.235 0.291

High subjective knowledge 0.262
Self-confidence
Objective knowledge
Low subjective knowledge -0.363 0.320
Self-confidence
Sensory
Objective knowledge
High subjective knowledge
Self-confidence 0.363 0.241
Objective knowledge
Low subjective knowledge -0.334 0.302
Self-confidence
186
8.4.7 Self-confidence segments (chardonnay)
Table 8.19 shows the computed average attribute importance and individual average
utility values for each level, according to the high and low self-confidence clusters.
There was minimal difference in opinion regarding attributes or levels between those
with higher and lower levels of self-confidence. For both segments the extrinsic cues
were more powerful than the cited intrinsic cue, with price again dominating the quality
ratings. Results of testing (Table 8.20) show that there were no significant differences
in average utility values according to self-confidence clusters. Comparison of results
with the sensory experiment revealed some influence of self-confidence when the
products were being experienced; those with high self-confidence rated the importance
of acid much higher than any other segment analyzed. Interestingly, the utility value of
the high acid level was much lower for those surveyed than for respondents who tasted
the low quality product.
Table 8.19 Comparison of part worths and utilities self-confidence
Attribute High self-confidence Low self-confidence

Ave imp COO 4.34 15.66 19.44 16.88
Chile -0.0422 -0.1310 -0.1984 -0.2864
U.S. -0.0578 -0.2902 -0.1413 -0.1878
France 0.0156 0.4212 0.3397 0.4742
Ave imp price 74.28 71.19 70.83 69.38
$ 6.00 -0.8222 -1.6683 -0.9698 -1.6385
$16.00 -0.0689 0.1028 0.0984 0.1502
$53.00 0.8911 1.5655 1.0683 1.4883
Ave imp acid 21.39 13.14 9.72 13.75
Average 0.2378 0.2919 0.2222 0.2629
Above average 0.0178 -0.0133 0.1511 0.0939
High -0.2556 -0.3051 -0.1289 -0.3568
N = 78 N = 70 N = 58 N = 71
Pearsons r 0.963 sig .000 1.000 sig. .000 0.990 sig. .000 0.997 sig. .000
Kendalls tau 0.817 sig .002 1.000 sig. .001 0.833 sig. .001 0.889 sig. .004
187
a
Test Statistics - Respondents with high and low self confidence Chardonnay (survey)
Above
average
Mann-Whitney U 2054.000 2111.500 2330.500 2259.000 2234.500 2312.500 2296.000 2179.500 2227.500
Wilcoxon W 4610.000 4389.500 4608.500 4537.000 4512.500 4868.500 4574.000 4457.500 4783.500
Z -1.385 -1.140 -.205 -.510 -.615 -.281 -.352 -.851 -.644
Asymp. Sig. (2-tailed) .166 .254 .838 .610 .539 .778 .725 .395 .519
a. Grouping Variable: Grouped self confidence
Table 8.21 Self-confidence summary comparison of significant correlations
Above
ave
High
Objective knowledge -0.235

Subjective knowledge -0.286 0.262
High self-confidence
Objective knowledge
Low self-confidence -0.251
Sensory
Objective knowledge
subjective knowledge -0.246 0.371
High Self-confidence
Objective knowledge
Low self-confidence 0.285 -0.354
Analysis of correlations between the consumer characteristics and attribute average
importance for the high self-confidence group revealed 2 significant relationships. A
weak and positive (0.240) correlation was found between subjective knowledge and the
average importance of price, and a low and negative (-0.303) correlation existed
between subjective knowledge and acid. For the low self-confidence group, 1
significant relationship was found between self-confidence and the average importance
of acid (0.261) (See Appendix 8 Tables A 52 A 55). The correlations with subjective
knowledge closely replicate those found in the analysis of the subjective knowledge
segments. As discussed in chapter 7, the segments are not mutually exclusive
respective to all segments. While the same respondent cannot possess high and low
188
levels of a specific attribute, they may be represented in more than one segment;
therefore, some replication of results is likely to occur.
Scrutiny of the correlation matrixes for the high self-confidence segments and attribute
utilities reveal 3 significant relationships, and 2 emerged from testing undertaken in the
low self-confidence group (Table 8.21). All of the relationships found were weak and
only 1 involved self-confidence. This is seen in the analysis for the low self-confidence
group, with a weak and negative (-0.251) relationship between self-confidence and the
utility value for $53.00. In comparison, the data from the sensory experiments revealed
no relationships between self-confidence and price, confirming earlier findings in
relation to subjective knowledge and COO utilities.
8.5 Chardonnay summary
Results of analysis of the survey data generally support the findings of the sensory
data, both demonstrating the power of extrinsic cues to dominate consumer quality
perceptions, and also the ability of the survey conjoint analysis method to predict these
results. The sample was clustered to isolate those respondents with respectively high
and low levels of knowledge (both types) and self-confidence and each segment was
tested to quantify the number and of nature of any significant relationships between
these consumer characteristics and cue usage. While a number of significant
correlations were found between the independent and dependant variables tested in the
survey data (consistent with analysis of the sensory data) they were few in number, only
weak to moderate in magnitude and confined exclusively to the extrinsic cues of price
and COO. Comparison of utility values between high and low level clusters showed
general differences between groups tested, but only 1 instance of a significant
difference was found. Therefore, no consistent pattern of strong relationships between
189
knowledge and self-confidence and cue usage was found to exist in this stage of the
study
In terms of comparison between the analysis of results for sensory data and the survey
data, the opportunity to taste the wine and experience the effect of higher levels of acid
can be seen in the comparison of average importance and utility levels for that attribute,
particularly for the high and low self-confidence and subjective knowledge groups.
However, even though these respondents experiencing sour wine showed a degree of
reaction to it, the credence they placed on price remained stronger. Given the vastly
different circumstances of participation for the survey respondents and compared to the
strictly controlled environment experienced by the sensory participants the consistency
in outcomes is significant. However, scrutiny of test results for both aspects of the
research show no significant correlations were found in either aspect of the study
between the independent variables and any of the 3 utility levels for acid. Hence no
empirical link between the consumer characteristics of knowledge (both types) and self-
confidence and the use of intrinsic cues to assess quality has been established using
either methodology. Rather, the review shows 14 significant relationships between the
tested independent variables and price utilities. In the main, these reflect a universal
consumer belief in the price/value schema as similar relationships emerged for clusters
regardless of knowledge or self-confidence levels. A further 12 significant relationships
were revealed between knowledge and self-confidence and the utility values for COO.
These tended to illustrate a preference for France as a source country over the Chile
and the U.S., confirming both groups are influenced in their quality judgments by a
positive CI congruent with the product evaluated. Based on the findings across both
studies, only 1 of the proposed hypotheses was supported in the testing specific to
chardonnay (Table 8.22).
190
Table 8.22 Hypotheses testing outcomes (chardonnay)
Objective knowledge is positively associated with the relative

H1 Not
contribution of intrinsic cues to an individuals assessment of product Supported
quality.
Objective knowledge is positively associated with the relative contribution of an Not

H1a
intrinsic cue (acid) to an individuals assessment of product quality (wine). Supported
Objective knowledge is negatively associated with the relative

Not
H2 contribution of extrinsic cues to an individuals assessment of product Supported
quality.
Objective knowledge is negatively associated with the relative contribution of an Not
H2a
extrinsic cue (COO) to an individuals assessment of product quality (wine / cheese). Supported

H2b
extrinsic cue (price) to an individuals assessment of product quality (wine / cheese). Supported
Subjective knowledge is positively associated with the relative

Not
H3 contribution of intrinsic cues to an individuals assessment of product Supported
quality.
Subjective knowledge is positively associated with the relative contribution of an Not
H3a
intrinsic cue (acid) to an individuals assessment of product quality (wine). Supported
Subjective knowledge is negatively associated with the relative

Not
quality.
Subjective knowledge is negatively associated with the relative contribution of an Not
H4a

H4b
Self-confidence is positively associated with the relative contribution Not

H5 Supported
of intrinsic cues to an individuals assessment of quality.
Self-confidence is positively associated with the relative contribution of an intrinsic Not
H5a
cue (acid) to an individuals assessment of product quality (wine). Supported
Self-confidence is negatively associated with the relative contribution Not

H6 Supported
of extrinsic cues to an individuals assessment of quality.
Self-confidence is negatively associated with the relative contribution of an extrinsic Not

H6a
cue (COO) to an individuals assessment of product quality (wine / cheese). Supported

H6b
cue (price) to an individuals assessment of product quality (wine / cheese). Supported
The paper based conjoint analysis method will be predictive of the sensory
H7 based test in terms of the relative average importance of tested product Supported
attributes (intrinsic and extrinsic)
191
9 Conjoint analysis survey results for brie
9.1 Introduction
Chapter 8 provided an overview of the methodology used to conduct the conjoint
analysis by survey and subsequent data analysis results specific to chardonnay; the
chapter also included a comparison of the survey data analysis with the sensory data
analysis for chardonnay. The chapter concluded with a presentation of hypotheses
testing outcomes. Chapter 9 illustrates the results of survey data analysis and
comparative discussion respective to brie, concluding with a presentation of hypotheses
testing outcomes.
9.2 Brie conjoint analysis results
As found with chardonnay, price was considered the most important attribute
contributing to perception of brie quality; however, these respondents considered fat as
important as COO in forming their opinions (Table 9.1). Further, while price was still
considered the most important, the view appears more balanced across the attributes
with COO at 24% as compared to 40% for price, and fat only approximately 5% less
influential than price at 35%. Results of the conjoint analysis showed remarkable
consistency between the opinions of respondents in both stages of the research in
relation to which attributes were most important to rating quality. However, in contrast
to the findings from the sensory experiment, the full fat brie was considered very
undesirable for the survey group with a utility value of -0.5337, with both higher fat
cheeses receiving favorable values. Overall, the triple cream was believed to be the
best option, in line with the literature and expectations, but in stark contrast to the taste
testing data. Many respondents who tasted the triple cream brie said they found it too
rich and somewhat sickening, and rated these cheeses according to this negative
perception. In the survey situation, respondents were left to imagine (or remember if
192
they had product experience) what the product would taste like. Clearly, they imagined
(or remembered) the experience favorably and rated the products accordingly. Table
9.1 illustrates the average utility value for each profile and shows that the combination
of higher price and desirable fat levels result in higher utility values. However, the third
ranked brie illustrates the more balanced view across the attributes where a low price
is accepted when the fat level and the COO are the most favored. Due to this reversal
of beliefs compared to the sensory data, profile rankings for brie are considerably
different for each of the two stages of the study (Table 9.2).
Table 9.1 Summary of part worths and utilities brie
Attribute Research stage

and level SENSORY SURVEY
Ave imp COO 25.07 24.33
Argentina -0.1841 -0.2320
Canada 0.0208 -0.1200
France 0.1633 0.3520
$28.95 -0.3405 -0.4692
$49.00 0.2689 -0.0361
$69.95 0.0717 0.5053
Fat 30.95 35.07
Full cream -0.0148 -0.5337
uble cream 0.2218 0.2255
Triple cream -0.2070 0.3082
N = 263 N = 274
Table 9.2 Comparative average values per profile brie (sensory and survey)
Min Max Mean SD

Profile Attribute levels
sens sur. sens sur. sens sur. sens sur.
621 France $69.95 Double cream 0.78 3.33 10.89 11.00 7.29 8.16 1.73 1.31
380 Argentina $69.95 Triple cream 0.33 2.78 10.22 10.44 6.51 7.66 2.36 1.59
735 France $28.95 Triple cream 0.56 1.78 11.44 11.56 6.45 7.27 2.23 1.58
178 Canada $49.95 Triple cream 0.44 1.89 11.22 10.33 6.90 7.23 2.33 1.46
507 Argentina $49.95 Double cream 1.11 1.56 11.67 9.78 7.13 7.03 1.87 1.52
367 Canada $69.95 Full cream 1.11 1.22 11.11 20.78 6.91 6.93 1.91 2.39
709 France $49.95 Full cream 1.11 1.44 11.22 10.22 7.24 6.86 1.88 1.85
266 Canada $28.95 Double cream 1.44 0.89 11.22 10.22 6.73 6.71 1.84 1.54
139 Argentina $28.95 Triple cream 0.11 0.11 10.78 12.11 6.10 6.68 2.12 1.67
810 France $28.95 Full cream 0.78 1.11 10.89 10.89 7.29 6.43 1.73 1.79
393 Argentina $28.95 Full cream 1.11 0.78 10.33 10.00 6.29 5.84 1.85 1.78
N = 263 (sensory) N = 274 (survey)
193
9.2.1 Brie profiles and likelihood of purchase
Table 9.3 Average brie profile values by purchase intentions (survey)
Yes No
Mean SD Mean SD
810 6.46 1.81 6.32 1.74 6750.00 -0.39 .696
139 7.30 1.47 6.09 1.55 5257.00 -6.04 .000
367 7.83 1.51 6.62 2.57 5075.00 -3.81 .000
266 7.24 1.29 5.90 1.56 4414.50 -7.07 .000
709 7.46 1.33 6.54 2.01 6361.00 -3.55 .000
380 8.29 1.23 7.45 1.67 5069.50 -3.62 .000
735 7.75 1.33 6.27 1.59 3928.00 -6.95 .000
393 6.80 1.37 5.25 1.75 4361.50 -7.11 .000
621 8.21 1.21 8.14 1.37 8626.00 -0.23 .818
178 7.61 1.12 7.01 1.60 7052.00 -2.76 .006
507 7.56 1.03 6.80 1.64 6060.50 -3.40 .001
N = 274
Table 9.3 shows the average total utility value for each brie sample grouped by
consideration of purchase. As with chardonnay, the mean values are generally
significantly higher for those products respondents considered purchasing as compared
to those they would not, linking the perception of quality positively with purchase
intention.
9.2.2 Knowledge levels brie
Consistent with previous results, the general level of objective knowledge amongst this
group was found to be quite low. Results from the objective knowledge test found that
respondents achieved an average of 4.47 correct answers out of a possible 14 and 83%
scored 7 correct answers or less. Respondents have consistently overestimated their
expertise in relation to this product even in comparison to relatively modest levels of
self-assessed knowledge (Table 9.4). As self-confidence is not product specific, the
score determined previously will be used for the analysis of brie also. Table 9.5 shows
194
the results of testing for significant differences between the samples in terms of
knowledge. As found with chardonnay, there is no significant difference in average
objective knowledge levels; but there is one for subjective knowledge. Whilst the
difference is significant, the degree of difference is minimal.
Table 9.4 Mean scores for knowledge (sensory and survey)
Mean SD
sens. surv. sens. surv.
Subjective knowledge brie 4.59 4.01 0.80 1.94
Objective knowledge brie (standardized) 2.95 2.88 0.51 1.80
Self-confidence (not product specific) 6.34 6.21 1.54 1.51
N = 263 (sensory) N = 274 (survey.)
Table 9.5 Mann-Whittney U test for comparison of knowledge between groups
Objective Subjective
knowledge knowledge
brie brie
Mann-Whitney U 35359.000 32274.000
Wilcoxon W 70075.000 66990.000
Z -.376 -2.091
Asymp. Sig. (2-tailed) .707 .037
(brie)
Tables 9.6 and 9.7 illustrate the correlation coefficients calculated to investigate
significant relationships between knowledge and self-confidence and tested product
cues. As found previously with chardonnay (sensory and survey), there was significant
inter-correlation (0.548) between subjective and objective knowledge due to the
simultaneous attainment of both (Alba and Hutchinson 1987; Park, Mothersbaugh et al.
1994; Alba 2000). Also, consistent with previous analysis (both products: survey and
sensory), significant correlations were found between levels within an attribute as well
as between the attributes themselves, representing the relative degree of relative trade
195
off occurring between the attributes/levels as part of the conjoint analysis utility
calculations. However, unlike the results of previous analysis, for this group 4
significant correlations emerged between consumer characteristics and the average
importance of 2 product attributes. There was a weak and negative relationship (-
0.208) between objective knowledge and price, and a correspondingly weak and
positive (0.215) relationship between objective knowledge and fat. These relationships
are replicated with subjective knowledge. Whilst these correlations are weak, they
provide the first suggestion that knowledge type and levels may influence consumer use
of cues.
Table 9.6 Spearmans rho Average attribute importance brie

Self
knowledge knowledge importance importance
confidence
brie brie COO price
Objective knowledge brie Correlation Coefficient 1.000
Sig. (2-tailed) .
Subjective knowledge brie Correlation Coefficient .548 1.000
Sig. (2-tailed) .200 .002 .
Average importance COO Correlation Coefficient -.041 .083 -.013 1.000
Sig. (2-tailed) .497 .173 .828 .
Average importance price Correlation Coefficient -.208 -.230 .000 -.394 1.000
Sig. (2-tailed) .001 .000 .997 .000 .
Average importance fat Correlation Coefficient .215 .160 .004 -.317 -.623
Sig. (2-tailed) .000 .008 .949 .000 .000
N = 274
Table 9.7 reveals the basis of these correlations in relationships found between the
tested consumer characteristics and the attribute utility values from the conjoint
analysis. Weak but positive and significant correlations were found between objective
knowledge and double cream (0.186) and triple cream (0.303) and a correspondingly
weak and negative correlation (-0.289) with full cream. There is also an indication that
higher levels of objective knowledge led to a lesser reliance on price to assure quality.
Similar relationships were found between these utilities and subjective knowledge.
COO did not appear to effect opinions in a significant way, with the exception of 1 low
196
and negative (-0.137) between subjective knowledge and the utility for Canada. For
self-confidence, no significant relationships emerged.
Table 9.7 Spearmans rho Utilities brie

Objective knowledge brie Correlation Coefficient -.004 -.116 .113 .126 .020 -.127 -.289 .186 .303
Sig. (2-tailed) .948 .055 .061 .037 .745 .036 .000 .002 .000
Subjective knowledge brie Correlation Coefficient .028 -.137 .089 .215 -.056 -.165 -.215 .076 .298
Sig. (2-tailed) .643 .023 .143 .000 .359 .006 .000 .212 .000
Self confidence Correlation Coefficient -.020 .003 -.021 -.043 -.076 .067 .019 -.096 .054
Sig. (2-tailed) .742 .955 .727 .475 .211 .272 .751 .112 .377
Sig. (2-tailed) .
Sig. (2-tailed) .000 .000 .
$28.95 per kilo Correlation Coefficient .018 -.143 .170 1.000
Sig. (2-tailed) .767 .018 .005 .
$49.00 per kilo Correlation Coefficient .003 -.045 .054 -.352 1.000
Sig. (2-tailed) .958 .454 .375 .000 .
$69.95 per kilo Correlation Coefficient -.039 .203 -.192 -.742 -.277 1.000
Sig. (2-tailed) .520 .001 .001 .000 .000 .
Full Cream Correlation Coefficient -.102 .192 -.089 -.264 -.056 .341 1.000
Sig. (2-tailed) .091 .001 .142 .000 .356 .000 .
Double Cream Correlation Coefficient .091 -.151 .046 .172 .158 -.280 -.749 1.000
Sig. (2-tailed) .134 .013 .449 .004 .009 .000 .000 .
Triple Cream Correlation Coefficient .112 -.191 .089 .254 -.053 -.266 -.774 .245 1.000
Sig. (2-tailed) .065 .002 .143 .000 .379 .000 .000 .000 .
N = 274
As with previous components of study, the next step was to segment the sample and
cluster those individuals with the most extreme levels of knowledge and self-confidence.
The methodology for segmentation was identical to that used previously: according to
quartile score values where the lowest 25% would be deemed to possess low levels of
the characteristic and those in the highest 25% of the data array would be deemed to
have high levels with all other respondents ignored. Table 9.8 illustrates the quartile
values used for segmentation purposes; the previously computed segments for self-
confidence used for testing respective of brie. Correlation matrices for all segments are
seen in Appendix 9 Tables A68 through A 78.
197
Subjective Objective Self-

Percentiles
knowledge knowledge confidence
25th 2.38 2.00 5.20
50th 4.00 4.00 6.20
th
75 5.50 7.00 7.20
N = 274
9.2.3 Objective knowledge segments (brie)
Table 9.9 Comparison of part worths and utilities objective knowledge
Attribute High objective knowledge Low objective knowledge

Ave imp COO 35.63 24.85 9.69 26.61
Argentina -0.2495 -0.3040 -0.1250 -0.2397
Canada -0.0448 -0.1698 -0.0119 0.0079
France 0.2943 0.4738 0.1131 0.2317
Ave imp price 39.46 25.89 35.05 67.74
$28.95 -0.3372 -0.3827 -0.5179 -0.5730
$49.00 -0.2651 -0.0448 0.3929 -0.0540
$69.95 0.0721 0.4275 0.1250 0.6270
Ave imp fat 24.90 49.26 53.27 5.65
Full cream -0.2144 -0.9799 0.3452 -0.0587
Double cream 0.1657 0.4182 0.4821 0.0413
Triple cream -0.0487 0.5617 -0.8274 0.0175
N = 57 N = 72 N = 56 N = 70
Pearsons r 0.979 sig .000 0.995 sig .000 0.994 sig .000 0.990 sig .000
Kendalls tau 0.944 sig .000 0.986 sig .000 0.944 sig .001 1.000 sig .000
Table 9.9 illustrates the comparison of part worths and utilities for the objective
knowledge segments and also provides the comparison of results from the analysis of
sensory data. The average importance given to COO is comparable for both segments,
with identical ranking of countries. However, for price, the low objective knowledge
group placed far more importance on price than did those with high objective
knowledge, due to their somewhat indifferent response to varying fat levels.
Conversely, those with better understanding of the implications of fat level variances
198
paid more credence to that cue than to price. Results from the sensory experiment
provide insight into the potential power of experience over expectations. When tasting,
the lower knowledge group graded products more strictly according to their sensory
experiences whilst those participating in the survey with low knowledge could only base
their quality ratings on expectations. Without objective knowledge to draw on, the low
objective knowledge survey segment reacted more to differences in price, and to a
lesser degree, COO. However, comparison of mean utility values between groups
(Table 9.10) show these differences to only reflect the degree of extremities in scores
as none of the averages are found to be significant.
a
Test Statistics - Respondents with High and Low Objective Knowledge Brie (survey)
Mann-Whitney U 368.000 336.000 304.000 300.000 358.500 338.500 276.000 284.500 270.500
Wilcoxon W 1071.000 1039.000 535.000 531.000 1061.500 1041.500 979.000 515.500 501.500
Z -.333 -.852 -1.374 -1.437 -.488 -.811 -1.824 -1.691 -1.916
Asymp. Sig. (2-tailed) .739 .394 .169 .151 .626 .417 .068 .091 .055
a. Grouping Variable: Objective knowledge brie
Examination of the correlation matrix for attribute average importance and consumer
characteristics (high objective knowledge) revealed 1 weak and negative relationship
(-0.242) between price and subjective knowledge. There was also 1 weak and negative
(-0.310) correlation between subjective knowledge and the utility value for $49.00 (See
Appendix 9 Tables A 68 A 71). The correlation matrix for attribute average
importance and consumer characteristics (low objective knowledge) showed 3
significant correlations. The first was a weak and positive (0.257) relationship between
subjective knowledge and COO, the other 2 were between objective knowledge and
price (-0.251) and fat (0.265), suggesting weak links between knowledge and the
influence of respective cues for this group. For the low objective knowledge cluster, the
correlation matrix illustrating consumer characteristics and utility values revealed 5
significant correlations, with 4 relationships involving objective knowledge. A summary
199
of these significant relationships, and those found from analysis of the sensory data, is
shown in Table 9.11. Whilst none of the relationships could be considered strong (the
highest is 0.377) there is evidence for this cluster, that knowledge is linked to cue
usage. Amongst the low knowledge group, those with better product understanding
paid less attention to the extrinsic cues, favoring the intrinsic indicators of quality
instead.
Table 9.11 Objective knowledge summary comparison of significant correlations (brie)
Full Double Triple

Survey Argentina Canada France $28.95 $49.00 $69.95
cream cream cream
High objective knowledge 0.310
Self-confidence
Low objective knowledge 0.295 -0.271 -0.285 0.377
Self-confidence
Sensory
High objective knowledge
Subjective Knowledge -0.370
Low objective knowledge
Subjective Knowledge
Self-confidence
9.2.4 Subjective knowledge segments (brie)
Comparison of part worths and utilities for the high and low subjective knowledge
groups (Table 9.12) illustrated the respective differences in category experience
between the 2 groups and the impact this had on their opinions regarding fat in relation
to quality ratings. Average importance of attributes for the high subjective knowledge
group is comparable to those for the high objective group, with an identical ranking of
utility values for each. For those with low levels of subjective knowledge, price
overwhelmed the influence of COO and fat level again. This group assessed their own
knowledge as low; therefore, their heavy reliance on price, moderate use of COO and
disregard for fat levels are not surprising. Comparison with the results from the sensory
data shows similar results as those discussed for the high and low objective knowledge
200
groups. Not bound by assumptions based on category experience, this group went
along with an expectation that a higher price means higher quality. Conversely,
respondents who were able to taste the different products rated according to their taste
preferences. Comparison of means testing shows, however, that average utility values
were not significantly different in spite of the variance in utility values (Table 9.13).
Attribute High subjective knowledge Low subjective knowledge

Ave imp COO 28.37 26.53 11.38 26.00
Argentina -0.1732 -0.2029 -0.0415 -0.3094
Canada -0.0458 -0.2850 -0.0614 -0.0171
France 0.2190 0.4879 0.1028 0.3265
Ave imp price 29.43 21.89 44.48 58.28
$28.95 -0.2222 -0.2705 -0.3847 -0.2274
$49.00 0.1846 -0.0966 0.2570 -0.0376
$69.95 0.0376 0.3671 0.1277 0.7316
Ave imp fat 42.20 51.58 44.14 15.72
Full cream -0.3350 -0.8986 0.2322 -0.2274
Double cream 0.0866 0.2947 0.1725 0.1573
Triple cream 0.2484 0.6039 -0.4046 0.0701
N = 68 N = 69 N = 67 N = 65
Table 9.13 Comparison between high and low subjective knowledge groups
b
Test Statistics - Respondents with High and Low Subjective Knowledge Brie (survey)
Mann-Whitney U 19.000 2.000 13.000 10.500 17.500 11.500 8.500 16.500 7.000
Wilcoxon W 22.000 5.000 203.000 200.500 20.500 14.500 11.500 206.500 197.000
Z .000 -2.051 -.720 -1.022 -.181 -.902 -1.261 -.301 -1.443
Asymp. Sig. (2-tailed) 1.000 .040 .472 .307 .857 .367 .207 .763 .149
Exact Sig. [2*(1-tailed Sig.)] 1.000a .038a .533a .343a .857a .400a .238a .771a .190a
a. Not corrected for ties.
b. Grouping Variable: Subjective knowledge brie
201
As with the sensory data analysis, no significant relationships were found to exist in the
survey data and for either cluster between knowledge, self-confidence and the attribute
average importance (See Appendix 9 Tables A 72 A 75). Table 9.14 illustrates the
significant correlations found between utility values for the respective groups and also
illustrates a comparison with those found in the analysis of the sensory data. For the
high subjective knowledge segment, only 1 significant correlation (0.279) was found and
that is between objective knowledge and the utility for double cream. For the low level
cluster there were 2; the first was a weak and negative relationship (-0.302) between
objective knowledge and the utility for full cream, and the other was a weak and positive
correlation between self-confidence and the utility for Canada. The relationship
between the utility value and self-confidence was similar in nature to instances identified
in other components of the data analysis (both sensory and survey). Those with higher
levels of self-confidence were less likely to punish a CI not congruent with the
particular product being assessed (e.g. Canada and cheese). The relationship between
objective knowledge and the utility for full cream is a reflection of a better understanding
of the impact of fat levels on quality. However, as found previously, the relationships
were weak and indicate that knowledge and self-confidence exerted very little influence
overall. Importantly, no significant correlations were found between subjective
knowledge (high or low) and any of the utility values. In comparison, 2 significant
correlations were found to exist between the tested consumer characteristics and utility
values in the analysis of the sensory data; although these also indicated little overall
influence.
202
Table 9.14 Subjective knowledge summary comparison of significant correlations (brie)
Full Double Triple

cream cream cream
Objective knowledge 0.279
Self-confidence
Low subjective knowledge
Sensory
Objective knowledge
Objective knowledge 0.310 0.295 0.262
Low subjective knowledge 0.260 0.246
Self-confidence
9.2.5 Self-confidence segments (brie)
Table 9.15 shows the comparison of part worths and utilities for the high and low self-
confidence segments. Results indicate little difference in opinions between the groups
in the survey in relation to attribute average importance or utility values. This is
confirmed in table 9.16 showing there were no significant differences between average
utility values for the clusters. Both segments in the survey sample showed a higher
regard for price than fat, with COO also very influential with these segments exhibiting
the highest reliance on extrinsic cues compared to the intrinsic cue described. This
may have been a reflection of stubbornness, supported by their confidence, related to
strongly held beliefs regarding price or COO. Reviewing the correlation matrices for
these segments suggests that the influence of self-confidence on cue usage was
negligible. Consistent with the findings for knowledge segments, those actually tasting
the triple cream brie often found it unpleasant, in the case of anticipated liking, this level
of fat is expected to be more desirable than the lower levels. Interestingly, those with
low levels of self-confidence in the sensory experiment placed the highest credence on
taste in comparison to all other segments across both studies. This is surprising, as
203
these respondents would be expected to rely most heavily on the extrinsic cues
provided. In contrast, those with higher levels were more responsive to COO attribute
levels than others indicating a strong reluctance to abandon beliefs regarding country
associations. Respondents in the survey, basing opinions on expectations alone
showed little variation in their expectations based on self-confidence levels.
Table 9.15 Self-confidence summary comparison of significant correlations (brie)
Attribute High self-confidence Low self-confidence

Ave imp COO 34.21 27.73 15.28 29.66
Argentina -0.2151 -0.2570 -0.1871 -0.2254
Canada -0.0142 -0.1426 0.1345 -0.1549
France 0.2293 0.3997 0.0526 0.3803
Ave imp price 27.63 41.60 43.33 37.01
$28.95 -0.1980 -0.44.61 -0.5673 -0.3474
$49.00 0.1610 -0.0929 0.3450 -0.0610
$69.95 0.3700 0.5390 0.2222 0.4085
Ave imp fat 38.16 30.67 41.39 33.33
Full cream 0.0880 -0.4113 0.1345 -0.4366
Double cream 0.2037 0.0962 0.3684 0.2441
Triple cream -0.2920 0.3151 -0.5029 0.1925
N = 78 N = 67 N = 71 N = 65
Scrutiny of the correlation matrices showing relationships between attribute average
importance and knowledge and self-confidence for the high and low self-confidence
clusters revealed only 1 significant correlation (See Appendix 9 Tables A 76 A 78).
For the high self-confidence group, an isolated relationship (0.277) between objective
knowledge and fat was found, reflecting earlier findings. This outcome is consistent
with the findings from the sensory data, where no significant relationships were found.
Table 9.16 shows a summary of comparisons of significant relationships between
consumer characteristics and utility values (by segment). The significant correlations
revealed in the analysis of sensory data for the same segments are also shown. Whilst
there were no significant relationships noted for self-confidence in the sensory data, 1
weak and negative (-0.254) correlation was found for France amongst the high self-
204
confidence cluster (survey respondents). This, once again, demonstrates (albeit
weakly) that high levels of self-confidence can lessen the influence of a strong CI.
However, in examining the results for both segments, in both stages of research, there
was no pattern of strong influence found that can be specifically attributed to self-
confidence levels.
Table 9.16 Self-confidence summary comparison of significant correlations (brie)
Full Double Triple

cream cream cream
Objective knowledge 0.327 0.250
Subjective knowledge 0.312 0.249
High self-confidence -0.254
Objective knowledge
Low self-confidence
Sensory
High self-confidence
Low self-confidence
a
Test Statistics - Respondents with High and Low Self Confidence Brie (survey)
Mann-Whitney U 2306.500 2370.000 2376.000 2277.500 2098.500 2212.000 2316.500 2115.000 2012.500
Wilcoxon W 4584.500 4648.000 4932.000 4555.500 4376.500 4768.000 4594.500 4393.000 4568.500
Z -.307 -.036 -.011 -.431 -1.198 -.711 -.264 -1.126 -1.563
Asymp. Sig. (2-tailed) .759 .971 .991 .667 .231 .477 .791 .260 .118
a. Grouping Variable: Grouped self confidence
205
9.3 Brie summary
The power of price and COO to influence expectations of quality was illustrated again in
the analysis of results for Brie. Price was found to be the most important influence on
product expectations for the sample (as whole) and for all segments except those
possessing high levels of objective and subjective knowledge. For these consumers,
the intrinsic cue was the most influential. Fat levels also exceeded the importance
placed on COO for the group, and for all segments with the exception of the low self-
confidence cluster. Overall, COO levels contributed approximately 25% to the overall
quality ratings, and therefore, should not be discounted.
In the survey, respondents were assessing products according to their expectations
rather than sensory experience and this critical difference is clearly seen in the utility
values computed for the various levels of fat. Many participating in the taste tests found
the triple cream brie unpleasant; conversely, most completing the survey favored the
description of higher cream levels in expectation of better taste (a reaction in line with
expectations and congruent with the literature). Whilst those with high levels of
objective and subjective knowledge correctly placed the greatest importance on fat
levels over price and COO, these outcomes were not supported with a strong pattern of
significant correlations between utility values and knowledge (either type), although
several weak and isolated examples were found. Conversely, for the low objective
knowledge group, 4 instances (albeit weak) of significant correlations were found
between objective knowledge and 2 fat and 2 price utility values. Corresponding
relationships were also found in the analysis of the low subjective knowledge and self-
confidence segments. As explained previously, the emergence of repeated or similar
correlations across different segments is to be expected as some respondents may be
present in more than 1 cluster since segments were only mutually exclusive according
to the specific consumer characteristic under investigation. These relationships provide
206
the first credible evidence of a moderating effect of these consumer characteristics and
product cue usage. However, the lack of significant differences between utility values
between respective groups means we must be cautious in interpreting these
correlations, given the generally weak relationships they represented.
Overall, the conjoint analysis findings and correlation testing results for the survey are
comparable to those found in the analysis of the sensory data in terms of the attitudes
towards intrinsic and extrinsic cues. One important difference in opinions identified
between the sensory experiment and survey results is those completing the survey
generally associated higher cream levels with better quality brie. Conversely, in the
sensory experiment, only those with high levels of category and self-assessed
knowledge favored the triple cream level, with a number of respondents preferring the
double or full cream variety. This highlights the critical nature of the terms used in
product descriptions, as the reality for many consumers may not be as expected. Also,
the survey findings provide the first empirical evidence linking knowledge to cue usage,
although interpretations must be tempered, with the acknowledgement that the
influence is quite weak and appears isolated to lower knowledge segments. In
summary, the findings show that for brie the influence of the intrinsic cue is more
powerful than the extrinsic cues investigated, although price and COO were not totally
discounted in forming respondent opinions. Generally, knowledge and self-confidence
were not found significantly influential in forming respondent opinions of quality. In the
case of the sensory experiments, taste overwhelmed expectations and underpinned
quality ratings while anticipation of taste (participants expecting the creamier the
better) drove quality ratings in the survey. Table 9.18 provides a summary of results in
terms of hypotheses tested.
207
Table 9.18 Hypotheses testing outcomes (brie)
Objective knowledge is positively associated with the relative

H1 Partially
contribution of intrinsic cues to an individuals assessment of product Supported
quality.
Objective knowledge is positively associated with the relative contribution of an Partially

H1a
intrinsic cue (fat) to an individuals assessment of product quality (cheese). Supported
Objective knowledge is negatively associated with the relative

Not
quality.
H2a

H2b
Subjective knowledge is positively associated with the relative

Partially
H3 contribution of intrinsic cues to an individuals assessment of product Supported
quality.
Subjective knowledge is positively associated with the relative contribution of an Partially
H3a
intrinsic cue (fat) to an individuals assessment of product quality (cheese). Supported
Subjective knowledge is negatively associated with the relative

Not
quality.
H4a

H4b
Self-confidence is positively associated with the relative contribution Not

H5 Supported
of intrinsic cues to an individuals assessment of quality.
Self-confidence is positively associated with the relative contribution of an intrinsic Not
H5a
cue (fat) to an individuals assessment of product quality (cheese). Supported
Self-confidence is negatively associated with the relative contribution Not

H6 Supported
of extrinsic cues to an individuals assessment of quality.

H6a
cue (COO) to an individuals assessment of product quality (wine / cheese). Supported

H6b
cue (price) to an individuals assessment of product quality (wine / cheese). Supported
The paper based conjoint analysis method will be predictive of the sensory
Partially
H7 based test in terms of the relative average importance of tested product
Supported
attributes (intrinsic and extrinsic)
208
10 C
onclusions, limitations and implications
10.1 Introduction
This final chapter summarizes the thesis. A review of research is provided in addition to
a summary of key findings and their implications for marketing managers. Limitations to
the research are discussed and suggestions for future research are put forward.
10.2 Overview of the research and findings
Chapter 2 provides a prcis of the substantial body of literature examining the influence
of intrinsic and extrinsic cues (COO and price specifically) on consumer evaluation of
expected and experienced product quality. The likely moderating effect of the
consumer characteristics of knowledge and self-confidence are also explored.
However, in reviewing the existing research in these areas, several gaps in our current
understanding of the interactions between these variables were identified. Moreover,
this research offered an opportunity to test the ability of conjoint analysis as a credible
tool to predict actual consumer opinions regarding product attribute preferences.
Therefore, this research was undertaken to quantify the power of selected extrinsic and
intrinsic cues of consumer evaluations of both experienced and expected product
quality, and further, to measure the respective influences of objective knowledge,
subjective knowledge and self-confidence on these quality assessments. Another
intention was to determine if a survey conducted measuring expectations of quality
would yield comparable results with quality assessments based on actual product
experience. The investigation involved a series of sensory experiments where
respondents tasted samples of wine and cheese and was followed by a second study
administered by survey, where only product description profiles were provided. The
209
analysis of results from both the sensory experiment and the conjoint analysis survey
show findings that are remarkably similar in most respects.
For chardonnay, price was consistently found to exert the most powerful influence on
consumer quality assessments with an average attribute importance of 71.81% for the
sensory experiment, and 73.68% for the survey (Table 8.5). Surprisingly, those
participating in the sensory experiments thought COO was more persuasive than
experienced acid levels with an average importance of 15.08% for COO and 13.10% for
acid. This is in spite of experiencing wines of considerably diminished quality,
described as sour, like vinegar and foul by volunteers in earlier aspects of the study.
In the survey, respondents considered COO to be marginally less important than acid
with respective attribute average importance scores of 12.27% for COO and 14.05% for
acid. Scrutiny of average profile utility values across both stages of the research
confirms the overall parity of opinion between the two groups. The rank order of wine
profiles from the highest utility value profile to the lowest were identical in both stages,
with the exception of a reverse order for the profiles ranking 8th and 9th respectively, with
profiles also sharing similar average scores (Table 8.6). COO and price did not affect
objective quality; however, even when experienced quality was diminished, their
influence was not substantially reduced.
The influence of knowledge (objective and subjective) and self-confidence was found to
be sporadic and weak. In the case of knowledge, this is likely to be due to respondents
general lack of objective knowledge in both stages of the survey. In the case of self-
confidence, results are surprising given that respondents in both studies exhibited
reasonably healthy degrees of self-confidence. Whilst significant relationships emerged
between these consumer characteristics and attribute utility levels, no interpretable
pattern emerged and correlation coefficients were generally weak. Further, testing
210
revealed that few significant differences existed between average utility values and the
high and low level knowledge and self-confidence clusters.
For brie, the results were similar but with one noteworthy difference in opinion between
the respondents participating in the sensory experiment and the survey. Whilst price
was again the most important attribute influencing quality (experienced and expected)
with an average attribute importance of 43.99% for the sensory results and 40.60% for
the survey, fat levels were found to be more important than COO for both groups with
an average attribute importance of 30.95% in the experiment and 35.07% in the survey
(Table 8.22). The average importance of COO for both groups was consistent at
25.07% in the sensory experiment, and 24.33% in the survey. Utility values for the
various levels revealed a wider range of opinion generated from those assessing
product descriptions only. For example, while both samples believed that Argentina
would provide the worst quality brie and France the best, the utility value for Argentina
was -0.1841 (sensory) and -0.2320 (survey), and 0.1633 (sensory) and 0.3520 (survey)
for France. Whilst both groups considered fat a major contributor to quality, many in the
sensory experiment found the triple cream unpleasant and rated those profiles
accordingly, whereas (working on expectations alone) those in the survey strongly
favored this fat level. Opinions on the double cream were very similar, with differences
noted also on judgments of the full cream variety. Those who tasted the full cream brie
found the cheese marginally less appealing, whereas those completing the survey
punished this level with harsh ratings. The strong halo over the triple cream brie may
be to some extent due to pleasant associations with the term cream for many
consumers. For them, if double cream is expected to be good, then triple cream must
be even better. Analysis of results for this product provides the first indications of the
influence of knowledge, both subjective and objective, on cue usage. Hence, those
respondents with higher levels of knowledge considered fat to be considerably more
important that those with lower levels, with correspondingly diminished importance for
211
price. For example, the average importance of fat for high objective knowledge group
(survey) is 49.26% as opposed to 5.65% for the low level cluster. Correspondingly,
those with high subjective knowledge (survey) determined its average importance as
51.58% compared to 15.72% for the low level segment. These findings are supported
by the correlations matrix illustrating attribute average importance for fat and price and
knowledge (both types) for the survey data where significant relationships were found to
exist between these variables. However, these outcomes must be tempered by
examination of the correlation matrixes between groups for knowledge (both types) and
self-confidence and attribute utility values where, as with the sensory data, all significant
correlations revealed signified weak relationships. Moreover, no reliable and
interpretable pattern was found between the variables tested and few differences in
average utility values were significant.
Overall, the results of the data analysis for brie, for both stages of the study, show
consistent opinions in regard to the relative average importance for the attributes tested.
Price was found for both groups to be the most influential, but fat levels and COO were
both also major contributors to the quality evaluation. The major noteworthy difference
in the findings from the sensory experiments and survey is the opposite response to the
experienced assessment of triple cream brie versus the expected liking of respondents
for this product. For brie, conjoint analysis has fairly accurately identified the respective
contribution of the product attributes tested, but the importance of meaningful product
descriptions must be highlighted. Whilst both sample groups were consistent in their
determination of fat average importance (30.95% for the sensory group and 35.07% for
the survey sample), there was considerable divergence in their opinions regarding the
most desirable levels. The survey respondents were attracted to the brie with the
highest cream content, expecting the taste to be very good. However, in reality,
consuming brie containing that much fat actually results in a very negative the sensory
experience for many. These consistent findings across 2 different product categories
212
indicates that conjoint analysis by survey is a valid and appropriate method for
measuring expected consumer liking for hypothetical product attributes, with the proviso
that the product descriptors used must be accurately interpreted by respondents both
pre and post-product experience.
In summary, for chardonnay only H7 is supported, as there is no robust empirical
evidence to support that high levels of knowledge or self-confidence lead to a heavier
reliance on intrinsic rather than extrinsic cues. Rather, significant correlations between
the independent and dependent variables tested tend to illustrate that heightened
knowledge and self-confidence may result in strong support for stereotypical country
images and the need to pay more to acquire better quality wine. For brie, credible
evidence was found supporting a link between knowledge levels and consideration of
the intrinsic cue and its varying levels. Thus demonstrating that for this product
category, experience and an objective understanding of a critical product attribute can
result in increased consideration of that attribute when assessing quality, both expected
and experienced. Therefore, for brie H1, H1a, H3, H3a and H7 are found to be partially
supported. However, these results are tempered by the lack of support for the
remaining hypotheses, demonstrating the power of extrinsic cues to overcome the
credence placed on intrinsic cues, knowledge levels notwithstanding.
10.3 Contribution to the literature
Previous research has shown that consumers vary in their reliance on both intrinsic and
extrinsic cues as well as in their ability to accurately assess product cues accurately
(Alba 2000; Kardes, Kim et al. 2001). Two extrinsic cues found by researchers to be
used consistently in this process are COO and price (Zeithaml 1988; Dodds 1991; Chao
and Rajendran 1993; Al-Sulaiti and Baker 1998). However, the degree to which COO
and price have the power to override sensory perceptions of quality has not previously
213
been tested to this extent. Moreover, while consumer knowledge and self-confidence
have been found previously to moderate consumers reliance on extrinsic cues, results
are ambiguous often as a result of inconsistencies in definitions and measurement
(Alba and Hutchinson 1987; Alba and Hutchinson 2000). While empirical evidence
exists in relation to various aspects of these specific variables, several gaps remain in
the literature in relation to their combined effects on product quality determination.
Further, many studies have investigated the influence of extrinsic cues using
experimental designs, where respondents experience various product offerings, or
quasi-experiments through conjoint analysis, where respondents rate products or
choose their preferred product option from a number of described product profiles. The
influence of extrinsic cues, however, has never been tested using different types of
methodology in the same study. Firstly, in a confirmatory sense using experienced
quality as evaluated through sensory perceptions and, secondly, using expected quality
as determined through product description. This research has examined and quantified
differences in both outcomes and explored the ability of conjoint analysis to predict
actual quality evaluations. Further, the research has also provided invaluable incite
into the consumer knowledge construct, both in terms of quantification and potential
impact on quality assessment (pre and post trial). The study has also provided a
validated scale for the measurement of consumer personal self confidence, as distinct
from purchasing confidence, and quantified its influence on product cue usage. This
information provides marketing managers with an invaluable tool for developing
effective marketing strategies rather than risking allocations of valuable resources
promoting product characteristics that consumers do not understand or value.
10.3.1 Consumer reliance on extrinsic cues: COO and
price
Consumers have been found more reliant on extrinsic cues such as price and COO
when there is little other specific and reliable information available for consumers to
214
consider, or when consumers are evaluating high cost/high involvement products
(Chao 1989 b; Han 1990; Chao 1992; Chao and Rajendran 1993; Piron 2000). In the
case of the sensory experiments, reliance on the extrinsic cues tested was found to
remain extremely robust even when all intrinsic cues (through sensory experience)
were available for respondent evaluation. In the case of the conjoint survey, results
which confirmed expectations of quality were also highly reliant on the extrinsic cues
communicated. Whilst the influence of extrinsic cues was expected to be stronger in
the case of wine as compared to cheese (Quester and Smart 1998; Piron 2000), in both
instances price was considered more important than the intrinsic cue described and
experienced, demonstrating that even when evaluating a relatively low involvement
product, consumer belief in the price value schema dominates quality assessment.
This research has significantly advanced the understanding of the consumer use of
extrinsic cues (price and COO specifically), and their respective influence in their
determination of both expected and experienced quality.
10.3.2 Knowledge and self-confidence
This study goes further than the existing literature by clearly delineating between
objective knowledge, subjective knowledge and self-confidence and quantifying specific
levels for each. The study also provides a template for testing objective knowledge by
developing, employing and validating the measure used (pilot, sensory, survey). While
subjective knowledge and objective knowledge are linked, they are 2 distinct constructs
reflecting differing understanding of product characteristics; therefore, future studies
measuring consumer knowledge should specify which characteristic is being
investigated and ensure that an appropriate instrument is used. The scale used for
measuring self-confidence was determined through the testing and validation using a
set of 10 statements originally developed by Day and Hamblin (1964) and used by Bell
(1967). Whilst used in earlier research, the items had not been previously reduced and
215
validated as a measurement scale; hence, this research has provided a legitimate tool
for future researchers wishing to measure this characteristic.
Empirical evidence has established that consumers, in the main, do not possess the
level or quality of objective knowledge they believe they do (Alba and Hutchinson 1987;
Heimbach, Johansson et al. 1989; Alba and Hutchinson 2000; Alba 2000). Given this, it
is not surprising that many consumers often misjudge product quality through erroneous
interpretation of both intrinsic and extrinsic cues. Therefore, whilst there was an
anticipation of low consumer knowledge in this study, actual levels were still well below
expectations. This suggests that for these products also, the majority of consumers are
basing decisions largely on preconceived ideas regarding price and the moderate
influence of COO. Given that the tests and scales were completed by over 800
respondents in total during the course of study, the results are beyond doubt. Given
this, it is not surprising that the influence of knowledge on consumer quality
expectations was found to be limited or nonexistent. Previous studies have found that
experts use intrinsic cues more accurately and consistently than those with low levels
of knowledge (Brucks 1985; Park, Mothersbaugh et al. 1994; Mason and Bequette
1998; Kardes, Kim et al. 2001; Wirtz and Mattila 2003). However, testing of
respondents in this study suggest that true product experts are few and, for the
products tested, still very likely to give strong credence to extrinsic cues over intrinsic
ones even when the intrinsic cues are experienced. These finding add significantly to
the existing body of knowledge in relation to the implications of consumer knowledge on
product quality expectations and assessment.
Previous studies have found that levels of self-confidence can have an important
influence on product cue usage (Wilson and Brekke 1994; Wansink, Park et al. 2000;
Bearden, Hardesty et al. 2001; Jover, Montes et al. 2004). However, in this study little
was found to support these previous findings. Unlike determined knowledge levels, in
216
general, respondents in all stages of the research possessed quite healthy levels of
self-confidence. However, even those in the extreme clusters of this attribute showed
no significant differences in testing suggesting the influence may somewhat
overestimated in previous studies or perhaps linked to other personality traits.
Therefore, the findings of this study show firstly, that Australians are found (by the
measure used) to be generally self-confident in their approach, and secondly, levels
have little effect on cue usage for the products tested.
10.4 Implications for marketers
It is critical that marketing managers understand the respective influence of extrinsic
cues in consumers quality assessment processes to ensure that marketing efforts are
focused towards enhancing those attributes most likely to influence consumers
opinions regarding quality. Given that there are relatively few true experts in most
consumer markets, the credence given to extrinsic cues by these consumers cannot be
underestimated.
Consumers have been found to consistently rely on the extrinsic cues provided as
surrogate indicators of quality; in the case of the respondents in the sensory experiment
for chardonnay, the influence of price was found to even overwhelm the taste of poor
wine. These results were determined even though the range of acidity in the wines
tasted was considerably wider than the likely available range in the market. Hence, the
conclusions are irrefutable, given that even with such extreme acidity taste was not
nearly as influential as price and similar in influence to COO. Therefore, in the market
place where the objective quality between products is often comparable, the influence
of extrinsic cues may be critical to consumer quality assessment. Further, testing of
respondents for knowledge and self-confidence confirmed that objective knowledge is
generally low indicating a very limited understanding of intrinsic product attributes and a
217
likely misinterpretation of those considered. This means that marketers must not
assume that knowledge is present at high levels and must ensure that marketing
communications (including packaging and labeling) is designed in such a way as to
facilitate clear understanding of the implications of any intrinsic cues communicated.
For example, in the case of brie the use of cream in the product descriptions is
accurate and describes a critical intrinsic attribute to the taste experience. However,
most respondents reading the descriptions expected that as the level of cream
increased, their liking would increase correspondingly. However, many of those
experiencing the triple cream brie found this level of fat distasteful. Only those
participants with experience and knowledge (and therefore more sophisticated palates)
enjoyed the extremely rich taste and mouth feel of this brie. Conversely, those
evaluating the chardonnay essentially discounted the impact of the acid, in both the
experience and the expectation from the descriptions. Testing revealed that very few
respondents understood the implications of acid on taste and so were not able to
understand the implications of high acid on quality, or believed that a high price wine
(particularly from France) would be good in spite of any stipulated acid amounts. Those
actually tasting the wine were often overcome by their beliefs in price over taste,
perhaps believing their own palates to be at fault. Given that wine labels provide acid
levels in grams per bottle, rather than a prescribed description, accurate understanding
of the repercussions of this important intrinsic cue by the majority of consumers is even
less likely.
These outcomes lead to threats and opportunities alike. Marketers working with a base
level product in the wine or cheese category, for example, could leverage up the relative
position of their product in the minds of consumers through the use of a prestige pricing
strategy supported by other congruent extrinsic cues such as COO (if possible),
selective distribution and quality oriented packaging and communications. Further,
aside from the legal requirements specific to the product in relation to labeling and
218
packaging, intrinsic cues that are not conducive to the product position desired could be
played down or not disclosed. Given that consumers may possess a low level of
objective knowledge regarding the product category and are likely to give credence to
extrinsic cues provided, desired brand values and product quality expectations could be
established based purely on essentially non-intrinsic product characteristics.
Conversely, for those wishing to differentiate on intrinsic product attributes unforeseen
challenges may await. Consumers may not be able to appreciate the intrinsic
difference offered and may not value it in any case. Further, descriptions or
specifications may not be precisely understood. For marketers in this situation,
consumer education (based on research to gain an accurate understanding of
consumer knowledge levels and perceptions) through marketing communications is
likely to be needed. This approach will still require the support of congruent extrinsic
cues to underpin the strategy, as even experts have been found to rely on extrinsic
cues to make decisions. For example, it would be extremely difficult to convince
consumers that high quality can be obtained at a bargain price. Results of self-
confidence testing in the research show that, in the main, Australians enjoy a relatively
high level of self-confidence. This means they are likely to be somewhat stubborn in
their beliefs regarding product attributes irrespective of accuracy. Therefore, marketers
may find re-educating consumers in this situation challenging.
Whilst the effect of consumer belief in the price value schema is quite straightforward,
the implications of COO are less clear cut. The findings of the pilot study revealed an
example of ethnocentric backlash against foreign wines and cheeses and serve as a
reminder that consumers may be sensitive to some foreign sourced products even
when the product and the country are congruent (e.g. wine from France). Whilst overall
products tested were more highly favored when associated with a more attractive CI,
interestingly, the similar utility values for the U.S. and Chile in the chardonnay testing
show that views do change over time and as a result of marketing efforts. It was
219
expected that respondents would strongly favour the U.S. over Chile; however, results
show that the value placed on both countries was variable highlighting the need to
investigate consumer perceptions of source countries rather than assume a positive or
negative COO effect.
The research has also provided evidence that the conjoint analysis by survey is a
credible means of identifying attributes that are most highly valued and determining
those levels deemed most and least attractive. This assertion is made with the proviso
that thorough research is undertaken prior to deciding on which attributes to investigate
and that realistic levels for each attribute are set. By using this type of methodology,
marketers will possess powerful information for use in developing a meaningful and
effective marketing strategy, prior to the introduction of a new product, or making
significant changes to an existing marketing mix.
10.5 Limitations
A number of limitations are evident in this study. First, the use of convenience samples
can limit the ability to generalize results. However, care was taken to source
respondents exclusively from evening classes across a number of faculties and from
various sources in the general community in order to derive the closest possible
matches to the relevant demographics of the Australian population. While some minor
biases are reported, overall these are not expected to severely limit results. While total
sample sizes of 263 (sensory) and 274 (survey) were achieved, clusters used for testing
those with extreme levels of knowledge and self-confidence were relatively small. A
sample size in excess of 30 is considered statistically acceptable for measuring
correlations, however, a small sample size does contribute to error due to non-
representation of the total population (Malhotra, Hall et al. 2002). Statistical results
must therefore be considered in light of this limitation.
220
Scales employed to measure the subjective knowledge and self-confidence resulted in
acceptable total variances explained for each variable, however, the percentages of
variance left unexplained by each scale used must be considered a limitation to the
results (Malhotra, Hall et al. 2002).
There are limitations inherent in the use of conjoint analysis methodology. In reality,
products and services are comprised of a combination of hundreds (perhaps
thousands) of intrinsic and extrinsic cues and the methodology only allows the
researcher to test a few. The generalizability is therefore dependent upon choosing
those attributes and levels that are most critical to the quality evaluation and/or the
buying decision (Jaeger, Hedderley et al. 2000). Whilst careful scrutiny of the existing
literature, plus analysis of the data derived from the focus groups and the results of the
pilot study were all employed to ensure that appropriate choices were made to minimize
these limitations, they still must be considered a risk. Further, the use of rating scales is
a possible limitation due to consumer fatigue and disinterest, seen where each profile is
rated consistently low, high or average resulting in little discrimination between the
profiles (Cohen 2003).
The use of a laboratory environment can enhance accuracy and also result in
limitations. Respondents in the lab are more likely to concentrate fully on the task at
hand leading to a heightened examination of products tested, beyond what would be
provided by them in real life. Further, the provision of a large number of samples may
lead to fatigue, resulting disinterest or diminished sensitivity. Steps discussed in the
methodology sections of the thesis were taken to minimize these potential sources of
error, however, they must be acknowledged (Schifferstein 1996).
221
An important objective of the research was to determine the degree to which knowledge
moderates consumer use of intrinsic and extrinsic cues. However, test results clearly
show that objective knowledge in relation to the 2 products tested was consistently low.
Therefore, even the high knowledge segments could not be considered true product
experts. This means that the impact of high levels of objective knowledge on product
cues has been limited; however, it also shows that the majority of consumers for these
products are unlikely to be knowledgeable and results are an accurate reflection of
consumer ability in this regard.
10.6 Directions for future research
The research provides a platform for a number of subsequent studies. The limitation of
low consumer knowledge discussed previously suggests that the results of the study
could be extended and enhanced by collecting data exclusively from sourced and
qualified product experts, thus furthering the body of literature in relation to the
moderating effect of objective knowledge. Following from this, the development of a
measure of consumer expertise is required. This research, and others, demonstrates
that there is a link between objective and subjective knowledge, in which an individual
possessing high levels of both, is deemed to be an expert (Alba and Hutchinson 1987;
Alba and Hutchinson 2000; Alba 2000). What is required is a validated measure of
expertise incorporating both aspects of the construct, and from this, further research
can be undertaken in relation to cue usage.
The study also revealed the potential for consumers to misunderstand the implications
of commonly used descriptive terms, confirming the risk to marketers developing
communications and labeling strategies. Further studies investigating the expectations
consumers develop as a result of their interpretations of terms regularly used in
222
conjunction with wine and/or food products would provide managers with valuable
insight into consumer understanding of communicated messages.
The use of rating scales in conjoint analysis has identified limitations that have been,
very recently, addressed in research through the use of maximum difference rating
methodology known as Best versus Worst (Cohen 2003). Using this approach,
consumers are asked to consider a number of selected attributes/levels and instead of
rating the overall profile, they are simply asked to identify the best attribute/level and
the worst attribute/level. This eliminates the risk of consumers rating each profile
similarly irrespective of its components, as they must evaluate each element provided in
the individual profile and identify the most and least attractive element. In addition to
eliminating that potential limitation from the conjoint analysis method, analysis of the
elements most often considered the best and worst and those ranked after is a
comparatively simple computation. There is the potential to test this application in a
sensory experiment, thus providing the opportunity to extend the development of this
technique.
10.7 Summary
This chapter has provided a review of the findings of the research, the respective
contributions to the literature and marketing professionals. Limitations and suggestions
for future research close the thesis. In summary, the findings demonstrate the
importance to consumers of cues that are irrelevant to objective product quality,
reminding marketers that consumers buy more than just the actual product, they
purchase a bundle of attributes, the value of which is determined largely by their
perceptions, experiences and preconceived beliefs. Further, while consumers may be
familiar with products and product categories as a result of marketing efforts and
223
purchases, this cannot be confused with objective product knowledge and accurate
understanding of product attributes, be they intrinsic or extrinsic.
224
11 R
eferences
Aaron, J. I., D. J. Mela, et al. (1994). "The Influences of Attitudes, Beliefs and Label
Information on Perceptions of Reduced-fat Spread." Appetite 22(1): 25 - 37.
Acebron, L. B. and D. C. Dopico (2000). "The importance of intrinsic and extrinsic cues
to expected and experienced quality: and empirical application for beef." Food and
Quality Preference 11: 229 - 238.
Agarwal, S. and S. Sikri (1996). "Country Image: consumer evaluation of product
category extensions." International Marketing Review 13(4): 23 - 39.
Agarwal, S., R. K. Teas, et al. (1997). "Entity context and format induced instability in
multiattribute ratings of country image." International Marketing Review 14(6): 486
- 504.
Agrawal, J. and W. A. Kamakura (1999). "Country of origin: a competitive advantage?"
International Journal of Research in Marketing 16: 255 - 267.
Ahmed, S. A. and A. d'Astous (2004). "Perceptions of countries as producers of
consumer goods
A T-shirt study in China." Journal of Fashion Marketing and Management 8(2): 187 -
200.
Ahmed, Z. U., J. P. Johnson, et al. (2002). "Country-of-origin and brand effects on
consumers' evaluations of cruise lines." International Marketing Review 19(3): 279
- 302.
Ahmed, Z. U., J. P. Johnson, et al. (2004). "Does country of origin matter for low-
involvement products?" International Marketing Review 21(1): 102 - 120.
Alba, J. and J. W. Hutchinson (2000). "Knowledge calibration: what consumers know
and what they think they know." Journal of Consumer Research 27(2): 123 - 156.
Alba, J. W. (2000). "Dimensions of Consumer Expertise ... or lack thereof." Advances in
Consumer Research 27: 1 - 9.
Alba, J. W. and J. W. Hutchinson (1987). "Dimensions of Consumer Expertise." Journal
of Consumer Research 13(March): 411 - 454.
Al-Sulaiti, K. I. and M. J. Baker (1998). "Country of origin effets: a literature review."
Marketing Intelligence & Planning 16(3): 150 - 199.
Amine, L. S. and S. H. Shin (2002). "A comparison of Consumer Nationality as a
Determinant of COO Preferences." Multinational Business Review(Spring): 45 -
53.
Andersen, P. H. and P. Chao (2003). "Country-of-origin effects in global industrial
sourcing: toward an integrated framework." Management International Review
43(4): 339 - 361.
Andreassen, T. W. and B. Lindestad (1998). "Customer loyalty and complex services."
International Journal of Service Industry Management 9(1): 7 - 23.
225
Australian-Bureau-of-Statistics (2003 - 04). Household Characteristics (Report No.
65230), Australian Government.
Badri, M. A., D. L. Davis, et al. (1995). "Decision support for global marketing strategies:
the effect of country of origin on product evaluation." Journal of Product & Brand
Management 4(5): 49 - 64.
Balabanis, G. and A. Diamantopoulos (2004). "Domestic Country Bias, Country-of-
Origin Effects, and Consumer Ethnocentrism: A Multidensional Unfolding
Approach." Journal of the Academy of Marketing Science 32(1): 80 - 96.
Balabanis, G., R. Mueller, et al. (2002). "The human values' lenses of country of origin
images." International Marketing Review 19(6): 582 - 610.
Baldy, M. W. (1993). The University Wine Course, A wine Appreciation Text & Self
Tutorial. San Francisco, The Wine Appreciation Guild.
Bandyopadhyay, S. (2001). "Competitiveness of Foreign Products as Perceived by
Consumers in the Emerging Indian Market." 11(1): 53 - 64.
Bandyopadhyay, S. and B. Banjeree (2002). "A country of origin analysis of foreign
products by Indian Consumers." Journal of International Consumer Marketing
15(2): 85 - 109.
Bargh, J. A. and T. L. Chartrand (1999). "The Unbearable Autonomy of Being."
American Psychologist 54(July): 462 - 479.
Bearden, W. O., D. M. Hardesty, et al. (2001). "Consumer Self-Confidence:
Refinements in Conceptualization and Measurement." Journal of Consumer
Research 28(June): 121 - 134.
Beaudoin, P., M. A. Moore, et al. (1998). "Young fashion leaders' and follower's
attitudes toward American and imported apparel." Journal of Product & Brand
Management 7(3): 193 - 207.
Beaverland, M. and A. Lindgreen (2002). "Using Country of Origin in Strategy: the
Importance of Context and Strategic Action." Brand Management 10(2): 147 -
167.
Becker, T. (2000a). "Consumer perception of fresh meat quality: a framework for
analysis." British Food Journal 102(3): 158 - 176.
Becker, T., E. Benner, et al. (2000b). "Consumer perception of fresh meat quality in
Germany." British Food Journal 102(3): 246 - 266.
Bell, G. D. (1967). "Self-Confidence and Persuasion in Car Buying." Journal of
Marketing Research IV(February): 46 - 52.
Berg, B. v. d., M. Al, et al. (2005). "Economic valuation of informal care: The conjoint
measurement method applied to informal care-giving." Social Science & Medicine
61: 1342 - 1355.
Bernues, A., A. Olaizola, et al. (2003). "Extrinsic attributes of red meat as indicators of
quality in Europe: an application for market segmentation." Food and Quality
Preference 14: 265 - 276.
Bettman, J. R. and C. W. Park (1980). "Effects of Prior Knowledge and Experience and
Phase of the Choice Process on Consumer Decision Processes." Journal of
Consumer Research(7 (December)): 234 - 248.
226
Bhuian, S. N. (1997). "Marketing cues and perceived quality: Perceptions of Saudi
consumers towards products of the U.S., Japan, Germany, Italy, U.K. and
France." Journal of Quality Management 2(2): 217 - 235.
Bilkey, W. J. and E. Nes (1982). "Country-of-origin Effects on Product Evaluations."
Journal of International Business Studies(Spring/Summer): 89 - 99.
Bitta, A. J. D. and K. B. Monroe (1982). "A Multivariate Analysis of the Perception of
Value from Retail Price Advertisements." Advances in Consumer Research 8(1):
161 - 165.
Blackwell, L. (1995). "Visual cues and their effects on odour assessment." Nutrition &
Food Science 5(September / October): 24 - 28.
Bourke, A. (2000). "A Model of the Determinants of International Trade in Higher
Education." The Service Industries Journal 20(1): 110 - 138.
Bredahl, L. (2003). "Cue utilisation and quality perception with regard to branded beef."
Food and Quality Preference 15: 65 - 75.
Brucks, M. (1985). "The effects of product class knowledge on information search
behaviour." Journal of Consumer Research 12: 1 - 16.
Bruning, E. R. (1997). "Country of origin, national loyalty and product choice: the case
of international air travel." International Marketing Review 14(1): 59 - 74.
Buchanan, B., M. Givon, et al. (1987). "Measurement of Discrimination Ability in Taste
Tests: An Empirical Investigation." Journal of Marketing Research 24(May): 154 -
163.
Buttery, E. A. and T. K. P. Leung (1998). "The difference between Chinese and
Western negotiations." European Journal of Marketing 32: 374 - 389.
Cateora, G. P. (2002). International Marketing 11th Ed, McGraw-Hill Higher Education.
Chao, P. (1989 a). "Expert and reverse investment: Strategic implications for newly
industrialized countries." Journal of International Business Studies 20(1): 75 - 91.
Chao, P. (1989 b). "The impact of country affiliation of the credibility of product attribute
claims." Journal of Advertising Research(April - May): 35 - 41.
Chao, P. (1993). "Partitioning country of origin effects: Consumer evalations of a hybrid
product." Journal of International Business Studies(September & November): 291
- 306.
Chao, P. (2001). "The Moderating Effects of Country of Assembly, Country of Parts,
and Country of Design on Hybrid Product Evaluations." Journal of Advertising
XXX(4): 67 - 81.
Chao, P. (2005). "Celebrity and foreign brand name as moderators of country-of-origin
effects." International Journal of Advertising 24(2): 173 - 192.
Chao, P. and P. B. Gupta (1995). "Information search and efficiency of consumer
choices of new cars." International Marketing Review 12(6): 47 - 59.
Chao, P. and K. N. Rajendran (1993). "Consumer Profiles and Perceptions: Country-of-
origin Effects." International Marketing Review 10(2).
227
Chen, H.-C. and A. Pereira (1999). "Product entry in international markets: the effect of
country-of-origin on first-mover advantage." Journal of Product & Brand
Management 8(3): 218 - 231.
Chiou, J.-s. (2003). "The Impact of Country of Origin on Pretrial and Post trial Product
Evaluations: The Moderating Effect of Consumer Expertise." Psychology &
Marketing 20(10): 935 - 954.
Choi, C. J. (1992). "Marketing Barriers Facing Developing Country Manufactured
Exporters: A Comment." The Journal of Development Studies 29(Oct): 166 - 171.
Chung, J.-E. and D. T. Pysarchik (2000). "A model of behavioural intention to buy
domestic versus imported products in a Confucian culture." Marketing Intelligence
& Planning 18(5): 281 - 291.
Cignetti, F. F. and M. Caldara (1997). "Comparing tools for service quality evaluation."
International Journal of Quality Science 3(4): 356 - 367.
Clarke, I., M. Owens, et al. (2000). "Integrating country of origin into global marketing
strategy, a review of UW Marketing statutes." International Marketing Review
17(2): 114 - 126.
Coakes, S. J., L. Steed, et al. (2006). SPSS version 13.0 for Windows, Analysis without
Anguish. Milton Qld, John Wiley & Sons Australia Ltd.
Cohen, S. H. (2003). Maximum Difference Scaling: Improved Measures of Importance
and Preference for Segmentation. Research Paper Series. S. Software. Sequim
Washington.
Conover, J. N. (1986). "The Accuracy of Price Knowledge: Issues in Research
Methodology." Advances in Consumer Research 13: 589 - 593.
Cordell, V. (1992). "Effects of consumer preferences of foreign sourced products."
Journal of International Business Studies 23(2): 251 - 269.
Cowley, E. J. (1994). "Recovering Forgotten Information: A Study in Consumer
Expertise." Advances in Consumer Research 21: 58 - 63.
Crawford, J. C. and C. W. Lamb (1981). "Source preferences for imported products."
Journal of Purchasing and Material Management 17: 28 - 33.
Cronin, J. J. and A. T. Steven (1994). "SERVPERF Versus SERVQUAL: Reconciling
Performance-Based and Perceptions-Minus-Expectations Measurement of
Service Quality." Journal of Marketing 58: 125 - 131.
Curry, J. (1996). "Understanding Conjoint Analysis is 15 Minutes. Research Paper
Series." Sawtooth Software.
Day, R. C. and R. L. Hamblin (1964). "Some Effects of Close and Punitive Styles of
Supervision." The American Journal of Psychology 69(March): 599 - 511.
Dean, D. H. (2004). "Evaluating potential brand associations through conjoint analysis
and market simulation." Journal of Product & Brand Management 13(7): 506 -
513.
Delong, M., M. Bao, et al. (2004). "Perception of US Branded apparel in Shanghai."
Journal of Fashion Marketing and Management 8(2): 141 - 153.
228
Devlin, J. F. (2002). "Customer knowledge and choice criteria in retail banking." Journal
of Strategic Marketing 10: 273 - 290.
Dickson, P. R. and A. G. Sawyer (1990). "The Price Knowledge and Search of
Supermarket Shoppers." Journal of Marketing 54(July): 42 - 53.
Dodds, W. (1991). "In Search of Value: How price and store name information influence
buyers' product perceptions." The Journal of Consumer Marketing 8(2): 15 - 24.
Ede, F. O. and B. Panigrahi (2000). "African-American consumer attitudes toward
domestic and foreign-made automobiles." Management Research News 23(5): 1 -
19.
Erickson, G. M. and J. K. Johansson (1985). "The Role of Price in Multi-Attribute
Product Evaluations." Journal of Consumer Research 12(September).
Eriksson, K. and A. Hadjikhani (2000). "Perceptual product connection in an
international context." International Business Review 9: 301 - 320.
Eroglu, S. A. and K. A. Machleit (1988). "Effects of Individual and Product-specific
Variables on Utilizing Country of Origin as a Product Quality Cue." International
Marketing Review 6(6): 27 - 41.
Ettensen, R. (1993). "Brand Name and Country of Origin Effects in the Emerging
Market Economies of Russia, Poland and Hungary." International Marketing
Review 10(5).
Fazio, R. H. and M. P. Zanna (1978). "On the predictive validity of attitudes: The roles
of direct experience and confidence." Journal of Personality 46(2): 228 - 243.
Finn, D. W. and C. W. Lamb (1991). "An Evaluation of the SERVQUAL Scales in a
Retailing Setting." Advances in Consumer Research 18: 483 - 490.
Flynn, L. R. and R. E. Goldsmith (1999). "A Short, Reliable Measure of Subjective
Knowledge." Journal of Business Research 46: 57 -66.
Ford, J. B., M. Joseph, et al. (1999). "Importance-performance analysis as a strategic
tool for service marketing: the case of service quality perceptions of business
students in New Zealand and the USA." The Journal of Services Marketing 13(2):
171 - 186.
Franceschini, F. and M. Cignetti (1998). "Comparing tools for service quality
evaluation." International Journal of Quality Science 3(4): 256 - 167.
Gatchalian, M. M. (1999). "Quality assessment through statistically based sensory
evaluation methods." The TQM Magazine 11(6): 389 - 395.
Geel, L., M. Kinnear, et al. (2004). "Relating consumer preferences to sensory attributes
of instant coffee." Food and Quality Preference In Press: 1 - 8.
Ger, G. (1999). "Experiential Nature of Product-Place Images: Image as a Narrative."
Advances in Consumer Research 26: 165 - 169.
Glitsch, K. (2000). "Consumer perceptions of fresh meat quality: cross-national
comparison." British Food Journal 102(3): 177 - 194.
Gluckman, R. L. (2001). "A Consumer Approach to Branded Wines." European Journal
of Marketing 20(6): 21 - 35.
229
Goldsmith, R. E., F. d'Hauteville, et al. (1997). "Theory and measurement of consumer
innovativeness." European Journal of Marketing 32(3/4): 340 - 353.
Green, P. E. and V. Srinivasan (1978). "Conjoint analysis in consumer research: Issues
and outlook." Journal of Consumer Research 5(2): 103 - 123.
Green, P. E. and V. Srinivasan (1990). "Conjoint analysis in marketing: New
developments with implications for research and practice." Journal of Marketing
54(4): 3 - 19.
Greenwald, A. G. (1968). Cognitive learning, cognitive response to persuasion, and
attitude change. Psychological foundations of attitudes (p 147 - 170). A. G.
Greenwald, T. C. Brock and T. M. Ostrom, New York Academic.
Grunert, K. G. (1997). "What's in a Steak? A cross-cultural study on the quality
perception of beef." Food and Quality Preference 8(3): 157 174.
Gurhan-Canli, Z. and D. Maheswaran (2000). "Cultural variations in country of origin
effects." Journal of Marketing Research 37(3): 309 - 317.
Hair, J. F., R. E. Anderson, et al. (1995). Multivariate Data Analysis with Readings,
Prentice-Hall International Inc.
Hamilton, J., B. Knox, et al. (2000). "Reduced fat products, Consumer perceptions and
preferences." British Food Journal 102(7): 494 - 506.
Hamin and G. Elliot (2006). "A less-developed country perspective of consumer
ethnocentrism and 'country of origin' effects: Indonesian evidence." Asia Pacific
Journal of Marketing 18(2): 79 - 92.
Han, C. M. (1989). "Country Image: Halo or Summary Construct?" Journal of Marketing
Research 26(May): 222 - 229.
Han, C. M. (1990). "Testing the Role of Country Image in Consumer Choice Behaviour."
European Journal of Marketing 6: 24 - 39.
Han, C. M. and V. Terpstra (1988). "Country-of-origin effects for uni-national and bi-
national products." Journal of International Business Studies 19(2): 235 - 255.
Harrison-Walker, L. J. (1995). "The relative effects of national stereotype and
advertising information on the selection of a service provider: an empirical study."
Journal of Services Marketing 9(1): 47 - 59.
Hastak, M. and S. Hong (1991). "Country of origin effects on product quality judgments:
An information integration perspective." Psychology and Marketing 8(2): 129 -
143.
Haubl, G. (1996). "A cross-national investigation of the effects of country of origin and
brand name on the evaluation of a new car." International Marketing Review
13(5): 76 - 97.
He, C. (2003). "Location of foreign manufacturers in China: Agglomeration economies
and country of origin effects." Papers in Regional Science 82(3): 351- 373.
Heimbach, A. E., J. K. Johansson, et al. (1989). "Product Familiarity, Information
Processing and Country-of-Origin Cues." Advances in Consumer Research 16:
460 - 467.
230
Henderson, C. and D. Reibstein (1985). "Holistic Conjoint." Advances in Consumer
Research: 282 - 285.
Herche, J. (1994). "Ethnocentric Tendencies, Marketing Strategy and Import Purchase
Behaviour." International Marketing Review 11(3): 4 - 16.
Heslop, L. A., J. P. Liefeld, et al. (1987). An experimental study of the impact of country-
of-origin information. Marketing. R. E. Turner. Toronto, Canada, ASAC. 8: 179 -
185.
Heslop, L. A., N. Papadopoulos, et al. (1998). "An Interregional Intercultural Perspective
on Sub-cultural Differences in Product Evaluations." Canadian Journal of
Administrative Sciences 15(2): 113 - 127.
Hofestede, G. (1991). "National Cultures in Four Dimensions." International Studies of
Man and Organization 12: 46 - 74.
Hoffmann, R. (2000). "Country of origin - a consumer perception perspective of fresh
meat." British Food Journal 102(3): 211 - 229.
Hofstede, G. (1983). "National Cultures in Four Dimensions A Research-based Theorgy
of Cultural Differences among Nations." Int. Studies of Man. & Org. XIII(1-2): 46 -
74.
Hong, S. and R. S. Wyer (1989). "Effects of country-of-origin and product attribute
information or product evaluation: An information processing perspective." Journal
of Consumer Research 16(2): 175 - 187.
Hong, S. and R. S. Wyer (1990). "Determinants of product evaluation: Effects of the
time interval between knowledge of a product's country-of-origin and information
about specific attributes." 17(3): 17 - 33.
Horowitz, I. and L. Lockshin (2002). "What Price Quality? An Investigation into the
Prediction of Wine-quality Ratings." Journal of Wine Research 13(1): 7 - 22.
Huang, C.-W. and A.-P. Tai (2003). "A Cross-Cultural Comparison of Customer Value
Perceptions for Products: Consumer Aspects in East Asia." Cross Cultural
Management 10(4): 43 - 60.
Huber, J. (1997). What We Have Learned from 20 Years of Conjoint Research: When to
use Self-Explicated, Graded Pairs, Full Profiles or Choice Experiments. Research
Paper Series. S. Software. Sequim, Sawtooth Software Inc.
Huber, J. C., D. R. Wittink, et al. (1991). An Empirical Comparison of ACA and Full
Profile Judgments. Sawtooth Software Conference, Ketchum, Populus.
Huffman, C. (1997). "Beyond Brand Equity: Managing Mature Brands." Advances in
Consumer Research 24: 251 - 252.
Hui, M. K. and L. Zhou (2003). "Country-of-manufacture effects for known brands."
European Journal of Marketing 37(1/2): 133 - 153.
Hunt, S. D. (1991). Modern Marketing Theory: Critical Issues in the Philosophy of
Marketing Science. Cincinnati OH, South Western.
Hurling, R. and R. Shepherd (2003). "Eating with your eyes: effect of appearance on
expectations of liking." Appetite 41: 167 - 174.
231
Hussey, J. and R. Hussey (1997). Business Research: A Practical Guide for
Undergraduate and Post Graduate Students. London, Mac Millan Press.
Imram, N. (1999). "The role of visual cues in consumer perception and acceptance of a
food product." Nutrition & Food Science 5(September / October 1999): 224 - 228.
Insch, G. S. and J. B. McBride (2004). "The impact of country-of-origin cues on
consumer perceptions of product quality: A bi-national test of the decomposed
country-of-origin construct." Journal of Business Research 57: 256 - 265.
Jaeger, S. R., D. Hedderley, et al. (2000). "Methodological issues in conjoint analysis: a
case study." European Journal of Marketing 35(11/12): 1217 1237.
Jaffe, E. D. and I. D. Nebenzahl (1984). "Alternative questionnaire formats for country
image studies." Journal of Marketing Research 21(4): 463 - 471.
James, W. L. and B. S. Sonner (2001). "Just Say No to Traditional Student Samples."
Journal of Advertising Research September: 63 - 71.
Janda, S. and C. P. Rao (1997). "The effect of country-of-origin related stereotypes and
personal beliefs on product evaluation." Psychology & Marketing 14(7): 689 - 702.
Janiszewski, C. and M. C. Jr. (2004). "The Influence of Price Discount Framing on the
Evalution of a Product Bundle." Journal of Consumer Research 30(534 - 546).
Jarvis, C. B., S. B. MacKenzie, et al. (2003). "A Critical Review of Construct Indicators
and Measurement Model Misspecification in Marketing and Consumer Research."
Journal of Consumer Research 30(September): 199 - 218.
Javalgi, R. G., B. D. Cutler, et al. (2001). "At your Service! Does country of origin
research apply to services." Journal of Services Marketing 15(7): 565 - 582.
Jo, M.-S., K. Nakamoto, et al. (2003). "The shielding effects of brand image against
lower quality countries-of-origin in global manufacturing." Journal of Business
Research 56(8): 637 - 346.
Jover, A. J. V., F. J. L. Montes, et al. (2004). "Measuring perceptions of quality in food
products: the case of red wine." Food and Quality Preference 15: 453 - 469.
Kardes, F. R., M. L. Cronley, et al. (2004). "The Role of Selective Information
Processing in Price-Quality Inference." Journal of Consumer Research
31(September): 368 - 374.
Kardes, F. R., J. Kim, et al. (2001). "Consumer Expertise and the Perceived
Diagnosticity of Inference." Advances in Consumer Research 19: 409 - 410.
Katoshevski, R. and H. Timmermans (2001). "Using Conjoint Analysis to Formulate
User-centred guidelines for Urban Design: The Example of New Residential
Development in Israel." Journal of Urban Design 6(1): 37 - 53.
Kaynak, E. and A. Kara (2000). "Consumer perceptions of foreign products An analysis
of product-country images and ethnocentrism." European Journal of Marketing
36(7/8): 928 - 949.
Kaynak, E., O. Kucukemiroglu, et al. (1999). "Consumers' county-of-origin (COO)
perceptions of imported products in a homogenous less-developed country."
European Journal of Marketing 34(9/10): 1221 - 1241.
232
Kennedy, O., B. Stewart-Knox, et al. (2004). "The influence of context upon consumer
sensory evaluation of chicken-meat quality." British Food Journal 106(3): 158 -
165.
Keown, C. and M. Casey (1995). "Purchasing behaviour in the Northern Ireland wine
market." British Food Journal 97(1): 17- 20.
Klein, J. G. (2002). "Us Versus Them, or Us Versus Everyone? Delineating Consumer
Aversion to Foreign Goods." Journal of International Business Studies 33(2): 345 -
363.
Kleppe, I. A., N. M. Iversen, et al. (2001). "Country Images in Marketing Strategies:
Conceptual issues and an empirical Asian illustration." Brand Management
10(September): 61 - 74.
Knight, G. A. and R. J. Calatone (2000). "A flexible model of consumer country-of-origin
perceptions - a cross-cultural investigation." International Marketing Review 17(2):
127 - 145.
Koch, C. and E. C. Koch (2003). "Preconceptions of taste based on color." Journal of
Psychology 137(3): 233 - 242.
Kotler, P. and D. Gertner (2002). "Country as brand, product and beyond: A place
marketing and brand management." Brand Management 9(4): 249 - 261.
Krishna, A., M. Wagner, et al. (2006). "Effects of Extreme-Priced Products on
Consumer Reservation Prices." Journal of Consumer Psychology 16(2): 176 -
190.
Krosnick, J. A., D. S. Boniger, et al. (1993). "Attitude strength: one construct or many
related constructs?" Journal of Personality and Social Psychology 65(6): 1132 -
1151.
Kupiec, B. and B. Revell (2001). "Measuring consumer quality judgments." British Food
Journal 103(1): 7 - 22.
Kuusela, H., M. T. Spence, et al. (1998). "Expertise effects on pre-choice decision
processes and final outcomes: A protocol analysis." European Journal of
Marketing 32(5/6): 559 - 576.
Lampert, S. I. and E. D. Jaffe (1997). "A dynamic approach to country-of-origin effect."
European Journal of Marketing 32: 61 - 78.
Langer, E. J. (1983). The Psychology of Control. Beverly Hills CA, Sage.
Lawrence, C., N. E. Marr, et al. (1992). "Country-of-origin stereotyping: A case study in
the New Zealand moter industry." European Journal of Marketing 26(3): 37 - 51.
Leclerc, F., B. H. Schmitt, et al. (1994). "Foreign branding and its effects on product
perceptions and attitudes." Journal of Marketing Research 31(2): 263 - 270.
Lee, D. and G. Ganesh (1999). "Effects of partitioned country image in the context of
brand image and familiarity." International Marketing Review 61(1): 19 - 39.
Lee, M. and C.-C. Lou (1996). "Consumer reliance on intrinsic and extrinsic cues in
product evaluations: A conjoint approach." Journal of Applied Business Research
12(1): 21 - 30.
233
Leonidou, L. C., J. Hadjimarcou, et al. (1998). "Bulgarian consumers' perceptions of
products made in Asia Pacific." International Marketing Review 16(2): 126 - 142.
Li, Z. G., L. W. Murray, et al. (2000). "Global Sourcing, Multiple Country-of-Origin
Facets, and Consumer Reactions." Journal of Business Research 47: 121 - 133.
Liefeld, J. P., L. A. Heslop, et al. (1996). "Dutch consumer use of intrinsic, country-of-
orign and price cuses in product evaluation and choice." Journal of International
Consumer Marketing 9(1): 57 - 81.
Lim, K. and A. O'Cass (2001). "Consumer brand classifications: an assessment of
culture-of-origin versus country-of-origin." Journal of Product & Brand
Management 10(2): 120 - 136.
Lin, L.-W. and B. Sternquist (1994). "Taiwanese Consumers' Perceptions of Product
Information Cues: Country of Origin and Store Prestige." European Journal of
Marketing 28(1): 5 - 18.
Llusar, J. C. B. and C. C. Zornoza (2000). "Validity and reliability in perceived quality
measurement models: An empirical investigation in Spanish ceramic companies."
International Journal of Quality & Reliability Management 17(8): 899 - 918.
Loeffler, M. (2000). "A multinational examination of the "(non-)domestic product effect."
International Marketing Review 19(5): 482 - 498.
Lorr, M. (1991). "A Redefinition of Dominance." Personality and Individual Differences
12(9): 877 - 879.
Louviere, J. (1988b). "Analyzing Decision Making: Metric Conjoint Analysis." Sage
University Papers Series In Quintitative Applications in Social Sciences 67.
Louviere, J. J. (1979). "Modeling individual residential preferences: a totally
disaggregated approach." Transportation Research A(13): 373 - 384.
Louviere, J. J. (1988). Analyzing Decision Making: Metric Conjoint Analysis. Beverly
Hills, CA, Sage Publications.
Lundstrom, W. J., O. W. Lee, et al. (???). "Factors Influencing Taiwanese Consumer
Preference for Foreign-made White Goods: USA Versus Japan." Asia Pacific
Journal of Marketing and Logistics: 5 - 29.
Lusk, J. L. and F. B. Norwood (2005). "Effect of Experimental Design on Choice-Based
Conjoint Valuation Estimates." American Journal of Agricultural Economics 87(3):
771 - 785.
Maheswaran, D. (1994). "Country of Origin as a Stereotype: Effects of Consumer
Expertise and Attribute Strength on Product Evaluations." Journal of Consumer
Marketing 21: 354 - 365.
Maheswaran, D., B. Sternthal, et al. (1996). "Acquisition and Impact of Consumer
Expertise." Journal of Consumer Psychology 5(2): 115 - 133.
Malhotra, N. (1986). "An approach to the Measurement of Consumer Preferences Using
Limited Information." Journal of Marketing Research(February): 33 - 40.
Malhotra, N., J. Hall, et al. (2002). Marketing Research an Applied Orientation, Prentice
Hall.
234
Malhotra, N. K. (1981). "A scale to measure self-concepts, person concepts and
product concepts." Journal of Marketing Research 18(4): 456 - 464.
Malhotra, N. K., J. Agarwal, et al. (1996). "Methodological issues in cross-cultural
marketing research." International Marketing Review 13(5): 7 - 43.
Mangan, J., C. Lalwani, et al. (2004). "Combining Quantitative and Qualitative
Methodologies in Logistics Research." International Journal of Physical
Distribution and Logistics Management 34(7): 565 - 578.
Mannion, M. A., C. Cowan, et al. (2000). "Factors associated with perceived quality
influencing beef consumption behaviour in Ireland." British Food Journal 102(3):
195 - 210.
Manrai, L. A., D.-N. Lascu, et al. (1998). "Interactive effects of country of origin and
product category on product evaluations." International Business Review 7: 591 -
615.
Maronick, T. J. (1995). "An empirical investigation of consumer perceptions of 'made in
USA' claims." International Marketing Review 12(3): 15 - 30.
Martin, I. M. and S. Eroglu (1993). "Measuring a Multi-Dimensional Construct: Country
Image." Journal of Business Research 28: 191 - 210.
Mason, K. and J. Bequette (1998). "Product experience and consumer product attribute
inference accuracy." Journal of Consumer Marketing 15(4): 343 - 357.
McAlexander, J. H., D. Kaldenberg, et al. (1994). "Service Quality Management.
Examination of dental practices sheds more light on the relationships between
service quality, satisfaction, and purchase intentions in a health care setting."
Journal of Health Care Marketing 3: 34 - 39.
McCollough, J. and R. Best (1979). "Conjoint measurement: Temporal Stability and
Structural Reliability." Journal of Marketing Research(February): 26 - 31.
McIlveen, H. and J. Buchanan (2001). "The impact of sensory factors on beef purchase
and consumption." Nutrition & Food Science 31(6): 286 - 292.
Mitchell, A. A. and P. A. Dacin (1996). "The Assessment of Alternative Measures of
Consumer Expertise." Journal of Consumer Research 23(December): 219 - 239.
Mohamad, O., Z. U. Ahmed, et al. (2000). "Does 'Made in..." Matter to Consumers? A
Malaysian Study of Country of Origin Effect." Multinational Business Review(Fall):
69 - 73.
Monroe, K. B. (1976). "The influence of price differences and brand familiarity on brand
perferences." Journal of Consumer Research 3: 42 - 49.
Monroe, K. B. (1982). "The influence of price on product perceptions and product
choice." Advances in Consumer Research 9(1): 206 - 209.
Moorman, C. (2001). "Consumer Confidence and Knowledge Calibration." Advances in
Consumer Research 28: 47.
Moy, J. W. and K. F. Lam (2003). "Selection criteria and the impact of personality on
getting hired." Personnel Review 33(5): 521 - 535.
Murphy, M., C. Cowan, et al. (2000). "Irish consumer preferences for honey: a conjoint
approach." British Food Journal 102(8): 585 - 597.
235
Nagashima, A. (1970). "A comparison of Japanese and U.S. Attitudes Toward Foreign
Products." Journal of Marketing 34: 68 - 74.
Naslund, D. (2002). "Logistics needs qualitative research: especially action research."
International Journal of Physical Distribution and Logistics Management 32(5):
321 - 338.
Nebanzahl, I. and E. D. Jaffe (1996). "Measuring the joint effect of brand and country
image in consumer evaluation of global products." International Marketing Review
13(4): 5 - 22.
Ness, M. R. and H. Gerhardy (1994). "Consumer Preferences for Quality and
Freshness Attributes of Eggs." British Food Journal 96(3): 26 - 34.
Niss, H. (1996). "Country of origin marketing over the product life cycle." European
Northen, J. R. (2000). "Quality attributes and quality cues, Effective communication in
the UK meat supply chain." British Food Journal 102(3): 230 - 245.
Obermiller, C. and E. R. Spangenberg (1998). "Development of a Scale to Measure
Consumer Skepticism toward Advertising." Journal of Consumer Psychology 7(2):
159 - 186.
Obermiller, C. and E. Spangenbert (1988). "Exploring the Effects of Country of Origin
Labels: An information Processing Framework." Advances in Consumer Research
16(454 - 459).
O'Cass, A. and K. Lim (2002). "Understanding the Younger Singaporean Consumers'
Views of Western and Eastern Brands." Asia Pacific Journal of Marketing and
Logistics 14(4): 54 - 79.
Okechuku, C. (1994). "The importance of Product Country of Origin: A conjoint Analysis
of the United States, Canada, Germany and the Netherlands." European Journal
of Marketing 28(4): 5 - 19.
Olsen, S. O. (1999). "Strength and conflicting valence in measurement of food attitudes
and preferences." Food and Quality Preference 10: 483 - 494.
Olsen, S. O. and U. H. Olsson (2002). "Multi-entity Scaling and the Consistency of
Country of Origin Attitudes." Journal of International Business Studies 33(1): 149 -
167.
Olson, J. C. (1972). "Cue Utilization of the Quality Perception Process: A Cognitive
Model and an Empirical Test." Purdue University.
Ophuis, P. A. M. O. and H. C. M. V. Trijp (1995). "Perceived quality: a market driven
and consumer oriented approach." Food and Quality Preference 6: 177 - 183.
Oppewal, H. and M. Vriens (2000). "Measuring perceived service quality using
integrated conjoint experiments." International Jurnal of Bank Marketing 18(4):
154 - 169.
Owens, T. J. (1993). "Accentuate the Positive and the Negative: Rethinking the Use of
Self-Esteem, Self Depreciation and Self-Confidence." Social Psychology Quarterly
56(4): 288 - 299.
Papadopoulos, N. and L. A. Heslop (1989). "National Stereotypes and Product
Evaluations in a Socialist Country." International Marketing Review 7(1): 32 - 47.
236
Papadopoulos, N. and L. A. Heslop (1993). Product Country Images - Impact and Role
in International Marketing., International Press.
Papadopoulos, N. and L. A. Heslop (2002). "Country equity and country branding:
Problems and prospects." Brand Management 9(April): 294 - 314.
Parameswaran, R. and R. H. Pisharodi (2002). "Assimilation effects in country image
research." International Marketing Review 19(3): 259 - 278.
Parasuraman, A., L. L. Berry, et al. (1993). "More on Improving Service Quality
Measurement." Journal of Retailing 69(1): 140 - 147.
Park, C. W. and V. P. Lessig (1981). "Familiarity and Its Impact on Consumer Decision
Biases and Hueristics." Journal of Advertising(8 (September)): 223 - 230.
Park, C. W., D. L. Mothersbaugh, et al. (1994). "Consumer knowledge assessment."
Journal of Consumer Research 8: 71 - 82.
Pechmann, C. and S. Ratneshwar (1992). "Consumer Co variation Judgments: Theory
or Data Driven." Journal of Consumer Research 19(December): 373 - 386.
Pelsmacker, P. D., L. Driesen, et al. (2005). "Do Consumers Care about Ethics?
Willingness to Pay for Fair-Trade Coffee." Journal of Consumer Affairs 39(2): 363
- 385.
Pereira, A., C.-C. Hsu, et al. (2004). "Country-of-origin image: measurement and cross-
national testing." Journal of Business Research.
Peterson, R. A. (2001). "On the Use of College Student in Social Science Research:
Insights from a Second-Order Meta-analysis." Journal of Consumer Research
28(December): 450 - 461.
Peterson, R. A., G. Albaum, et al. (1985). "A Meta-Analysis of Effect Sizes in Consumer
Behaviour Experiments." Journal of Consumer Research 12(June): 97 - 103.
Phau, I. and G. P. Prendergast (2000). "Conceptualizing the Country of Origin of
Brand." Journal of Marketing Communications 6: 159 - 170.
Phau, I. and V. Suntornnond (2006). "Dimensions of consumer knowledge and its
impacts on country of origin effects among Australian consumers: a case of fast-
consuming product." Journal of Consumer Marketing 23(1): 34 - 42.
Piron, F. (2000). "Consumers' Perceptions of the Country-of-Origin Effect on
Purchasing Intention of Inconspicuous Products." Journal of Consumer Research
17(4): 308 - 321.
Pisharodi, R. M. and R. Parameswaran (1991). "Confirmatory Factor Analysis of a
Country-of-Origin Scale: Initial Results." Advances in Consumer Research 17: 704
- 714.
Puisais, J. and R. L. Chabanon (1974). Initiation into the Art of Wine Tasting,
Interpublish Inc.
Punj, G. N. and S. Staelin (1983). "A Model of Consumer Information Search Behaviour
for New Automobiles." Journal of Consumer Research 9(March): 366 - 380.
237
Quester, P. G., S. Dzever, et al. (2000). "Country-of-origin effects on purchasing agents'
product perceptions: an international perspective." Journal of Business &
Industrial Marketing 15(7): 479 - 490.
Quester, P. G., N. E. Marr, et al. (1996). "Country-of-origin effects: an Australian
experiment in shelf labeling." The International Review of Retail, Distribution and
Consumer Research 6(1): 113 - 133.
Quester, P. G. and J. Smart (1998). "The influence of consumption situation and
product involvement over consumers' use of product attribute." Journal of
Consumer Marketing 15(3): 220 - 238.
Raden, D. (1985). "Strength-related attitude dimensions." Social Psychology quarterly
48(4): 312 - 330.
Rao, A. R. and K. B. Monroe (1988). "The Moderating Effect of Prior Knowledge on Cue
Utilization in Product Evaluations." Journal of Consumer Research
15(September): 253 - 264.
Rao, A. R. and E. M. Olson (1990). "Information Examination as a Function of
Information Type and Dimension of Consumer Expertise: Some Exploratory
Findings." Advances in Consumer Research 17: 361 - 366.
Rao, V. R. and J. R. Hauser (2004). Conjoint Analysis, Related Modeling, and
Application. Market Research and Modeling: Progress and Prospects: A Trubute
to Paul Green. Y. Wind and P. E. Green, Kluwer Academic Publishers: 141 - 168.
Rexeisen, R. J. (1982). "Is There a Valid Price Quality Relationship?" Advances in
Consumer Research 9(1): 190 - 194.
Roper, B. W. (1969). "Sensitivity, Reliability, and Consumer Taste Testing: Some
'Rights and 'Wrongs'." Journal of Marketing Research VI(February): 102 - 106.
Ross, M. (2004). Confirmation of cheese questions - Michael Ross, (Director) The Diary
Farmers Group of Companies. R. Veale. Adelaide: Personal interview.
Sagar, H. A. and J. W. Scofield (1982). "Racial and Behavioral Cues in Black and White
Children's Perceptions of Ambiguously Aggressive Acts." Journal of Personality
and Social Psychology 39(October): 590 - 598.
ScanTrack-Liquor (2005). AC Nielsen Bottled Table Wine Market Update. ScanTrack
Liquor AC Neilsen, AC Nielsen.
Schaefer, A. (1997). "Consumer knowledge and country of origin effects." European
Schifferstein, H. N. J. (1996). "Cognitive Factors Affecting Taste Intensity Judgments."
Food and Quality Preference 7(3/4): 167 - 175.
Schooler, R. D. (1971). "Bias phenomena attendant to the marketing of foreign goods in
the U.S." Journal of International Business Studies 2: 71 - 81.
Seaman, C. E. A., A. H. Hughes, et al. (1993). "Consumers as Sensory Panelists."
British Food Journal 95(8).
Segal, M. (1982). "Reliability of Conjoint Analysis: Contrasting Data Collection
Procedures." Journal of Marketing Research 19: 139 - 143.
238
Showers, V. E. and L. S. Showers (1993). "The Effects of Alternative Measures of
Country of Origin on Objective Product Quality." International Marketing Review
10(4).
Siu, N. Y.-M. and H.-Y. Wong (2002). "The impact of product-related factors on
perceived product safety." Marketing Intelligence & Planning 20(3): 185 - 194.
Skuras, D. and A. Vakrou (2002). "Consumers' willingness to pay for origin labeled
wine: a Greek case study." British Food Journal 104(1): 898 - 912.
Smith, J. (2004). Confirmation of wine questions - Jim Smith (Director) Jim Smith
Consulting Pty Ltd. R. Veale. Adelaide South Australia: Personal Interview.
Smith, R. W. (1993). "Country-of-Origin Bias: A Regional Labeling Solution."
International Marketing Review 10(6): 4 - 12.
Snoj, B., A. Pisnik, et al. (2004). "The relationships among perceived quality ,perceived
risk and perceived product value." Journal of Product & Brand Management 13(3):
156 - 167.
Speece, M. W., Y. Kawahara, et al. (1994). "Imported Beer in the Hong Kong Market."
British Food Journal 96(1): 10 - 18.
Spence, M. T. and M. Brucks (1997). "The moderating effects of problem characteristics
on experts' and novices' judgments." Journal of Marketing Research(May): 223 -
247.
SPSS-Inc (1997). SPSS Conjoint 8.0. Chicago Il, SPSS Inc.
Srikatanyoo, N. and J. Gnoth (2002). "Nation Branding - Country image and
international tertiary education." Brand Management 10(2): 139 - 146.
Srinivasan, N., S. C. Jain, et al. (2004). "An experimental study of two dimensions of
country-of-origin (manufacturing country and branding country) using intrinsic and
extrinsic cues." International Business Review 13: 65 - 82.
Sullivan, G. L. and K. Burger (1987). "An Investigation of the Determinants of Cue
Utilization." Psychology & Marketing 4(1): 63 - 72.
Sweeney, J., G. N. Soutar, et al. (1999). "The role of Perceived Risk in the Quality-
Value Relationship: A Study in a Retail Environment." Journal of Retailing 75(1):
77 - 105.
Szbillo, G. J. and J. Jacoby (1974). "Intrinsic versus extrinsic cues as determinants of
perceived product quality." Journal of Applied Psychology 59(1): 74 - 78.
Tan, S. J. (1999). "Strategies for reducing consumers' risk aversion in Internet
shopping." Journal of Consumer Marketing 16(2): 163 - 180.
Teas, K. R. and S. Agarwal (2000). "The Effects of Extrinsic Product Cues on
Consumer's Perceptions of Quality, Sacrifice and Value." Journal of the Academy
of Marketing Science 28(2): 278 - 291.
Thakor, M. V. and C. S. Kohli (1996). "Brand origin: conceptualization and review."
Journal of Consumer Marketing 13(3): 27 - 42.
Thakor, M. V. and A. M. Lavack (2003). "Effect of perceived brand origin associations
on consumer perceptions of quality." Journal of Product & Brand Management
12(6): 394 - 407.
239
Tharp, M. and L. Marks (1990). "An Examination of the Effects of Attribute Order and
Product Order Biases on Conjoint Analysis." Advances in Consumer Research 17:
563 - 571.
Thode, S. F. and J. M. Maskulka (1998). "Place-based marketing strategies, brand
equity and vineyard valuation." Journal of Product & Brand Management 7(5): 379
- 399.
Thompson, K. E. and P. Panayiotopoulos (1999). "Predicting behavioral intention in a
small business context." Journal of Marketing Practice 5(3): 89 - 96.
Thorelli, H. B., J.-S. Lim, et al. (1989). "Relative Importance of Country of Origin,
Warranty, and Retail Store Image on Product Evaluations." International
Marketing Review 6(1).
Till, B. D. and M. Busler (1998). "Matching products with endorsers: attractiveness
versus expertise." Journal of Consumer Marketing 15(6): 576 - 586.
Toyado, H. and M. Mayanagi (2001). "A factor analysis model for an experiment having
repeated measures: A sensory evaluation of ice cream." Japanese Journal of
Behaviour-metrics 28(1): 1 - 7.
Tse, A. C. B. (1999). "Factors affecting consumer perceptions on product safety."
European Journal of Marketing 33(9): 911 - 925.
Verlegh, P. W. J. (2002). "Country-of-Origin Stereotypes and the Precession of Ads: A
Tomato-Field Experiment." Advances in Consumer Research 29: 166 - 167.
Verlegh, P. W. J. and J. E. M. Steenkamp (1999). "A review and meta analysis of
country of origin research." Journal of Economic Psychology 20(5): 521 - 546.
Vranesevic, T. and R. Stancec (2003). "The effect of the brand on perceived quality of
food products." British Food Journal 105(11): 811 - 825.
Wakefield, K. L. and J. J. Inman (1993). "Who are the Price Vigilantes? An Investigation
of Differentiating Characteristics Influencing Price Information Processing."
Journal of Retailing 69(2): 216- 233.
Wansink, B., S. B. Park, et al. (2000). "How soy labeling influences preferences and
taste." International Food and Agribusiness Management Review 3: 85 - 94.
Watson, J. J. and K. Wright (1999). "Consumer Ethnocentrism and Attitudes Towards
Domestic and Foreign Products." European Journal of Marketing 34(9/10): 1149 -
1166.
White, P. D. and E. W. Cundiff (1978). "Assessing the quality of industrial products."
Journal of Marketing 42: 80 - 86.
Wilson, T. D. and N. Brekke (1994). "Mental Contamination and Mental Correction:
Unwanted Influences on Judgments and Evaluations." Psychological Bulletin
116(1): 117 - 142.
Wirtz, J. and A. S. Mattila (2003). "The effects of consumer expertise on evoked set
size and service loyalty." Journal of Services Marketing 17(7): 649 - 665.
Witt, J. and C. P. Rao (1992). "Country of origin and brand effects on consumers'
perceived risk." Journal of Global Business 3(2): 27 - 33.
240
Wood, R. C. (1996). "Talking to themselves: food commentators, food snobbery and
market reality." British Food Journal 98(10): 5 - 11.
www.Dairy-Australia.com.au (2007). Dairy Australia, Cheese production statistics 2006.

March 2007.
Yong, Z. (1995). "Chinese consumers' evaluation of foreign products: the influence of
culture, product types and product presentation format." European Journal of
Marketing 30(12): 50 - 68.
Yong, Z. (1997). "Country-of-origin effect The Moderating function of individual
difference in information processing." International Marketing Review 14(4): 266 -
287.
Zain, O. M. and N. M. Yasin (1997). "The importance of country-of-origin information
and perceived product quality in Uzbekistan." International Journal of Retail and
Distribution Management 25(4): 138 - 145.
Zeithaml, V. A. (1988). "Consumer Perceptions of Price, Quality, and Value: A Means-
End Model and Synthesis of Evidence." Journal of Marketing 52(July): 2 - 22.
Zellner, D. A. and P. Durlach (2003). "Effect of color on expected and experienced
refreshment, intensity and liking of beverages." American Journal of Psychology
116(4): 633 - 647.
Zhang, Y. (1996). "Country of origin effect the moderating function of individual
difference in information processing." International Marketing Review 14(4): 266 -
287.
241
12 Appendices
1 Focus group discussion guide
242
Focus group discussion guide
March 22 / 23, 2005
1. WINE (1):
1. What types of wine do you like to drink?
a. What kinds of wine do you usually buy?

b. Do you tend to buy different wines for different occasions?
2. When thinking about the wine you drink / buy, what kinds of things do you
consider when making your decisions? (list).
3. Lets rate these things in terms of importance summated scale.
4. When thinking about wine, say a chardonnay, where would you expect the best
ones to come from? What countries come to mind? (list)
5. Lets try some wines! (Hand out the form for wine)
a. On your form, please put your name, and allocate each wine a score out
of 10, where 10 is excellent really good and 0 is terrible.
b. Also indicate if you would buy this wine based on taste.
6. Between each taste test, please have a drink of water and cracker.
2. CHEESE (1):
7. What types of cheeses do you like to eat?
a. What kinds of cheeses do you usually buy?
8. When thinking about the cheese you eat / buy what kinds of things do you
consider when making your decisions? (list).
9. Lets rate these things in terms of importance summated scale.
10. When thinking about cheese, say cheddar, where would you expect the best
ones to come from? What countries come to mind? (list)
11. Lets try some cheese! (Hand out the form for cheese)
a. On your form, please put your name, and allocate each cheese a score
out of 10, where 10 is excellent really good and 0 is terrible.
b. Also indicate if you would buy this cheese based on taste.
12. Between each taste test, please have a drink of water and cracker.
243
3. WINE (2):
13. This time, were going to try some wines again.
a. But, you wont know where they are from.

b. Dont assume they are all the same as the first 4
c. Please again give each a score out of 10.
d. Indicate whether or not you would buy it based on taste.
e. Between each taste test, please have a drink of water and cracker.
4. CHEESE (2):
14. This time, were going to try some cheeses again this time 4.
a. But, this time you wont know where they are from.
b. Dont assume they are all the same as the first 3
c. Please again give each a score out of 10.
d. Indicate whether or not you would buy it based on taste.
e. Between each taste test, please have a drink of water and cracker.
244
Appendix
2 Pilot study questionnaire
245
School of Commerce
Research Project Food Product Preferences
Hello!
I am a student at the University of Adelaide, working on a research project about preferences for
wine and cheese products. This questionnaire should take between 10 to 15 minutes to complete.
Thank you very much for participating in this study.
Why is the research being conducted?

Were doing some basic research on peoples evaluation and decision processes.
What you will be asked to do

In the following pages are descriptions of chardonnay and camembert. Please give a score for
each to show how much, or how little, you like them. Then, let us know whether or not you would
consider buying them.
The expected benefits of the research
While this study is pretty simple, it will help us to develop statistical tools with applications as
diverse as market share forecasting, political votes or environmental policy.
Risks to you
None. (Except for perhaps, a mild risk of wine or cheese cravings!)
Your confidentiality
We dont need to identify you, so this study is entirely confidential.
Distribution and return
When you have completed the questions, please hand back this questionnaire to the person who
gave it to you.
Questions / further information

Regardless of your participation, wed be delighted to give you further information about this project
or to give you copies of our results. Contact Roberta Veale on Roberta.Veale@tafesasouth.org, or
Professor Pascale Quester on pascale.quester@adelaide.edu.au.
Consent
By completing and returning the questionnaire, you have consented to participation in this
research. Please detach this sheet and retain it for your later reference.
246
Chardonnay Questions
Part 1 (a)
Your task:
The following pages include 11 descriptions of Chardonnay wine. Look over the descriptions and give each
Chardonnay a rating on the scale where 1 = Highly undesirable and 10 = Highly desirable. While you
may not be a regular Chardonnay Drinker (or even like Chardonnay personally) assess each profile in light
of the need to make a possible purchase. All Wine options are 2005 vintage.
Were interested in your first impressions, so work quickly and steadily.

There are no right or wrong answers, we are only interested in your opinion!
Chardonnay No. 1
Produced in France
Retail Price $14.00
Highly undesirable Highly desirable
1 2 3 4 5 6 7 8 9 10

Chardonnay No. 2
Produced in United States
Retail Price $6.00
1 2 3 4 5 6 7 8 9 10

Chardonnay No. 3
Acidity High
Produced in Chile
Retail Price $14.00
1 2 3 4 5 6 7 8 9 10

247
Chardonnay No. 4
Acidity High
Produced in France
Retail Price $6.00
1 2 3 4 5 6 7 8 9 10

Chardonnay No. 5
Retail Price $14.00
1 2 3 4 5 6 7 8 9 10

Chardonnay No. 6
Acidity Pronounced
Produced in Chile
Retail Price $39.00
1 2 3 4 5 6 7 8 9 10

Chardonnay No. 7
Acidity Pronounced
Retail Price $6.00
1 2 3 4 5 6 7 8 9 10

248
Chardonnay No. 8
Produced in France
Retail Price $39.00
1 2 3 4 5 6 7 8 9 10

Chardonnay No. 9
Produced in Chile
Retail Price $6.00
1 2 3 4 5 6 7 8 9 10

Chardonnay No. 10
Acidity Pronounced
Produced in France
Retail Price $14.00
1 2 3 4 5 6 7 8 9 10

Chardonnay No. 11
Acidity High
Retail Price $39.00
1 2 3 4 5 6 7 8 9 10

This completes the assessment of the different Chardonnay wines,

please continue the questionnaire by turning the page.
249
Part 1 (b)
In this part of the questionnaire, we would like to ask you some questions about yourself and your general
knowledge and experiences with purchasing Chardonnay. Please (3) the box that best reflects your
knowledge and experience.
Strongly Strongly
Disagree Agree
1 2 3 4 5 6 7 8 9
1. I feel confident about my knowledge of chardonnay.
2. I feel that I know how to judge the quality of chardonnay.
3. I do not feel very knowledgeable about chardonnay.
4. Among my circle of friends, Im one of the experts on chardonnay.
5. Compared to most buyers, I know less about chardonnay.
6. I know of most of the chardonnays around in shops.
7. When it comes to chardonnay, I really dont know a lot.
8. I can tell if a chardonnay is worth the price or not.
9. How often would you purchase bottle of 10. If you have a favourite brand of Chardonnay,
Chardonnay for yourself or others? (on please list it below.
average)
Less than once per month 1
1 to 2 times per month 2
5 or more times per month 4
Camembert Questions
Part 2 (a)
Your task:
The following pages include 11 descriptions of camembert cheese. Look over each of the descriptions and
give each camembert a rating on the scale where 1 = Highly undesirable and 10 = Highly desirable. While
you may not be a regular Camembert consumer (or even like Camembert personally) assess each profile in
light of the need to make a possible purchase. All cheeses are 250 gram packs.

Camembert No. 1
Made with: 50% reduced fat
Produced in France
Retail Price $3.00
1 2 3 4 5 6 7 8 9 10

250
Camembert No. 2
Made with: Triple cream
Produced in Argentina
Retail Price $3.00
1 2 3 4 5 6 7 8 9 10

Camembert No. 3
Produced in Canada
Retail Price $8.00
1 2 3 4 5 6 7 8 9 10

Camembert No. 4
Made with: Full cream
Produced in Canada
Retail Price $3.00
1 2 3 4 5 6 7 8 9 10

Camembert No. 5
Produced in France
Retail Price $5.00
1 2 3 4 5 6 7 8 9 10

251
Camembert No. 6
Retail Price $8.00
1 2 3 4 5 6 7 8 9 10

Camembert No. 7
Produced in France
Retail Price $3.00
1 2 3 4 5 6 7 8 9 10

Camembert No. 8
Retail Price $3.00
1 2 3 4 5 6 7 8 9 10

Camembert No. 9
Produced in France
Retail Price $8.00
1 2 3 4 5 6 7 8 9 10

252
Camembert No. 10
Produced in Canada
Retail Price $5.00
1 2 3 4 5 6 7 8 9 10

Camembert No. 11
Retail Price $5.00
1 2 3 4 5 6 7 8 9 10

Part 2 (b)
In this part of the survey, we would like to ask you some questions about your general knowledge and
experiences with purchasing camembert. Please (3) the box that best reflects your knowledge and
experience about camembert.
Strongly Strongly
Disagree Agree
1 2 3 4 5 6 7 8 9
1. I feel confident about my knowledge of camembert.
2. I feel that I know how to judge the quality of camembert.
3. I do not feel very knowledgeable about camembert.
4. Among my circle of friends, Im one of the experts on camembert.
5. Compared to most buyers, I know less about camembert.
6. I know of most of the camembert cheeses in the shops.
7. When it comes to camembert, I really dont know a lot.
8. I can tell if a camembert is worth the price or not.
How often would you purchase Camembert for If you have a favourite brand of Camembert,
9. 10.
yourself or others? (on average) please list it below.
253
Part 3 (a) Chardonnay
In this part of the survey, we would like to ask you some questions relating to your general knowledge about
wine. Please (3) the box that best reflects your knowledge and experience. If youre unsure about
the answer to any particular question, thats no problem just (3) the box indicating this.

Is to remove the cork for a time, prior to drinking. 1 Wine mature quickly. 1
Means aerating it so it can react with the oxygen. 2 Wine mature slowly 2
Always improves the flavour of wine. 3 Keep pests away from wine 3
3. Which one of the following white varietals is 4. Which one of the following white varietals is most
most likely to be aged in oak? likely to improve with aging?

Bitterness. 1 Among the finest grown for white wine. 1
Tartness. 2 Not usually used in sparkling wine. 2
Astringency. 3 Often used to make sweet wines. 3
7. Chardonnay, typically has an aging potential of: 8. Terms often linked with the taste of Chardonnay
are:
2 or 3 years. 1 Apple, peach, citrus 1
3 or 4 years. 2 Plum, spice, mint 2
9. What percentage (%) of the wine in the bottle 10. When thinking about matching foods with wines,
must be made from grapes harvested and trying to achieve the most complimentary
crushed in the year named, if a Vintage date is combinations, it is important to remember that:
given?
85% 1 Very sweet food with counter the acid in the wine. 1
95% 2 Very salty foods counter acid in the wine. 2
100% 3 Very acid foods will bring out the acid in the wine. 3
11. An oak aged Chardonnay will, typically, be: 12. Champagne is an excellent choice to accompany:
Less full bodied than many other white wines 1 Smoked salmon. 1
More full bodied than other many other white wines 2 Chinese food. 2
Comparable in body to many other white wines 3 Most foods. 3
13. The term green is often used to describe a 14. Chilling wine (even red wine) will often:
wines:
Colour 1 Improve the taste of a poor wine. 1
Acidity 2 Make tannins less noticeable. 2
Age 3 Make no real difference to perceived quality. 3
254
Part 3 (b) Camembert
cheese. Please (3) the box that best reflects your knowledge and experience. If youre unsure about

Fresh curds aged in warm temperatures. 1 The conversion of milk solids to curd and whey. 1
Fresh curds ripened by introducing surface mould. 2 The basis of cheese making. 2
Never found in quality cheeses. 3 Both of the above. 3
3. Rennet is: 4. Camembert:

A form of yeast used to make cheese. 1 Is eaten younger than Brie. 1
An enzyme extracted from stomach linings. 2 Is eaten older than Brie. 2
Neither of the above. 3 Matures in the same time as Brie. 3
5. Rind is: 6. Camembert:

An important influence on flavour development. 1 Has a slightly thicker mould than Brie. 1
Not found on Camembert. 2 Has a slightly thinner mould than Brie. 2
Never washed in making quality cheeses. 3 Has identical mould to Brie. 3
7. Camembert: 8. Camembert is nicest when made from milk

Never smells of mushrooms and yeast. 1 Produced in spring and autumn. 1
Usually smells of mushrooms and yeast. 2 Produced in the winter and summer 2
Can sometimes smell like old socks. 3 Produced in particularly wet summers 3
9. Triple Cream Camembert: 10. Camembert rind is:

Feels greasy with you eat it. 1 Comprised of soft white mould. 1
Is made from milk with extra cream added later. 2 Comprised of grey or white mould. 2
Doesnt exist as a product. 3 Sometimes orange. 3
11. When choosing wine to drink with Camembert: 12. Camembert cheese:
Its Highly desirable to make sure its well chilled. 1 Is a product where price usually influences quality. 1
A chardonnay will suit, but never a red. 2 Is a product where price is no indication of quality. 2
3 Is a product where all brands cost pretty much the 3
A full bodied, earthy red is a sound choice.
same, no matter where the cheese comes from.
13. The inside of a Camembert that is ready to 14. When you press the rind of a Camembert
eat is: it should:
Creamy, buttery and smooth. 1 Feel soft, but your finger leaves no mark or indent. 1
Rubbery, buttery and soft. 2 Feel soft to the touch, your finger leaving an indent 2
A little runny, but nice and smooth. 3 Feel quite firm. 3
255
Part 4
For the purpose of analysis, we would be grateful if you would provide some personal information about
yourself. We do not need your name or address, only information that will help us analyse your responses for
this research. Your anonymity is fully guaranteed. Your answers will only appear in aggregate and average
number and cannot be associated with any one respondent.
Please (3) the box that best reflects your agreement with the following statements.
Strongly Strongly
Disagree Agree
1 2 3 4 5 6 7 8 9

It doesnt bother me to have to enter a room where other people have

3.
already gathered and are talking.
4. In group discussions, I usually feel my opinions are inferior.
5. I dont make a very favorable first impression on people.
When confronted by a group of strangers, my first reaction is always

6.
one of shyness and inferiority.
It is extremely uncomfortable to accidentally go to a party wearing the
7.
wrong thing.
When in a group, I very rarely express an opinion for fear of being

9.
laughed at.
I am never at a loss for words when I am introduced to someone I dont
10.
know.
11. Please indicate your gender: 12. Please indicate your age category:
Male 1 18 yrs to 25 yrs 1
Female 2 26 yrs to 35 yrs 2
13. What is your occupation? 36 yrs to 45 yrs 3
46 yrs to 55 yrs 4
Over 55 yrs 5
14. Please indicate your household income 15. Please indicate highest level of education
(gross) completed:
Less than $25, 000 1 High School Certificate 1
$25,000 to $45,000 2 Diploma / Trade Qualification 2
$46,000 to $65,000 3 Bachelors Degree 3
Over $65,000 4 Post Graduate Degree 4
Thank you, once again, for completing this survey!

Please turn the page and complete the form to enter
the draw to win $250 in cash!
256
Appendix
3 Normality assumptions and testing
257
3.1 Assum ptions for normality
Parametric test results are not reliable unless variables are normally distributed. The
normal distribution of a population is a bell shaped, symmetrical curve with 68% of
the population lying within 1 standard deviation of the mean and 95% of the
distribution within 2 standard deviations of the mean and 99% within 3 (Malhotra, Hall
et al. 2002).
Normality can be confirmed through tests for skewness (height and width of curve)
and Kurtosis (deviation from symmetry). Scores for each test range from 0 to 2, with
a score of 0 in either test signifying perfect normality. Curves with scores clearly
different to 0 indicate distributions that are not normal and a non-parametric text
should be used. The third procedure commonly used is the Kolmogorov-Smirnov
test. A low significance (<0.05) indicates that the distribution of variables is not
normal with 95% confidence.
Normality test results for all quantitative stages of the research are illustrated in
specified tables in the following sections.
258
3.2 Pilot study data
Table A 1 Normality tests (pilot)
Kolmogorov
Skewness Kurtosis
Smirnov
Std. Std
Mean Statistic Statistic Statistic Sig
Error Error
Objective chardonnay 3.119 0.300 0.164 -0.491 0.327 1.808 0.003
Subjective chardonnay 4.244 0.071 0.164 -0.150 0.327 0.960 0.315
Self confidence 6.094 -0.397 0.164 0.657 0.327 0.814 0.522
Utility France -0.332 0.155 0.164 1.337 0.327 1.1186 0.120
Utility United States -0.240 -0.650 0.164 1.530 0.327 1.709 0.006
Utility Chile 0.572 0.430 0.164 1.693 0.327 1.372 0.046
Utility $39.00 -0.463 0.345 0.164 1.585 0.327 1.283 0.074
Utility $14.00 0.247 0.195 0.164 0.115 0.327 0.972 0.301
Utility $6.00 0.216 -0.102 0.164 0.400 0.327 1.062 0.209
Utility average 0.355 0.822 0.164 2.135 0.327 1.482 0.025
Utility above average 0.038 -0.156 0.164 1.517 0.327 1.602 0.012
Utility high -0.393 -0.202 0.164 1.477 0.327 1.003 0.267
Objective camembert 3.425 0.477 0.164 -0.622 0.327 1.791 0.003
Subjective camembert 4.223 -0.095 0.164 -0.500 0.327 1.127 0.158
Utility France 0.420 0.983 0.164 1.457 0.327 1.728 0.005
Utility Canada -0.059 -0.099 0.164 0.190 0.327 1.007 0.263
Utility Argentina -0.361 -1.170 0.164 2.154 0.327 1.882 0.002
Utility $8.00 0.079 -0.200 0.164 2.022 0.327 1.684 0.007
Utility $5.00 0.037 -0.106 0.164 1.987 0.327 1.422 0.035
Utility $3.00 -0.116 0.132 0.164 2.181 0.327 1.274 0.078
Utility 50% reduced 0.032 -0.169 0.164 1.001 0.327 1.211 0.106
Utility full 0.152 0.041 0.164 3.768 0.327 1.548 0.017
Utility triple -0.184 -0.205 0.164 1.121 0.327 0.981 0.291
N = 217
259
3.3 Sensory experiment data
Table A 2 Normality tests (sensory)
Kolmogorov
Skewness Kurtosis
Smirnov
Std. Std
Error Error
Self confidence 6.341 -0.489 0.150 -0.343 0.299 1.425 0.035
Ave importance COO 0.291 0.740 0.150 0.684 0.299 1.480 0.025
Ave importance price 0.425 0.097 0.150 -0.436 0.299 0.977 0.295
Ave importance acid 0.284 0.339 0.150 -0.390 0.299 1.139 0.049
Utility France 0.194 -0.335 0.150 0.464 0.299 1.243 0.091
Utility Chile -0.115 -0.081 0.150 0.788 0.299 1.045 0.225
Utility $53.00 0.943 -0.159 0.150 0.136 0.299 0.958 0.018
Utility $16.00 -0.0845 0.150 0.150 -0.099 0.299 1.238 0.003
Utility $6.00 -0.859 -0.325 0.150 0.527 0.299 0.835 0.488
Utility average 0.943 0.147 0.150 0.296 0.299 0.715 0.686
Utility high 0.042 -0.192 0.150 0.709 0.299 1.192 0.117
Objective brie 4.57 0.348 0.150 0.061 0.299 1.874 0.002
Subjective brie 4.453 0.143 0.150 -0.515 0.299 0.637 0.813
Ave importance COO 0.284 0.460 0.150 -0.277 0.299 1.127 0.158
Ave importance price 0.315 0.699 0.150 0.093 0.299 1.328 0.059
Ave importance fat 0.400 0.143 0.150 -0.825 0.299 0.939 0.341
Utility France 0.166 0.038 0.150 0.510 0.299 0.782 0.573
Utility Canada 0.019 0.017 0.150 0.359 0.299 0.839 0.482
Utility $69.95 0.075 0.074 0.150 0.576 0.299 0.698 0.714
Utility $49.00 0.262 -0.153 0.150 0.777 0.299 1.123 0.160
Utility $28.95.00 -0.337 -0.235 0.150 0.032 0.299 0.952 0.325
Utility full -0.013 -0.201 0.150 -0.174 0.299 1.048 0.222
Utility double 0.221 -0.106 0.150 0.181 0.299 0.810 0.528
Utility triple -0.208 -0.275 0.150 -0.334 0.299 1.464 0.028
N = 263
260
3.4 Conjoint survey data
Table A 3 Normality tests (survey)
Kolmogorov
Skewness Kurtosis
Smirnov
Std. Std
Error Error
Self confidence 6.206 -0.397 0.147 0.071 0.294 0.791 0.558
Ave importance price 0.553 -0.280 0.147 -0.786 0.294 1.472 0.026
Ave importance acid 0.220 0.972 0.147 0.785 0.294 1.789 0.003
Utility France 0.362 0.419 0.147 1.165 0.294 1.226 0.009
Utility Chile -0.160 -0.149 0.147 0.785 0.294 1.196 0.114
Utility $53.00 1.649 -0.054 0.147 -0.141 0.294 1.280 0.075
Utility $16.00 0.089 0.180 0.147 1.172 0.294 1.429 0.034
Utility $6.00 -0.738 -0.098 0.147 -0.182 0.294 1.009 0.261
Utility average 0.306 1.117 0.147 2.325 0.294 2.182 0.000
Utility high -0.340 -0.241 0.147 -0.093 0.294 1.306 0.066
Objective brie 4.47 0.129 0.147 -0.840 0.293 1.508 0.021
Subjective brie 4.006 0.224 0.147 -0.767 0.293 1.293 0.071
Ave importance price 0.372 0.492 0.147 -0.254 0.293 1.462 0.028
Ave importance fat 0.334 0.636 0.147 -0.228 0.293 1.833 0.002
Utility France 0.352 1.447 0.147 6.016 0.293 2.323 0.000
Utility Canada -0.120 0.719 0.147 10.302 0.293 1.854 0.002
Utility $69.95 0.505 0.499 0.147 1.938 0.293 1.373 0.046
Utility $49.00 -0.036 -0.354 0.147 2.111 0.293 1.649 0.009
Utility $28.95.00 -0.469 -0.235 0.147 0.654 0.293 0.976 0.296
Utility full -0.534 -0.854 0.147 2.250 0.293 2.188 0.000
Utility double 0.226 0.660 0.147 2.194 0.293 2.027 0.001
Utility triple 0.308 0.564 0.147 0.891 0.293 1.929 0.001
N = 274
261
Appendix
4 Scale reliability and validity
262
4.5 Reliability analysis for all scales
Scales are a set of correlated items that are designed to work together to measure
an underlying or latent variable. Reliability relates to the consistency with which
these items can be used to measure the same latent variable. Cronbachs Alpha, a
numerical coefficient of reliability, is used as a means of measuring scale reliability.
Calculation of the alpha is based on the degree of inter-correlation between scale
items; as inter-item correlation increases, alpha scores increase accordingly. Scores
range from 0 to 1, with a score of 0.7 the minimum reliability level acceptable in
social research (Malhotra et. al. 2002). Table A.6 illustrates scale Cronbach Alpha
scores are all in excess of 0.7 indicating sound reliability of all scales used for each
sample tested.
Table A 4 Reliability Coefficients of Scales

Reliability Coefficients all stage
N of Cronbachs
Scale N of items Stage of research
cases Alpha
Pilot survey 238 0.866
Subjective
knowledge 8 Sensory experiment 263 0.909
chardonnay
Conjoint survey 274 0.903

Subjective
knowledge 8 Sensory experiment 238 0.895
camembert/brie

Self 10 (pilot only)
Sensory experiment 263 0.716
Confidence 5
263
4.6 Construct validity
Factor analysis examines the relationships between a set of variables in order to
reveal the existence of an underlying, or latent, variable. One of the primary uses of
factor analysis is to reduce the total number of variables needed to measure a
construct through the identification of the important dimensions, or factors that
comprise it.
Eigenvalues (also called characteristic roots) are illustrated for factor measure
produced in the analysis. This value measures the variance in all the items that are
accounted for in that factor. If a factor has a low Eigenvalue, then it is contributing
little to the explanation of variances in the items. Factors with Eigenvalues of less
than 1 are discounted. Data rotation methods, such as Varimax, are employed in the
factor analysis process. The goal of rotation is to achieve a simpler factor solution by
maximizing variances between items and encouraging them to cluster together in a
manner that facilitates the interpretation (Malhotra et. al 2002). Through this process
any non-contributing items are identified and a broad understanding of the ability of
the items to measure the construct under investigation is achieved. Factor analysis
is a non-dependent procedure in that it does not assume that a dependent variable
is specified. Factor analysis results include a number of important statistics including
a correlation matrix, Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy,
Bartletts test of sphericity and the analyzed factor loadings. KMO is a statistic that
indicates the proportion of variance in scale items that is common to all,thus
indicating a latent variable. High values (close to 1.0) support the use of factor
analysis. Bartlett's test of sphericity indicates whether your correlation matrix is an
identity matrix, which would indicate that your variables are unrelated. The
significance level gives the result of the test. Very small values (less than 0.05)
264
indicate that there are probably significant relationships among your variables
(Coakes, Steed et al. 2006).
A summary of cumulative percentages of variance explained by factors with
Eigenvalues greater than 1 for each application of the scales employed is illustrated
below in table A.7.
Table A 5 Construct validity summary for all scales

Cum % of variance explained all stage
Cum % of variance No. factors
Scale Stage of research
explained Eigenvalues > 1
Pilot survey 69.6 2
Subjective
knowledge Sensory experiment 61.7 1
chardonnay
Conjoint survey 59.2 1
Pilot survey 70.4 2

Subjective
knowledge Sensory experiment 63.4 1
camembert/brie
Pilot survey 47.2 2

Self
Sensory experiment 47.8 1
Confidence
4.6.1 Assumptions
Normality:
Normality is preferred but not essential; however, if distribution is found to be normal
the validity of the solution is improved (Coakes, Steed et al. 2006). Normality testing
has revealed that the data collected is not normal in all instances and results must be
considered with this potential limitation.
265
Sample size:
A minimum of 5 respondents per variable is required, with sample in excess of 100
preferred; this is satisfied with every sample analyzed in this research (Coakes,
Steed et al. 2006).
Outliers:
Factor analysis solutions are sensitive to outliers (Coakes, Steed et al. 2006). At
each stage of the research, outliers in cases were evaluated, and where appropriate
deleted. However, outliers were not common and only removed where a suggestion
of respondent error resulted in an atypical response (Malhotra, Hall et al. 2002).
Multicollinearity and singularity:
This assumption is not relevant for principal component extraction (Coakes, Steed et
al. 2006).
Linearity:
Factor analysis solutions are based on correlation; therefore linearity is important
when assessing solutions (Coakes, Steed et al. 2006). This assumption is satisfied
due to the significant correlations existing between the majorities of scale items.
Selection bias:
Social researchers often use factor analysis to explore data in order to reveal
previously undetermined structures and underlying variables. Including items only
known to be relevant and eliminating items that dont load cleanly on specific factors
in order to reduce factors and achieve a tidier solution can result in interpretation
error. Factor analysis solutions require the application of common sense and
compromise in interpretation; an acknowledged challenge and limitation in using this
tool (Malhotra, Hall et al. 2002). In this research, the scales used were tested in
previous studies and are based on established research methodology and
application. Further each scale was used 3 times, each involving samples in excess
of 200 respondents, with assumption testing and validation completed each time.
Based on this, the assumption for selection bias has been met.
266
Factorability of the correlation matrix:
To be appropriate for factor analysis, the correlation matrices will have many sizable
and significant correlations (in excess of 0.3). Items found to have weak or
insignificant correlations will result in a high number of factors making small
contributions to the overall variance explained (Malhotra, Hall et al. 2002; Coakes,
Steed et al. 2006). These results provide little or no insight in the measurement (or
existence) of an underlying variable. However, intercorrelations that are extremely
high between all items can suggest a multicollinearity problem requiring the
combination or elimination of some items (Coakes, Steed et al. 2006), Correlations
between scale items were measured using the non-parametric measurement
Spearmans r. Bartletts tes of sphericity and the KMO measure of sampling
adequacy are used to determine the overall factorability of the matrix; where if the
Barltetts test statistic is large and significant, and the KMO measure is greater than
0.6 then factorability is assumed (Coakes, Steed et al. 2006)
Results confirm that significant (<0.05), and therefore satisfactory, levels of
correlation exist between the majority of items. The results of the correlation analysis
for each scale, at each stage, and associated Bartletts tests of sphericity and the
KMO measure of sampling adequacy, are illustrated in the tables that follow.
267
4.6.2 Subjective knowledge scale test results
(Chardonnay)
Pilot study:
Table A 7 (Correlation matrix) shows that all items tested as a scale to measure self
confidence correlate positively, with all significant with 95% confidence or higher.
Significant correlations range from 0.260 (positive but weak) to 0.838 (positive and
strong). These high correlations, supported with a significant KMO statistic of 0.855
(Table A 6), demonstrate the data is factorable.
Table A 6 Factorability of subjective knowledge scale chardonnay (pilot)
KMO and Bartlett's Test -Subjective knowledge

chardonnay (pilot)
Kaiser-Meyer-Olkin Measure of Sampling Adequacy.
.855
Bartlett's Test of Approx. Chi-Square 914

Sphericity df 28
Sig. .000

chardonnay (pilot)
Correlation Matrix - Subjective knowledge chardonnay (pilot)
Judge quality Knowledgeable Know most Don't know
Know about Expert on Know less
of about Chardonnay alot about
Wine Chardonnay about Wine
Chardonnay Chardonnay in shops Chardonnay
1. Know about Wine 1.000
2. Judge quality of Chardonnay .838 1.000
3. Knowledgeable about Chardonnay .575 .541 1.000
4. Expert on Chardonnay .486 .465 .275 1.000
5. Know less about Wine .358 .339 .575 .139 1.000
6. Know most Chardonnay in shops .589 .509 .329 .418 .210 1.000
7. Don't know alot about Chardonnay .568 .550 .669 .211 .610 .392 1.000
8. Can tell price of Chardonnay .591 .590 .347 .486 .260 .484 .320
Sig. 2. Judge quality of Chardonnay .000
3. Knowledgeable about Chardonnay .000 .000
4. Expert on Chardonnay .000 .000 .000
5. Know less about Wine .000 .000 .000 .020
6. Know most Chardonnay in shops .000 .000 .000 .000 .001
7. Don't know alot about Chardonnay .000 .000 .000 .001 .000 .000
268
Sensory experiment:
Significant correlations range from 0.402 (positive and moderate) to 0.800 (positive
and strong). These high correlations, supported with a significant KMO statistic of
0.912 (Table A 8), demonstrate the data is factorable.
Table A 8 Factorability of subjective knowledge scale chardonnay (sensory)
KMO and Bartlett's Test - Subjective Knowledge

Chardonnay (lab)
Kaiser-Meyer-Olkin Measure of Sampling
Adequacy. .912
Bartlett's Test of Approx. Chi-Square 1261.124

Sphericity df 28
Sig. .000

chardonnay (sensory)
Correlation Matrix - Subjective Knowledge Chardonnay (lab)
Variables
Judge Knowledgeable Know less Know most Don't know
Know about Expert on
quality of about about Chardonnay alot about
Wine Chardonnay
Chardonnay Chardonnay Wine in shops Chardonnay
2. Judge quality of Chardonnay .000
Sig. 5. Know less about Wine .000 .000 .000 .000
269
Conjoint survey:
Table A.11 (Correlation matrix) shows that all items tested as a scale to measure self
and strong). These high correlations, supported with a significant KMO statistic of
0.903 (Table A 10), demonstrate the data is factorable.
Table A 10 Factorability of subjective knowledge scale chardonnay (survey)
KMO and Bartlett's Test - Subjective Knowledge

Chardonnay (survey)
Adequacy. .903

Sphericity df 28
Sig. .000

chardonnay (survey)
Correlation Matrix - Subjective knowledge chardonnay (survey)
Judge quality Knowledgeable Know most Don't know
Know about Expert on Know less
of about Chardonnay alot about
Wine Chardonnay about Wine
Chardonnay Chardonnay in shops Chardonnay
2. Judge quality of Chardonnay .000
5. Know less about Wine .000 .000 .000 .000
270
4.6.3 Subjective knowledge scale test results
(camembert/brie)
Pilot Study:
strong). These high correlations, supported with a KMO statistic of 0.849 (Table A
12), demonstrate the data is factorable.
Table A 12 Factorability of Subjective knowledge scale camembert (pilot)

KMO and Bartlett's Test - Subjective knowledge
camembert (pilot)
.849
Sphericity df 28
Sig. .000
Table A 13 Spearmans r correlation coefficients Subjective knowledge scale brie

(survey)
Correlation Matrix - Subjective knowledge camembert (pilot)
Know Judge Know less Know most Don't know
Know about Expert on
about quality of about Camembert alot about
Camembert Camembert
Cheese Camembert Camembert in shops Camembert
1. Know about Cheese 1.000
2.Judge quality of Camembert .827 1.000
3. Knowledgeable about Camembert .533 .524 1.000
4, Expert on Camembert .514 .538 .273 1.000
5. Know less about Camembert .418 .457 .556 .201 1.000
6. Know most Camembert in shops .500 .550 .232 .504 .175 1.000
7. Don't know alot about Camembert .520 .541 .649 .242 .625 .354 1.000
8. Can tell price of Camembert .562 .669 .327 .473 .216 .525 .326
Sig. 2.Judge quality of Camembert .000
3. Knowledgeable about Camembert .000 .000
4, Expert on Camembert .000 .000 .000
5. Know less about Camembert .000 .000 .000 .001
6. Know most Camembert in shops .000 .000 .000 .000 .005
7. Don't know alot about Camembert .000 .000 .000 .000 .000 .000
8. Can tell price of Camembert .000 .000 .000 .000 .001 .000 .000
271
Sensory experiment:
strong). These high correlations, supported with a KMO statistic of 0.899 (Table A
14), demonstrate the data is factorable.
Table A 14 Factorability of Subjective knowledge scale brie (sensory)

KMO and Bartlett's Test - Subjective Knowledge Brie
(lab)
Adequacy. .899

Sphericity df 28
Sig. .000

(sensory)
Correlation Matrix - Subjective Knowledge Brie (lab)
Know Judge Knowledge Know most Don't
Expert Know less
about quality of able about Bries in know alot
on Brie about Brie
Cheese Brie Brie shops about Brie
2. Judge quality of Brie .846 1.000
3. Knowledgeable about Brie .660 .656 1.000
4. Expert on Brie .675 .665 .605 1.000
5. Know less about Brie .545 .518 .592 .492 1.000
6. Know most Bries in shops .655 .660 .543 .660 .435 1.000
7. Don't know alot about Brie .595 .563 .694 .566 .591 .563 1.000
8. Can tell price of Brie .537 .620 .399 .501 .439 .541 .342
2. Judge quality of Brie .000
3. Knowledgeable about Brie .000 .000
4. Expert on Brie .000 .000 .000
Sig. 5. Know less about Brie .000 .000 .000 .000
6. Know most Bries in shops .000 .000 .000 .000 .000
7. Don't know alot about Brie .000 .000 .000 .000 .000 .000
8. Can tell price of Brie .000 .000 .000 .000 .000 .000 .000
272
Conjoint survey:
and strong). These high correlations, supported with a KMO statistic of 0.926. (Table
A 16), demonstrate the data is factorable.
Table A 16 Factorability of Subjective knowledge scale brie (survey)
KMO and Bartlett's Test - Subjective Knowledge Brie (survey)

Adequacy. .926

Sphericity df 28
Sig. .000

(survey)
Correlation Matrix - Subjective knowledge brie (survey)
Judge Know most Don't know
Know about Knowledge Expert on Know less
quality of of the brie in alot about
Cheese about brie brie about brie
brie shops brie
2. Judge quality of brie .889 1.000
3. Knowledgeable about brie .714 .704 1.000
4. Expert on brie .682 .697 .607 1.000
5. Know less about brie .577 .552 .510 .445 1.000
6. Know most of the brie in shops .641 .670 .572 .581 .435 1.000
7. Don't know alot about brie .729 .723 .747 .580 .548 .658 1.000
8. Can tell price of brie .746 .776 .608 .648 .499 .649 .609
2. Judge quality of brie .000
3. Knowledgeable about brie .000 .000
4. Expert on brie .000 .000 .000
5. Know less about brie .000 .000 .000 .000
6. Know most of the brie in shops .000 .000 .000 .000 .000
7. Don't know alot about brie .000 .000 .000 .000 .000 .000
8. Can tell price of brie .000 .000 .000 .000 .000 .000 .000
273
4.6.4 Self-confidence scale
Pilot study:
Table A 19 shows that the majority of items tested as a scale to measure self
confidence correlate positively, with the majority significant with 95% confidence or
higher. Significant correlations range from 0.114 (positive but weak) to 0.573
(positive and moderate). These correlations although low to moderate, supported
with a KMO statistic of 0.748 (Table A 18) demonstrate the data is factorable.
Table A 18 Factorability of self-confidence scale (pilot)
KMO and Bartlett's Test - Self confidence (pilot)

Adequacy. .748

Sphericity df 45
Sig. .000
Table A 19 Spearmans r correlation coefficients Self-confidence scale (pilot)

Correlation Matrix - Self confidence (pilot)
What
Handling Opinions Favourable Shyness Wear the
Low Enter a people Laughed
myself are first and wrong
opinion room think of at
socially inferior impression inferiority thing
me
1. Handling myself socially 1.000
2. Low opinion .351 1.000
3. Enter a room .573 .314 1.000
4. Opinions are inferior (rev) .256 .102 .291 1.000
5. Favourable first impression (rev) .200 .012 .164 .389 1.000
6. Shyness and inferiority (rev) .216 .114 .193 .391 .496 1.000
7. Wear the wrong thing (rev) .117 .032 .221 .327 .174 .400 1.000
8. What people think of me .318 .159 .253 .160 .125 .145 .123 1.000
9. Laughed at (rev) .143 -.033 .071 .305 .295 .336 .179 -.099 1.000
10. Loss for words .353 .200 .310 .201 .017 .142 .120 .334 -.029
Sig. 2. Low opinion .000
3. Enter a room .000 .000
4. Opinions are inferior (rev) .000 .059 .000
5. Favourable first impression (rev) .001 .426 .006 .000
6. Shyness and inferiority (rev) .000 .040 .001 .000 .000
7. Wear the wrong thing (rev) .036 .311 .000 .000 .004 .000
8. What people think of me .000 .007 .000 .007 .027 .013 .029
9. Laughed at (rev) .014 .308 .137 .000 .000 .000 .003 .064
10. Loss for words .000 .001 .000 .001 .398 .014 .032 .000 .327
274
Sensory experiment:
Table A 21 shows that many of items tested as a scale to measure self confidence
correlate positively, with the most significant with 95% confidence or higher.
moderate). These correlations were found to be low to moderate, however,
supported with a significant KMO statistic of 0.752 (Table A 20) demonstrates the
data is factorable. Whilst these results are satisfactory, the decision was taken
(discussion in chapter 5) to eliminate the 5 positively phrased items and retain the
negatively phrased items (Tables A 22 and A 23).
Table A 20 Factorability of self-confidence scale (sensory)

KMO and Bartlett's Test - Self Confidence (lab)
Adequacy. .752

Sphericity df 45
Sig. .000
Table A 21 Spearmans r correlation coefficients Self-confidence scale (sensory)

Correlation Matrix - Self Confidence (lab)
Handling Opinions Favourable Shyness Wear the What
Low Enter a Laughed
myself are first and wrong people
opinion room at
socially inferior impression inferiority thing think of me
1. Handling myself socially 1.000
2. Low opinion .299 1.000
3. Enter a room .489 .327 1.000
4. Opinions are inferior .177 .054 .173 1.000
5. Favourable first impression .233 .085 .108 .446 1.000
6. Shyness and inferiority .263 .135 .400 .390 .417 1.000
7. Wear the wrong thing .102 .096 .197 .321 .146 .324 1.000
8. What people think of me .292 .099 .270 .168 .118 .186 .132 1.000
9. Laughed at .102 .131 .120 .385 .377 .330 .233 .005 1.000
10. Loss for words .230 .173 .241 .094 .184 .278 .083 .266 -.066
2. Low opinion .000
3. Enter a room .000 .000
4. Opinions are inferior .002 .194 .003
5. Favourable first impression .000 .088 .041 .000
Sig. 6. Shyness and inferiority .000 .015 .000 .000 .000
7. Wear the wrong thing .051 .062 .001 .000 .009 .000
8. What people think of me .000 .057 .000 .003 .029 .001 .017
9. Laughed at .051 .018 .027 .000 .000 .000 .000 .468
10. Loss for words .000 .003 .000 .067 .001 .000 .093 .000 .146
275
Test results with only 5 negatively phrased items produced a 1 factor solution with all
intercorrelations significant, ranging from 0.157 (low and positive) to 0.451 (moderate
and positive) as seen in table A 23. These finding are supported with a significant
KMO statistic of 0.763 (Table A 22) demonstrating the data is factorable.
Table A 22 Factorability of reduced self confidence scale (sensory)
KMO and Bartlett's Test - Self confidence negative items only

(lab)
.763

Sphericity df 10
Sig. .000

(sensory)
Correlation Matrix - Self confidence negative items only (lab)

Favourable
Opinions are Shyness and Wear the
first
inferior inferiority wrong thing
impression
4. Opinions are inferior 1.000
5. Favourable first impression .451 1.000
6. Shyness and inferiority .400 .425 1.000
7. Wear the wrong thing .329 .157 .330 1.000
9. Laughed at .389 .373 .333 .240
Sig. 5. Favourable first impression .000
6. Shyness and inferiority .000 .000
7. Wear the wrong thing .000 .006 .000
9. Laughed at .000 .000 .000 .000
276
Conjoint survey:
Significant correlations range from 0.213 (positive and low) to 0.541 (positive and
moderate). These high correlations, supported with a KMO statistic of 0.795. (Table
A 24), demonstrate the data is factorable.
Table A 24 Factorability of reduced self confidence scale (survey)
KMO and Bartlett's Test - Self confidence negative items only

(conjoint survey)
.795
Sphericity df 10
Sig. .000

(survey)
Correlation matrix - Self confidence negative items only (survey)
Favourable Wear the
Opinions are Shyness and
first wrong
inferior inferiority
impression thing
4. Opinions are inferior 1.000
5. Favourable first impression .491 1.000
6. Shyness and inferiority .541 .479 1.000
7. Wear the wrong thing .255 .213 .378 1.000
9. Laughed at .377 .379 .451 .277
5. Favourable first impression .000
6. Shyness and inferiority .000 .000
7. Wear the wrong thing .000 .000 .000
9. Laughed at .000 .000 .000 .000
277
4.6.5 Discriminant validity
Discriminant validity is necessary to confirm that the scales used are measuring
different constructs and not simply duplicating the measurement of the same
variable. Convergent validity is the degree to which each item in a scale is related to
all others in the same scale. To test for discriminant validity, factor analysis was
used to test the combined items of the subjective knowledge (for both products) and
self confidence scales for each stage of the research.
Discriminant validity (pilot):
Results confirmed no significant and interpretable relationships between any items of
the subjective knowledge and self confidence items in spite of a degree of cross
loading across scale item sets. The outcome is not surprising given that, for this
sample, the sample each scale produced a 2 factor rotated solution. However, the
results were deemed acceptable on balance, for the purposes of the pilot analysis.
Table A 26 Discriminant validity chardonnay and self confidence (pilot)
Rotated Component Matrix Subjective knowledge chardonnay and self confidence (pilot) a
Component
1 2 3 4 5
Handling myself socially .057 .192 .037 .783 .234
Low opinion .058 -.045 .073 .725 .023
Enter a room .115 .193 -.012 .721 .238
Opinions are inferior .040 .678 .138 .113 .185
Favourable first impression -.016 .707 -.010 .051 -.018
Shyness and inferiority .066 .775 .110 .088 .061
Wear the wrong thing -.008 .575 -.048 -.015 .168
What people think of me -.067 .084 -.016 .183 .744
Laughed at -.036 .617 .008 .128 -.399
Loss for words .029 .119 -.024 .263 .718
Know about Wine .782 -.008 .406 .185 -.033
Judge quality of Chardonnay .759 .040 .397 .130 -.025
Knowledgeable about Chardonnay .335 .053 .788 .037 .020
Expert on Chardonnay .767 .181 -.012 -.175 .246
Know less about Wine .088 .066 .841 .017 -.028
Know most Chardonnay in shops .725 -.113 .164 .076 -.033
Don't know alot about Chardonnay .306 .048 .826 .046 -.033
Can tell price of Chardonnay .788 .007 .111 .114 -.133
Extraction Method: Principal Component Analysis.
Rotation Method: Varimax with Kaiser Normalization.
a. Rotation converged in 6 iterations.
278
Table A 27 Discriminant validity brie and self confidence (pilot)
Rotated Component Matrix Subjective knowledge camembert and self confidence (pilot) a
Component
1 2 3 4 5
Handling myself socially .058 .035 .184 .796 .234
Low opinion .075 -.039 -.032 .740 -.001
Enter a room .003 .101 .175 .706 .277
Opinions are inferior .002 .119 .667 .130 .191
Favourable first impression -.007 -.115 .716 .087 -.043
Shyness and inferiority -.024 .087 .777 .096 .079
Wear the wrong thing -.005 .063 .564 -.081 .262
What people think of me .027 -.125 .077 .168 .743
Laughed at .016 .041 .635 .084 -.341
Loss for words .002 .071 .089 .253 .732
Know about Cheese .717 .467 .051 .082 -.075
Judge quality of Camembert .774 .459 .073 .055 .008
Knowledgeable about Camembert .251 .803 .053 .011 -.054
Expert on Camembert .777 .071 -.030 -.105 .172
Know less about Camembert .108 .841 .063 -.004 .011
Know most Camembert in shops .787 .075 -.081 .022 -.024
Don't know alot about Camembert .274 .813 .027 .058 -.001
Can tell price of Camembert .788 .135 .018 .185 -.065
Discriminant validity (sensory):
Following the decision to use the illustrated 5 reverse coded items as a measure of
self confidence, no significant cross loading across scale item sets is noted.
Table A 28 Discriminant validity chardonnay and self confidence (sensory)
a
Rotated Component Matrix Subjective knowlede chardonnay and self confidence (sensory)
Component
1 2
In group discussions usually feel my opinions are inferior -.007 .762
I don't make a very favourable first impression -.068 .719
When confronted by strangers, my first reaction is shyness and inferiority -.210 .710
It's extremely uncomfortable to go to a party wearing the wrong thing -.055 .536
When in a group, I rarely express an opinion for fear of being laughed at .097 .692
Know about Wine .897 -.056
Judge quality of Chardonnay .826 -.122
Knowledgeable about Chardonnay (rev) .776 .021
Expert on Chardonnay .786 .024
Know less about Wine (rev) .690 .117
Know most Chardonnay in shops .720 -.012
Don't know alot about Chardonnay (rev) .875 -.015
Can tell price of Chardonnay .674 -.110
279
Table A 29 Discriminant validity brie and self confidence (sensory)
a
Rotated Component Matrix Subjective knowledge brie and self confidence (sensory)
Component
1 2
In group discussions usually feel my opinions are inferior .001 .760
I don't make a very favourable first impression .063 .718
When confronted by strangers, my first reaction is shyness and inferiority -.121 .724
It's extremely uncomfortable to go to a party wearing the wrong thing -.053 .556
Know about Cheese .877 -.035
Judge quality of Camembert .876 -.034
Knowledgeable about Camembert (rev) .816 -.020
Expert on Camembert .818 .017
Know less about Camembert (rev) .715 .025
Know most Camembert in shops .796 .020
Don't know alot about Camembert (rev) .776 .017
Can tell price of Camembert .672 -.078
Discriminant validity (survey):
No significant cross loading across scale item sets is noted.
Table A 30 Discriminant validity chardonnay and self confidence (survey)

a
Rotated Component Matrix Subjective knowledge chardonnay and self confidence (survey)
Component
1 2
When confronted by strangers, my first reaction is shyness and inferiority .070 .816
It's extremely uncomfortable to go to a party wearing the wrong thing .063 .540
Know about Wine .843 .218
Judge quality of Chardonnay .850 .142
Knowledgeable about Chardonnay (rev) .805 .128
Expert on Chardonnay .776 -.096
Know less about Wine (rev) .638 .103
Know most Chardonnay in shops .725 -.055
Don't know alot about Chardonnay (rev) .807 .129
Can tell price of Chardonnay .643 .038
280
Table A 31 Discriminant validity brie and self confidence (survey)
a
Rotated Component Matrix Subjective knowledge brie and self confidence (survey)
Component
1 2
When confronted by strangers, my first reaction is shyness and inferiority .078 .814
It's extremely uncomfortable to go to a party wearing the wrong thing .101 .530
When in a group, I rarely express an opinion for fear of being laughed at -.039 .705
Know about Cheese .906 .092
Judge quality of Camembert .919 .024
Knowledgeable about Camembert (rev) .820 .129
Expert on Camembert .794 .042
Know less about Camembert (rev) .676 .051
Know most Camembert in shops .775 .145
Don't know alot about Camembert (rev) .841 .109
Can tell price of Camembert .842 .072
281
Appendix
5 Information pack and registration form
282
School of Commerce - Research Project
Food Product Preferences
Hello!
I am a student at the University of Adelaide, working on a
research project about preferences for wine and cheese products and
I would like to invite you to come and taste some Wine and Cheese!
During the months of April and May I will be inviting consumers to visit our tasting rooms and offer
their opinions regarding provided samples of both wine and cheese. For participating in this
research project, you will be provided with a $30 cash payment as a Thank You for giving us your
valuable time.
You certainly dont need to be a wine expert to come along; anyone that enjoy a glass of wine or a
bite of cheese is encouraged to register. There is no right or wrong in these taste tests, we are only
interested in your genuine opinions about the products you will try. All wine and cheese samples are
provided for tasting at no cost to you, and the tasting will take about two and a half hours in total.

Were doing some basic research to gain a better understanding of consumer wine
and cheese preferences. While this study is pretty simple, ultimately its hoped that
the information you provide will be used to produce better products for the market
place.
Risks to you?
None. (Except for perhaps, a mild risk of wine or cheese cravings!)
Your confidentiality?
Your confidentiality is assured. We only use information for statistical analysis and we
only keep your details to schedule the visits for tasting. We do not pass information on
to any other parties and once the study is completed this information will be deposed
of.
How do you participate in the tastings and receive my $30?

Please complete the form on the back of this letter and return it to me in the reply paid envelope
provided. Please make sure to indicate you first and second preferences in terms of times and
dates illustrated. The tastings will be conducted at the University of Adelaide, Waite Campus, Waite
Rd, Urrbrae. There is plenty of free parking and youll be given clear directions to the facilities.

Contact Roberta Veale on 0404833924 or Roberta.Veale@tafesasouth.org, or Professor Pascale
Quester on pascale.quester@adelaide.edu.au to receive further information about this project.
Participation in this study is open only to individuals over the age of 18 years.
This project is funded by a research grant from the Wine and Grape Research and Development Corporation.
283
Tasting Registration Form
Thank you very much for registering to come along to

taste the wines and cheeses!
Please complete the form below, providing contact details and ticking (3) the
box that indicates two times that would best suit you, and then please (3) to
indicate if this is your first or second preference. Return the form by post
using the enclosed reply paid envelope. Once registered, I will contact you by
post or email, confirming your attendance time and providing you with a map
of the Waite Campus and directions to get there. Its just off Cross Road,
Urrbrae and is very easy to find and access.
There are four possible dates available in April 2006, with three sessions per
day. In May, there are five as there is one Saturday included with an extra
session commencing at noon available on that day. Participation in this study
is only open to individuals over the age of 18 years.
-----------------------------------------------------------------------------------
Name: Contact Phone Email
Postal Address: Post Code
Please choose times that best suit you. You will be registered for only
one of your choices; your first choice will be given whenever possible.
1st 2nd
April 2006 Choice Choice
Thursday 20th 2 - 4 pm 4 - 6 pm 6 - 8 pm
Friday 21 st 2 - 4 pm 4 - 6 pm 6 - 8 pm
Friday 28 th 2 - 4 pm 4 - 6 pm 6 - 8 pm
1st 2nd
May 2006 Choice Choice
Friday 5 th 2 - 4 pm 4 - 6 pm 6 - 8 pm
Saturday 6th 12 - 2 pm 2 - 4 pm 4 - 6 pm 6 - 8 pm
Friday 12 th 2 - 4 pm 4 - 6 pm 6 - 8 pm
284
Appendix
6 Tasting registration spreadsheet and
respondent confirmation notice
285
6.1 Spreadsheet for registrations
Figure A 1 Session registrations and time table
286
6.2 Confirmation notice
Name: _______________
School of Commerce - Research Project

Food Product Preferences
This is to confirm your scheduled tasting session at the
Waite Campus of the University of Adelaide on the
12th of May from 2 to 4 pm.

Thank you for registering to participate in my research!
I have attached a copy of a PDF file showing the Waite Campus and
parking facilities. You can access the campus taking Waite Road
from Cross Road, Urrbrae. When you open the PDF file, youll see
Ive made some notes for you to help you find us, and highlighted
the map reference of D5 on the index page showing the reference
for Wine and Horticulture Reception (Building 30).
Once on Waite Road veer left at the roundabout taking Hartley

Grove Road, on your right you will see Paratoo Road Carpark (feel
free to park here). The gate leading to the facility is gate 2b (see
notes). There is some limited parking on this entrance you may
find a spot, or they may all be in use. Any queries or questions,
please feel free to email me or call me on 0404833924. An
incentive payment of $30 is paid to all participants.
See you then!
Regards, Roberta
If for any reason you cannot come

please ring me on 0404833924!
287
Appendix
7 Sensory experiment questionnaire
288
School of Commerce
Welcome to this Tasting Session
Judge No:
Date:
Time:
Important Information about participation in this study
Your participation in this research is entirely voluntary and you may withdraw at any time.
The first is a briefing regarding the procedures for the session and information regarding the
products that you will be trying and assessing.
During the course of the next 2 to 2.5 hours you will be presented with 11 samples of wine and 11
samples of cheese to taste.
You are required to smell and taste each wine sample, then importantly, you are instructed to
expectorate (spit) the sample into the sink provided in the tasting booths.
The cheese samples may be consumed if you wish, or disposed of using the tissues and
receptacles provided in the tasting booths.
Prior to the commencement of the tasting session you will be asked to complete a short
questionnaire, and post tasting you will be asked to complete another. These should not take
more than 10 to 15 minutes to complete in total.
At no time should you go back and review and change judgements made or answers given. Its
critically important to keep moving forward throughout the session.
Page 1
289
Consent Form
For participants in a research project
I.(please print name) consent to take part in the

research project described.
I have read the information provided on page 1 of this booklet.
I have had the project, so far as it affects me, fully explained to my satisfaction by the research
worker and my consent to participate is given freely.
I understand that I am free to withdraw from the project at any time.
Signed: . Date: ..
Witness
I have described to ..(name of participant) the nature of

the procedures to be carried out. In my opinion he/she understood my explanation.
Status in Project: PhD Student Name: Roberta Veale
Signed:..
Page 2
Part 1 (a)
yourself. Your answers will only appear in aggregate and average numbers.
Please (3) the box that best reflects your agreement Strongly Strongly
with the following statements. Disagree Agree
1 2 3 4 5 6 7 8 9

3.
5. I dont make a very favourable first impression on people.
6.
7.
wrong thing.
9.
laughed at.
10.
know.
Please turn the page and continue with Part 1 (b) Page 3
290
Part 1 (b)
In this part of the questionnaire, we would like to ask you some questions about your general knowledge
and experiences with purchasing Chardonnay.
Please (3) the box that best reflects your knowledge Strongly Strongly
and experience. Disagree Agree
1 2 3 4 5 6 7 8 9
average)
Please turn the page and continue with Part 1 (c) Page 4
Part 1 (c)
In this part of the survey, we would like to ask you some questions about your general knowledge
and experiences with purchasing Brie.
and experience about Brie. Disagree Agree
1 2 3 4 5 6 7 8 9
1. I feel confident about my knowledge of Brie.
2. I feel that I know how to judge the quality of Brie.
3. I do not feel very knowledgeable about Brie.
4. Among my circle of friends, Im one of the experts on Brie.
5. Compared to most buyers, I know less about Brie.
6. I know of most of the Brie cheeses in the shops.
7. When it comes to Brie, I really dont know a lot.
8. I can tell if a Brie is worth the price or not.
How often would you purchase Brie for yourself If you have a favourite brand of Brie, please list
9. 10.
or others? (on average) it below.
This is the end of part 1 Please do not turn the page until instructed to do so (thanks!)
Page 5
291
Part 2
Chardonnay and Cheese Samples
You will be given 11 samples of chardonnay and 11 samples of brie to taste.
These will be given to you 1 sample at a time, in a mixed order.
Each sample profile and rating scale is on a separate page of this booklet each with a unique and
specific identifying number. Check to see the sample number matches the number in your booklet.
Give each sample a rating on the scale where 1 = Low Quality and 10 = High Quality. Then please
indicate whether or not you would consider buying this product.
Rate each sample individually as you receive them. While you may not be a regular Chardonnay
drinker or Brie fancier (or even particularly like Chardonnay or Brie) assess each profile in light of the
need to make a possible purchase. All Wine options are 2005 vintage.
Between each sample, be sure to cleanse your palate with water and/or a bite of water biscuit.
Remember to expectorate the wine sample in the sink provided.
DONT GO BACK! Once you have assessed a sample, use the indicator switch in the booth to request
your next sample and then turn to the page to rate the next corresponding sample number.

Page 6
Chardonnay 253
Produced in France
Retail Price $16.00
1 2 3 4 5 6 7 8 9 10

Page 7
292
Chardonnay 582
Retail Price $6.00
1 2 3 4 5 6 7 8 9 10

Page 8
Chardonnay 481
Produced in Chile
Retail Price $16.00
1 2 3 4 5 6 7 8 9 10

Page 9
293
Brie 810
Produced in France
1 2 3 4 5 6 7 8 9 10

Page 10
Brie 139
1 2 3 4 5 6 7 8 9 10

Page 11
294
Brie 367
Produced in Canada
Retail Price $69.95
1 2 3 4 5 6 7 8 9 10

Page 12
Chardonnay 696
Produced in France
Retail Price $6.00
1 2 3 4 5 6 7 8 9 10

Page 13
295
Chardonnay 595
Retail Price $16.00
1 2 3 4 5 6 7 8 9 10

Page 14
Chardonnay 924
Produced in Chile
Retail Price $53.00
1 2 3 4 5 6 7 8 9 10

Page 15
296
Brie 266
Produced in Canada
1 2 3 4 5 6 7 8 9 10

Page 16
Brie 709
Produced in France
Retail Price $49.95
1 2 3 4 5 6 7 8 9 10

Page 17
297
Brie 380
Retail Price $69.95
1 2 3 4 5 6 7 8 9 10

Page 18
Chardonnay 152
Retail Price $6.00
1 2 3 4 5 6 7 8 9 10

Page 19
298
Chardonnay 823
Produced in France
Retail Price $53.00
1 2 3 4 5 6 7 8 9 10

Page 20
Chardonnay 494
Produced in Chile
Retail Price $6.00
1 2 3 4 5 6 7 8 9 10

Page 21
299
Brie 735
Produced in France
1 2 3 4 5 6 7 8 9 10

Page 22
Brie 393
1 2 3 4 5 6 7 8 9 10

Page 23
300
Brie 621
Produced in France
Retail Price $69.95
1 2 3 4 5 6 7 8 9 10

Page24
Chardonnay 950
Produced in France
Retail Price $16.00
1 2 3 4 5 6 7 8 9 10

Page 25
301
Chardonnay 279
Retail Price $53.00
1 2 3 4 5 6 7 8 9 10

Page 26
Brie 178
Produced in Canada
Retail Price $49.95
1 2 3 4 5 6 7 8 9 10

Page 27
302
Brie 507
Retail Price $49.95
1 2 3 4 5 6 7 8 9 10

That concludes the taste testing.

Please close your booklet at this time.
Page 28

In this part of the survey, we would like to ask you some questions relating to your general knowledge
about wine. Please (3) the box that best reflects your knowledge and experience. If youre unsure about


Please turn the page and continue Page 29
303
Part 3 (a) Chardonnay cont.
7. Chardonnay, typically has an aging potential 8. Terms often linked with the taste of Chardonnay
of: are:
crushed in the year named, if a Vintage date is combinations, it is important to remember that:
given?
Please turn the page and continue

Page 30

wines:
Color 1 Improve the taste of a poor wine 1
Part 3 (b) Brie

Please (3) the box that best reflects your knowledge and experience with Brie. If youre unsure about the
answer to any particular question, thats no problem just (3) the box indicating this.

3. Rennet is: 4. Brie:

A form of yeast used to make cheese. 1 Is eaten younger than Camembert 1
An enzyme extracted from stomach linings. 2 Is eaten older than Camembert 2
Neither of the above. 3 Matures in the same time as Camembert 3
Page 31
304
Part 3 (b) Brie cont.
5. Rind is: 6. Brie is nicest when made from milk produced:
An important influence on flavor development 1 In the spring and autumn 1
Not found on Brie 2 In the winter and summer 2
Never washed in making quality cheeses 3 In particularly wet summers 3
7. Brie: 8. Brie:
Never smells of mushrooms and yeast 1 Has slightly thicker mould than Camembert 1
Usually smells of mushrooms and yeast 2 Has slightly thinner mould than Camembert 2
Can sometimes smell like old socks 3 Has identical mould to Camembert 3
9. Triple Cream Brie: 10. Brie rind is:

Is made from milk with extra cream added later 2 Comprised of grey or white mould 2
Doesnt exist as a product 3 Sometimes orange 3
Page 32
Part 3 (b) Brie cont.
11. When choosing wine to drink with Brie: 12. Brie cheese:
Its critical to make sure its well chilled 1 Is a product where price usually influences quality 1
A chardonnay will suit, but never a red 2 Is a product where price is no indication of quality 2
Is a product where all brands cost pretty much the
A full bodied, earthy red is a sound choice 3 3
same, no matter where the cheese comes from
13. The inside of a Brie that is ready to 14. When you press the rind of a Brie
eat is: it should:
Page 33
305
Part 4
For the purpose of analysis, we would be grateful if you would provide some personal information
about yourself. Your anonymity is guaranteed, this data is only used for statistical purposes.

46 yrs to 55 yrs 4
Over 55 yrs 5
(gross) completed:
Thank you very much!

Page 34
306
Appendix
8 Correlation matrices (chardonnay)
307
8.1 Sensory experiment
Table A 32 High objective knowledge and average importance chardonnay

Self
confidence
Sig. (2-tailed) .
Sig. (2-tailed) .197 .073 .
Ave Importance COO Correlation Coefficient -.080 -.076 .057 1.000
Sig. (2-tailed) .447 .470 .589 .
Ave Importance Price Correlation Coefficient .116 .109 .037 -.545 1.000
Sig. (2-tailed) .269 .299 .724 .000 .
Ave Importance Acid Correlation Coefficient -.081 -.039 -.054 -.250 -.607
Sig. (2-tailed) .440 .713 .608 .016 .000
N = 92
Table A 33 High objective knowledge and utility values
United Above
States Average
Objective knowledge Chardonnay Correlation Coefficient -.036 .038 -.001 .146 -.147 .016 .051 -.045 .036
Sig. (2-tailed) .732 .721 .990 .165 .161 .877 .631 .672 .733
Subjective knowledge Chardonnay Correlation Coefficient -.162 .136 .067 -.029 .151 -.023 .034 .076 -.007
Sig. (2-tailed) .122 .197 .526 .783 .152 .827 .744 .474 .947
Self confidence Correlation Coefficient .130 .147 -.274 .072 .046 -.135 .200 -.100 -.088
Sig. (2-tailed) .217 .162 .008 .495 .666 .200 .055 .342 .403
Sig. (2-tailed) .
Sig. (2-tailed) .000 .000 .
Sig. (2-tailed) .007 .013 .814 .
Sig. (2-tailed) .927 .219 .296 .004 .
$53.00 Correlation Coefficient .187 -.188 -.022 -.718 -.371 1.000
Sig. (2-tailed) .074 .073 .832 .000 .000 .
Average Correlation Coefficient .274 -.046 -.258 -.072 -.142 .154 1.000
Sig. (2-tailed) .008 .660 .013 .494 .176 .142 .
Above Average Correlation Coefficient -.357 .341 .102 .009 .102 -.063 -.483 1.000
Sig. (2-tailed) .000 .001 .333 .936 .334 .550 .000 .
High Correlation Coefficient -.027 -.140 .134 .145 -.015 -.130 -.514 -.393 1.000
Sig. (2-tailed) .795 .184 .201 .168 .891 .215 .000 .000 .
N = 92
308
Table A 34 Low objective knowledge and average importance chardonnay
Self
confidence
Sig. (2-tailed) .
Sig. (2-tailed) .511 .415 .
Ave Importance COO Correlation Coefficient -.090 -.032 -.187 1.000
Sig. (2-tailed) .424 .775 .095 .
Sig. (2-tailed) .220 .729 .706 .000 .
Sig. (2-tailed) .820 .668 .121 .006 .000
N = 81
Table A 35 Low objective knowledge and utility values

United Above
States Average
Objective knowledge Chardonnay Correlation Coefficient .069 .043 -.089 -.208 .065 .224 -.041 .080 -.078
Sig. (2-tailed) .542 .706 .429 .062 .566 .044 .717 .476 .488
Subjective knowledge Chardonnay Correlation Coefficient -.072 .228 -.153 .051 .080 -.054 -.085 -.052 .152
Sig. (2-tailed) .525 .041 .172 .648 .477 .634 .453 .643 .175
Self confidence Correlation Coefficient -.113 .068 .045 .063 -.151 .060 -.024 -.018 .014
Sig. (2-tailed) .313 .549 .690 .576 .177 .594 .835 .877 .905
Sig. (2-tailed) .
Sig. (2-tailed) .000 .000 .
$6.00 Correlation Coefficient -.002 -.216 .258 1.000
Sig. (2-tailed) .984 .053 .020 .
$16.00 Correlation Coefficient .048 .011 -.065 -.456 1.000
Sig. (2-tailed) .670 .924 .564 .000 .
$53.00 Correlation Coefficient -.080 .271 -.215 -.681 -.284 1.000
Sig. (2-tailed) .476 .014 .054 .000 .010 .
Average Correlation Coefficient .084 -.145 .063 .085 -.073 -.087 1.000
Sig. (2-tailed) .458 .196 .574 .452 .516 .440 .
Above Average Correlation Coefficient -.272 .182 .123 -.103 .061 .088 -.426 1.000
Sig. (2-tailed) .014 .103 .272 .361 .588 .436 .000 .
High Correlation Coefficient .171 .005 -.207 -.037 -.005 .048 -.536 -.475 1.000
Sig. (2-tailed) .128 .964 .064 .740 .967 .672 .000 .000 .
N = 81
309
Table A 36 High subjective knowledge and average importance chardonnay
Self
confidence
Sig. (2-tailed) .
Sig. (2-tailed) .104 .166 .
Ave Importance COO Correlation Coefficient -.096 -.069 .028 1.000
Sig. (2-tailed) .428 .572 .816 .
Sig. (2-tailed) .840 .351 .797 .000 .
Ave Importance Acid Correlation Coefficient .164 .074 .060 -.173 -.526
Sig. (2-tailed) .174 .542 .621 .153 .000
N = 70
Table A 37 High subjective knowledge and utility values

United Above
States Average
Objective knowledge Chardonnay Correlation Coefficient -.047 .094 -.083 .213 -.062 -.212 -.012 .060 -.014
Sig. (2-tailed) .698 .438 .493 .077 .612 .078 .924 .621 .907
Subjective knowledge Chardonnay Correlation Coefficient -.123 -.078 .175 .068 -.090 -.033 .094 .050 -.074
Sig. (2-tailed) .311 .520 .148 .575 .459 .785 .441 .683 .543
Self confidence Correlation Coefficient -.113 .363 -.196 .241 -.186 -.127 .093 .056 -.074
Sig. (2-tailed) .352 .002 .104 .044 .123 .297 .442 .645 .545
Sig. (2-tailed) .
Sig. (2-tailed) .000 .001 .
Sig. (2-tailed) .088 .055 .995 .
Sig. (2-tailed) .510 .256 .783 .000 .
$53.00 Correlation Coefficient .103 -.125 .058 -.783 -.173 1.000
Sig. (2-tailed) .395 .304 .631 .000 .151 .
Average Correlation Coefficient .142 .116 -.298 -.156 -.164 .267 1.000
Sig. (2-tailed) .240 .338 .012 .198 .175 .025 .
Above Average Correlation Coefficient -.285 .193 .233 -.057 -.026 .081 -.346 1.000
Sig. (2-tailed) .017 .110 .052 .638 .833 .503 .003 .
High Correlation Coefficient .023 -.202 .122 .231 .135 -.355 -.574 -.437 1.000
Sig. (2-tailed) .847 .093 .314 .055 .267 .003 .000 .000 .
N = 70
310
Table A 38 Low subjective knowledge and average importance chardonnay
Self
confidence
Sig. (1-tailed) .
Self confidence Correlation Coefficient -.159 -.071 1.000
Sig. (1-tailed) .099 .285 .
Ave Importance COO Correlation Coefficient -.213 .014 -.079 1.000
Sig. (1-tailed) .042 .456 .262 .
Sig. (1-tailed) .133 .310 .280 .000 .
Ave Importance Acid Correlation Coefficient .131 -.015 .087 -.427 -.458
Sig. (1-tailed) .145 .453 .241 .000 .000
N = 67
Table A 39 Low subjective knowledge and utility values

United Above
States Average
Objective knowledge Chardonnay Correlation Coefficient .135 -.012 -.166 -.046 -.184 .222 -.027 .018 -.096
Sig. (2-tailed) .277 .921 .179 .709 .136 .071 .830 .887 .438
Subjective knowledge Chardonnay Correlation Coefficient -.334 .065 .302 .136 .022 -.144 -.057 -.078 .076
Sig. (2-tailed) .006 .599 .013 .271 .862 .245 .645 .529 .542
Self confidence Correlation Coefficient .217 -.161 -.024 .156 -.114 -.110 .048 .159 -.142
Sig. (2-tailed) .077 .192 .849 .209 .359 .374 .701 .200 .251
Sig. (2-tailed) .
Sig. (2-tailed) .000 .030 .
Sig. (2-tailed) .342 .768 .010 .
Sig. (2-tailed) .905 .785 .640 .004 .
Sig. (2-tailed) .889 .382 .109 .000 .000 .
Average Correlation Coefficient .212 -.269 .012 -.054 .111 -.093 1.000
Sig. (2-tailed) .086 .028 .922 .661 .373 .452 .
Above Average Correlation Coefficient -.245 .186 .051 -.027 .049 .036 -.497 1.000
Sig. (2-tailed) .045 .133 .682 .829 .694 .771 .000 .
High Correlation Coefficient .023 .118 -.114 .084 -.152 .062 -.487 -.419 1.000
Sig. (2-tailed) .851 .340 .357 .498 .218 .616 .000 .000 .
N = 67
311
Table A 40 High self-confidence and average importance chardonnay
Self
confidence
Sig. (2-tailed) .
Sig. (2-tailed) .990 .735 .
Ave Importance COO Correlation Coefficient .062 .086 -.077 1.000
Sig. (2-tailed) .587 .454 .505 .
Sig. (2-tailed) .067 .158 .408 .000 .
Sig. (2-tailed) .118 .104 .323 .001 .000
N = 78
Table A 41 High self-confidence and utility values

United Above
States Average
Objective knowledge Chardonnay Correlation Coefficient .046 .172 -.205 -.067 .056 -.008 .168 .056 -.171
Sig. (2-tailed) .687 .132 .072 .562 .625 .944 .142 .627 .136
Subjective knowledge Chardonnay Correlation Coefficient -.246 .371 -.034 -.013 .126 -.051 .081 .050 -.031
Sig. (2-tailed) .030 .001 .766 .907 .271 .655 .480 .666 .786
Self confidence Correlation Coefficient -.023 .000 -.041 -.012 .129 -.116 .021 .063 -.083
Sig. (2-tailed) .842 .997 .720 .916 .261 .314 .852 .584 .471
Sig. (2-tailed) .
Sig. (2-tailed) .000 .108 .
Sig. (2-tailed) .026 .220 .102 .
Sig. (2-tailed) .024 .080 .557 .001 .
Sig. (2-tailed) .637 .967 .255 .000 .000 .
Average Correlation Coefficient .157 -.075 -.109 .002 -.085 .028 1.000
Sig. (2-tailed) .169 .517 .343 .986 .457 .805 .
Above Average Correlation Coefficient -.261 .121 .191 -.116 -.093 .181 -.455 1.000
Sig. (2-tailed) .021 .290 .093 .311 .420 .112 .000 .
High Correlation Coefficient -.027 .138 -.073 .132 .067 -.162 -.543 -.397 1.000
Sig. (2-tailed) .812 .227 .524 .251 .561 .156 .000 .000 .
N = 78
312
Table A 42 Low self-confidence and average importance chardonnay
Objective Subjective Self Ave Ave
knowledge knowledge confidence Importance Importance
Chardonnay Chardonnay (sensory) COO Price
Sig. (2-tailed) .
Sig. (2-tailed)
.000 .
Self confidence (sensory) Correlation Coefficient .000 .094 1.000

Sig. (2-tailed) .992 .481 .
Ave Importance COO Correlation Coefficient -.240 -.105 -.068 1.000
Sig. (2-tailed) .070 .433 .609 .
Sig. (2-tailed) .704 .650 .245 .000 .
Ave Importance Acid Correlation Coefficient .205 .047 .221 -.018 -.613
Sig. (2-tailed) .123 .729 .095 .894 .000
N = 58
Table A 43 Low self-confidence and utility values

United Above
States Average
Objective knowledge Chardonnay Correlation Coefficient -.048 .082 -.024 -.028 -.146 .183 -.050 .045 .029
Sig. (2-tailed) .718 .541 .859 .834 .273 .168 .708 .736 .830
Subjective knowledge Chardonnay Correlation Coefficient .147 -.112 -.055 -.096 .122 .032 -.004 .035 .031
Sig. (2-tailed) .270 .402 .683 .472 .362 .812 .975 .793 .816
Self confidence Correlation Coefficient .285 -.037 -.354 .056 .020 -.131 -.067 -.219 .246
Sig. (2-tailed) .030 .782 .006 .674 .879 .326 .618 .098 .062
Sig. (2-tailed) .
Sig. (2-tailed) .001 .002 .
Sig. (2-tailed) .251 .979 .143 .
Sig. (2-tailed) .216 .213 .879 .000 .
Sig. (2-tailed) .677 .307 .397 .000 .541 .
Average Correlation Coefficient -.141 -.036 .205 -.040 -.052 .066 1.000
Sig. (2-tailed) .291 .786 .123 .768 .697 .622 .
Above Average Correlation Coefficient -.283 .164 .172 .087 .079 -.119 -.386 1.000
Sig. (2-tailed) .032 .220 .198 .514 .553 .372 .003 .
High Correlation Coefficient .372 -.096 -.342 -.019 -.041 .059 -.575 -.441 1.000
Sig. (2-tailed) .004 .473 .009 .888 .762 .662 .000 .001 .
N = 58
313
8.2 Conjoint survey (chardonnay)
Table A 44 High objective knowledge and average importance chardonnay

Self
confidence
Objective knowledge chardonnay Correlation Coefficie 1.000
Sig. (2-tailed) .
Subjective knowledge chardonnay Correlation Coefficie .153 1.000
Self confidence Correlation Coefficie .059 .319 1.000
Sig. (2-tailed) .654 .014 .
Average Importance COO Correlation Coefficie .066 .011 -.008 1.000
Sig. (2-tailed) .622 .935 .952 .
Average Importance Price Correlation Coefficie .200 -.008 .044 -.634 1.000
Sig. (2-tailed) .130 .952 .743 .000 .
Average Importance Acid Correlation Coefficie -.206 -.076 -.156 -.039 -.651
Sig. (2-tailed) .118 .570 .238 .772 .000
N = 59

United Above
States average
Objective knowledge chardonnay Correlation Coefficient -.032 .088 -.060 -.341 -.116 .422 .073 .040 -.032
Sig. (2-tailed) .810 .508 .652 .008 .383 .001 .585 .761 .813
Subjective knowledge chardonnay Correlation Coefficient -.170 .050 .184 -.181 .329 .028 .042 -.158 .053
Sig. (2-tailed) .198 .704 .162 .169 .011 .832 .751 .232 .689
Self confidence Correlation Coefficient -.148 .322 -.101 .050 -.130 -.007 -.094 -.087 .128
Sig. (2-tailed) .264 .013 .448 .706 .325 .959 .478 .510 .335
Sig. (2-tailed) .
Sig. (2-tailed) .000 .004 .
Sig. (2-tailed) .136 .829 .052 .
Sig. (2-tailed) .510 .276 .750 .019 .
Sig. (2-tailed) .205 .353 .017 .000 .261 .
Sig. (2-tailed) .001 .117 .224 .345 .114 .930 .
Above average Correlation Coefficient .131 -.095 -.027 -.097 -.038 .081 -.365 1.000
Sig. (2-tailed) .324 .474 .837 .465 .775 .541 .004 .
High Correlation Coefficient .362 -.116 -.174 -.157 .267 .032 -.799 -.181 1.000
Sig. (2-tailed) .005 .381 .186 .234 .041 .810 .000 .170 .
N = 59
314
Table A 46 Low objective knowledge and average importance chardonnay
Self
confidence
Sig. (2-tailed) .
Self confidence Correlation Coefficient .110 -.071 1.000
Sig. (2-tailed) .421 .602 .
Average Importance COO Correlation Coefficient -.164 -.109 -.188 1.000
Sig. (2-tailed) .227 .422 .164 .
Average Importance Price Correlation Coefficient .151 .104 -.019 -.701 1.000
Sig. (2-tailed) .267 .445 .891 .000 .
Average Importance Acid Correlation Coefficient -.104 -.064 .101 .238 -.801
Sig. (2-tailed) .446 .642 .457 .077 .000
N = 56

United Above
States average
Objective knowledge chardonnay Correlation Coefficient -.052 .059 .012 -.180 .056 .098 -.025 -.187 .168
Sig. (2-tailed) .703 .666 .928 .185 .684 .473 .855 .166 .215
Subjective knowledge chardonnay Correlation Coefficient .017 -.068 .002 -.054 .099 -.020 -.192 .028 .111
Sig. (2-tailed) .900 .620 .986 .693 .467 .881 .156 .840 .417
Self confidence Correlation Coefficient .303 -.159 -.248 .009 .071 -.077 -.131 -.084 .140
Sig. (2-tailed) .023 .242 .066 .949 .601 .572 .335 .539 .303
Sig. (2-tailed) .
Sig. (2-tailed) .000 .317 .
$6.00 Correlation Coefficient .041 .027 -.121 1.000
Sig. (2-tailed) .764 .845 .373 .
$16.00 Correlation Coefficient -.117 -.046 .220 -.244 1.000
Sig. (2-tailed) .392 .738 .104 .069 .
Sig. (2-tailed) .981 .965 .945 .000 .029 .
Average Correlation Coefficient -.083 .022 .139 .082 -.152 .023 1.000
Sig. (2-tailed) .543 .872 .307 .549 .264 .867 .
Above average Correlation Coefficient .128 .152 -.208 .222 -.297 -.108 -.333 1.000
Sig. (2-tailed) .346 .264 .124 .101 .026 .430 .012 .
High Correlation Coefficient -.054 -.166 .065 -.259 .354 .119 -.406 -.662 1.000
Sig. (2-tailed) .691 .220 .637 .054 .007 .384 .002 .000 .
N = 56
315
Table A 48 High subjective knowledge and average importance chardonnay
Self
confidence
Sig. (2-tailed) .
Sig. (2-tailed) .080 .243 .
Average Importance COO Correlation Coefficient -.041 -.109 .064 1.000
Sig. (2-tailed) .736 .368 .600 .
Average Importance Price Correlation Coefficient .196 .240 .004 -.650 1.000
Sig. (2-tailed) .104 .045 .972 .000 .
Average Importance Acid Correlation Coefficient -.244 -.303 -.104 .120 -.727
Sig. (2-tailed) .041 .011 .393 .323 .000
N = 70

United Above
States average
Objective knowledge chardonnay Correlation Coefficient -.106 .077 .034 -.235 -.010 .291 .053 -.168 .066
Sig. (2-tailed) .381 .528 .783 .050 .932 .015 .666 .163 .585
Subjective knowledge chardonnay Correlation Coefficient -.115 -.072 .129 -.286 .188 .262 -.005 -.098 .040
Sig. (2-tailed) .343 .555 .289 .017 .119 .029 .968 .420 .743
Self confidence Correlation Coefficient -.178 .092 .105 .025 -.131 -.012 .131 -.115 -.042
Sig. (2-tailed) .140 .450 .385 .835 .278 .919 .280 .345 .731
Sig. (2-tailed) .
Sig. (2-tailed) .000 .018 .
Sig. (2-tailed) .075 .847 .046 .
Sig. (2-tailed) .141 .361 .881 .000 .
$53.00 Correlation Coefficient .121 .017 -.247 -.875 .008 1.000
Sig. (2-tailed) .318 .890 .040 .000 .944 .
Average Correlation Coefficient -.360 .167 .172 .022 .009 -.020 1.000
Sig. (2-tailed) .002 .167 .155 .857 .944 .867 .
Above average Correlation Coefficient .193 -.137 -.029 .082 -.073 -.070 -.374 1.000
Sig. (2-tailed) .110 .256 .809 .500 .550 .566 .001 .
Sig. (2-tailed) .008 .657 .044 .274 .809 .301 .000 .082 .
N = 70
316
Table A 50 Low subjective knowledge and average importance chardonnay
Self
confidence
Sig. (2-tailed) .
Sig. (2-tailed) .720 .510 .
Sig. (2-tailed) .922 .082 .683 .
Sig. (2-tailed) .416 .004 .984 .000 .
Average Importance Acid Correlation Coefficient -.085 -.273 .057 .049 -.702
Sig. (2-tailed) .482 .021 .639 .685 .000
N = 71

United Above
States Average
Objective knowledge chardonnay Correlation Coefficient .073 -.059 -.093 -.141 .019 .083 .096 -.100 .018
Sig. (2-tailed) .544 .623 .439 .241 .877 .491 .426 .405 .880
Subjective knowledge chardonnay Correlation Coefficient .182 -.036 -.183 -.363 -.013 .320 -.098 -.132 .229
Sig. (2-tailed) .128 .764 .128 .002 .917 .007 .418 .272 .055
Self confidence Correlation Coefficient .197 -.076 -.193 .068 -.028 -.115 -.011 -.023 -.034
Sig. (2-tailed) .099 .531 .106 .575 .819 .340 .930 .846 .778
Sig. (2-tailed) .
Sig. (2-tailed) .000 .026 .
Sig. (2-tailed) .076 .420 .152 .
Sig. (2-tailed) .175 .085 .907 .165 .
Sig. (2-tailed) .584 .797 .216 .000 .000 .
Average Correlation Coefficient -.073 .129 -.006 .079 -.154 .030 1.000
Sig. (2-tailed) .548 .283 .957 .510 .199 .803 .
Above average Correlation Coefficient .097 -.032 -.027 .128 .016 -.156 -.336 1.000
Sig. (2-tailed) .419 .790 .822 .288 .894 .195 .004 .
High Correlation Coefficient -.011 -.036 -.039 -.224 .054 .210 -.458 -.604 1.000
Sig. (2-tailed) .928 .767 .744 .060 .654 .078 .000 .000 .
N = 71
317
Table A 52 High self-confidence and average importance chardonnay
Self
confidence
Sig. (2-tailed) .
Sig. (2-tailed) .080 .243 .
Sig. (2-tailed) .736 .368 .600 .
Sig. (2-tailed) .104 .045 .972 .000 .
Average Importance Acid Correlation Coefficient -.244 -.303 -.104 .120 -.727
Sig. (2-tailed) .041 .011 .393 .323 .000
N = 70

United Above
States average
Objective knowledge chardonnay Correlation Coefficient -.106 .077 .034 -.235 -.010 .291 .053 -.168 .066
Sig. (2-tailed) .381 .528 .783 .050 .932 .015 .666 .163 .585
Subjective knowledge chardonnay Correlation Coefficient -.115 -.072 .129 -.286 .188 .262 -.005 -.098 .040
Sig. (2-tailed) .343 .555 .289 .017 .119 .029 .968 .420 .743
Self confidence Correlation Coefficient -.178 .092 .105 .025 -.131 -.012 .131 -.115 -.042
Sig. (2-tailed) .140 .450 .385 .835 .278 .919 .280 .345 .731
Sig. (2-tailed) .
Sig. (2-tailed) .000 .018 .
Sig. (2-tailed) .075 .847 .046 .
Sig. (2-tailed) .141 .361 .881 .000 .
$53.00 Correlation Coefficient .121 .017 -.247 -.875 .008 1.000
Sig. (2-tailed) .318 .890 .040 .000 .944 .
Average Correlation Coefficient -.360 .167 .172 .022 .009 -.020 1.000
Sig. (2-tailed) .002 .167 .155 .857 .944 .867 .
Above average Correlation Coefficient .193 -.137 -.029 .082 -.073 -.070 -.374 1.000
Sig. (2-tailed) .110 .256 .809 .500 .550 .566 .001 .
Sig. (2-tailed) .008 .657 .044 .274 .809 .301 .000 .082 .
N = 70
318
Table A 54 Low self-confidence and average importance chardonnay
Self
confidence
Sig. (2-tailed) .
Self confidence Correlation Coefficient -.022 .192 1.000
Sig. (2-tailed) .856 .108 .
Average Importance COO Correlation Coefficient -.055 .035 .012 1.000
Sig. (2-tailed) .647 .774 .920 .
Average Importance Price Correlation Coefficient .055 -.020 -.192 -.715 1.000
Sig. (2-tailed) .650 .869 .109 .000 .
Average Importance Acid Correlation Coefficient -.003 .018 .261 .168 -.773
Sig. (2-tailed) .981 .881 .028 .161 .000
N = 71

United Above
States average
Objective knowledge chardonnay Correlation Coefficient .158 -.131 .000 -.105 .152 .038 .026 -.085 .059
Sig. (2-tailed) .188 .275 .998 .385 .205 .755 .830 .479 .625
Subjective knowledge chardonnay Correlation Coefficient .248 .007 -.106 -.029 -.007 .008 -.157 .054 .063
Sig. (2-tailed) .037 .955 .381 .808 .951 .945 .192 .653 .604
Self confidence Correlation Coefficient .092 .169 -.154 .199 .132 -.251 .098 .073 -.155
Sig. (2-tailed) .446 .159 .199 .097 .271 .035 .416 .547 .197
Sig. (2-tailed) .
Sig. (2-tailed) .000 .005 .
$6.00 Correlation Coefficient .076 -.027 -.119 1.000
Sig. (2-tailed) .528 .824 .324 .
Sig. (2-tailed) .415 .683 .926 .133 .
$53.00 Correlation Coefficient -.074 .010 .092 -.845 -.323 1.000
Sig. (2-tailed) .542 .931 .447 .000 .006 .
Sig. (2-tailed) .006 .990 .035 .209 .802 .267 .
Above average Correlation Coefficient .163 -.109 -.055 .034 -.243 .087 -.228 1.000
Sig. (2-tailed) .174 .363 .647 .777 .041 .472 .056 .
High Correlation Coefficient .156 .079 -.177 -.152 .134 .111 -.790 -.335 1.000
Sig. (2-tailed) .195 .513 .139 .205 .264 .357 .000 .004 .
N = 71
319
Appendix
9 Correlation matrices (brie)
320
9.1 Sensory experiment
Table A 56 High objective knowledge and average importance brie

Self
confidence
Brie Brie COO Price
Sig. (2-tailed) .
Sig. (2-tailed) .138 .751 .
Average Importance COO Correlation Coefficient -.041 .204 .055 1.000
Sig. (2-tailed) .764 .129 .687 .
Sig. (2-tailed) .425 .032 .614 .022 .
Average Importance Fat Correlation Coefficient .130 .130 -.069 -.452 -.640
Sig. (2-tailed) .335 .336 .608 .000 .000
N = 57

Objective Knowledge Brie Correlation Coefficient -.018 -.001 .014 .086 -.026 -.034 -.211 .090 .121
Sig. (2-tailed) .893 .993 .915 .526 .846 .803 .114 .506 .369
Subjective Knowledge Brie Correlation Coefficient .106 -.035 -.006 .051 .164 -.220 -.370 -.066 .302
Sig. (2-tailed) .434 .799 .964 .708 .223 .100 .005 .624 .023
Self confidence Correlation Coefficient .051 -.173 .135 .274 -.004 -.163 .192 -.096 -.081
Sig. (2-tailed) .707 .197 .316 .039 .978 .227 .152 .476 .548
Sig. (2-tailed) .
Sig. (2-tailed) .000 .102 .
$28.95 per kilo Correlation Coefficient -.111 .025 .237 1.000
Sig. (2-tailed) .413 .853 .076 .
$49.00 per kilo Correlation Coefficient -.066 -.056 .154 -.179 1.000
Sig. (2-tailed) .627 .677 .253 .183 .
Sig. (2-tailed) .494 .430 .015 .000 .000 .
Full Cream Correlation Coefficient -.136 .193 -.106 .098 .155 -.111 1.000
Sig. (2-tailed) .315 .150 .434 .469 .250 .410 .
Double Cream Correlation Coefficient -.018 .033 .013 .078 -.191 .056 .093 1.000
Sig. (2-tailed) .894 .810 .925 .564 .156 .678 .491 .
Triple Cream Correlation Coefficient .071 -.104 .056 -.108 .117 -.050 -.691 -.707 1.000
Sig. (2-tailed) .599 .440 .679 .425 .384 .714 .000 .000 .
N = 57
321
Table A 58 Low objective knowledge and average importance brie
Self
confidence
Brie Brie COO Price
Sig. (2-tailed) .
Subjective Knowledge Brie Correlation Coefficient -.080 1.000
Sig. (2-tailed) .296 .805 .
Sig. (2-tailed) .835 .902 .938 .
Average Importance Price Correlation Coefficient .069 -.079 -.266 .075 1.000
Sig. (2-tailed) .614 .564 .048 .582 .
Sig. (2-tailed) .861 .810 .218 .000 .000
N = 56

Objective Knowledge Brie Correlation Coefficient -.025 -.038 -.020 -.178 .121 .106 .016 -.129 .050
Sig. (2-tailed) .853 .780 .886 .190 .374 .437 .909 .342 .713
Subjective Knowledge Brie Correlation Coefficient -.035 .158 -.161 .069 -.052 -.063 -.066 -.016 .049
Sig. (2-tailed) .797 .245 .235 .613 .704 .642 .628 .906 .719
Self confidence Correlation Coefficient -.002 .007 -.066 .125 -.201 .042 .198 .083 -.190
Sig. (2-tailed) .989 .958 .629 .361 .136 .758 .144 .544 .161
Sig. (2-tailed) .
Sig. (2-tailed) .001 .000 .
Sig. (2-tailed) .023 .557 .032 .
$49.00 per kilo Correlation Coefficient -.245 .303 -.012 -.380 1.000
Sig. (2-tailed) .069 .023 .931 .004 .
$69.95 per kilo Correlation Coefficient .494 -.140 -.285 -.585 -.452 1.000
Sig. (2-tailed) .000 .303 .033 .000 .000 .
Full Cream Correlation Coefficient .027 -.103 .097 .124 -.005 -.060 1.000
Sig. (2-tailed) .842 .448 .477 .363 .971 .663 .
Double Cream Correlation Coefficient .018 -.042 .041 -.095 .202 -.102 .123 1.000
Sig. (2-tailed) .897 .757 .763 .485 .135 .454 .365 .
Triple Cream Correlation Coefficient .006 .103 -.125 -.043 -.148 .135 -.802 -.637 1.000
Sig. (2-tailed) .966 .448 .358 .753 .275 .321 .000 .000 .
N = 56
322
Table A 60 High subjective knowledge and average importance brie
Self
confidence
Brie Brie COO Price
Sig. (2-tailed) .
Sig. (2-tailed) .598 .488 .
Average Importance COO Correlation Coefficient .153 -.017 -.038 1.000
Sig. (2-tailed) .214 .889 .759 .
Sig. (2-tailed) .529 .237 .302 .100 .
Average Importance Fat Correlation Coefficient .001 .167 -.128 -.625 -.560
Sig. (2-tailed) .995 .172 .297 .000 .000
N = 68

Objective Knowledge Brie Correlation Coefficient .019 -.071 .090 -.031 .155 -.183 -.118 -.039 .158
Sig. (2-tailed) .879 .565 .463 .799 .208 .136 .339 .752 .198
Subjective Knowledge Brie Correlation Coefficient .037 .044 -.049 .125 -.031 -.144 -.160 -.092 .179
Sig. (2-tailed) .762 .723 .694 .310 .799 .242 .193 .458 .144
Self confidence Correlation Coefficient .111 -.233 .101 .248 -.232 .059 -.050 -.001 -.014
Sig. (2-tailed) .365 .056 .414 .041 .057 .634 .685 .991 .909
Sig. (2-tailed) .
Sig. (2-tailed) .000 .000 .
Sig. (2-tailed) .047 .728 .018 .
$49.00 per kilo Correlation Coefficient -.010 -.031 .122 -.458 1.000
Sig. (2-tailed) .937 .800 .320 .000 .
Sig. (2-tailed) .111 .352 .003 .000 .000 .
Full Cream Correlation Coefficient -.045 .037 -.006 .063 -.160 .079 1.000
Sig. (2-tailed) .715 .761 .962 .608 .191 .521 .
Double Cream Correlation Coefficient .013 .039 -.038 .238 -.300 .057 .009 1.000
Sig. (2-tailed) .914 .753 .757 .051 .013 .644 .943 .
Triple Cream Correlation Coefficient .027 -.021 .000 -.191 .327 -.133 -.655 -.687 1.000
Sig. (2-tailed) .830 .864 .998 .118 .006 .278 .000 .000 .
N = 68
323
Table A 62 Low subjective knowledge and average importance brie
Self
confidence
Brie Brie COO Price
Sig. (2-tailed) .
Self confidence Correlation Coefficient -.060 .193 1.000
Sig. (2-tailed) .627 .119 .
Sig. (2-tailed) .657 .109 .923 .
Average Importance Price Correlation Coefficient -.105 -.101 -.082 .111 1.000
Sig. (2-tailed) .396 .415 .509 .372 .
Average Importance Fat Correlation Coefficient .064 .190 .002 -.683 -.763
Sig. (2-tailed) .608 .125 .985 .000 .000
N = 67

Objective Knowledge Brie Correlation Coefficient -.051 .175 -.177 .310 -.213 -.113 -.140 -.295 .262
Sig. (2-tailed) .684 .158 .151 .011 .084 .362 .258 .015 .032
Subjective Knowledge Brie Correlation Coefficient -.016 .260 -.246 .104 -.065 -.058 -.156 -.070 .142
Sig. (2-tailed) .896 .033 .045 .404 .600 .641 .207 .575 .252
Self confidence Correlation Coefficient -.029 .119 -.066 .041 -.092 .103 -.127 -.129 .130
Sig. (2-tailed) .815 .337 .595 .742 .459 .406 .306 .298 .294
Sig. (2-tailed) .
Sig. (2-tailed) .001 .001 .
Sig. (2-tailed) .033 .244 .408 .
$49.00 per kilo Correlation Coefficient .157 -.087 -.025 -.649 1.000
Sig. (2-tailed) .203 .484 .840 .000 .
$69.95 per kilo Correlation Coefficient .197 -.097 -.127 -.472 -.312 1.000
Sig. (2-tailed) .109 .437 .307 .000 .010 .
Full Cream Correlation Coefficient .185 -.228 .092 -.052 .077 -.005 1.000
Sig. (2-tailed) .134 .063 .457 .678 .535 .967 .
Double Cream Correlation Coefficient .198 -.194 .030 -.034 .091 -.135 .109 1.000
Sig. (2-tailed) .108 .115 .808 .782 .463 .276 .381 .
Triple Cream Correlation Coefficient -.261 .286 -.057 .078 -.149 .115 -.724 -.710 1.000
Sig. (2-tailed) .033 .019 .649 .531 .228 .355 .000 .000 .
N = 67
324
Table A 64 High self-confidence and average importance brie
Self
confidence
Brie Brie COO Price
Sig. (2-tailed) .
Sig. (2-tailed) .918 .940 .
Average Importance COO Correlation Coefficient .081 -.056 .022 1.000
Sig. (2-tailed) .483 .628 .850 .
Sig. (2-tailed) .132 .658 .412 .452 .
Average Importance Fat Correlation Coefficient -.154 .052 -.135 -.563 -.719
Sig. (2-tailed) .177 .652 .237 .000 .000
N = 78

Objective Knowledge Brie Correlation Coefficient .059 -.069 .033 .211 -.063 -.127 -.202 -.243 .307
Sig. (2-tailed) .611 .550 .775 .063 .586 .267 .077 .032 .006
Subjective Knowledge Correlation Coefficient .075 -.220 .154 .164 -.184 .029 -.249 -.021 .198
Brie Sig. (2-tailed)
.512 .052 .178 .150 .107 .803 .028 .854 .083
Self confidence Correlation Coefficient -.091 .083 .035 -.039 .088 -.004 .114 .191 -.185
Sig. (2-tailed) .429 .469 .759 .737 .441 .971 .321 .094 .105
Sig. (2-tailed) .
Sig. (2-tailed) .000 .004 .
Sig. (2-tailed) .314 .073 .019 .
$49.00 per kilo Correlation Coefficient -.028 .117 -.051 -.496 1.000
Sig. (2-tailed) .808 .308 .658 .000 .
Sig. (2-tailed) .138 .345 .027 .000 .000 .
Full Cream Correlation Coefficient -.087 .076 -.010 .081 .046 -.108 1.000
Sig. (2-tailed) .448 .509 .931 .483 .688 .346 .
Double Cream Correlation Coefficient -.084 -.060 .125 .063 -.024 -.127 .077 1.000
Sig. (2-tailed) .465 .600 .274 .585 .835 .268 .502 .
Triple Cream Correlation Coefficient .159 -.049 -.047 -.094 -.042 .161 -.686 -.699 1.000
Sig. (2-tailed) .163 .668 .686 .416 .716 .158 .000 .000 .
N = 78
325
Table A 66 Low self-confidence and average importance brie
Self
confidence
Brie Brie COO Price
Sig. (2-tailed) .
Sig. (2-tailed) .030 .991 .
Average Importance COO Correlation Coefficient -.038 .224 -.168 1.000
Sig. (2-tailed) .775 .092 .208 .
Average Importance Price Correlation Coefficient .067 -.180 -.063 -.006 1.000
Sig. (2-tailed) .616 .178 .638 .965 .
Sig. (2-tailed) .721 .759 .283 .000 .000
N = 58

Objective Knowledge Brie Correlation Coefficient .089 .049 -.158 -.177 -.140 .265 -.143 -.132 .190
Sig. (2-tailed) .504 .716 .236 .183 .295 .044 .286 .323 .154
Subjective Knowledge Brie Correlation Coefficient -.003 .050 -.066 .078 -.060 .020 -.204 .122 .075
Sig. (2-tailed) .985 .712 .621 .560 .653 .880 .125 .360 .577
Self confidence Correlation Coefficient .055 -.035 -.024 .155 -.230 .065 -.014 -.106 .055
Sig. (2-tailed) .682 .797 .858 .246 .082 .628 .917 .428 .682
Sig. (2-tailed) .
Sig. (2-tailed) .000 .000 .
Sig. (2-tailed) .000 .261 .032 .
Sig. (2-tailed) .005 .121 .764 .001 .
Sig. (2-tailed) .352 .719 .076 .000 .000 .
Full Cream Correlation Coefficient .144 -.249 .178 .062 .124 -.089 1.000
Sig. (2-tailed) .282 .059 .181 .644 .353 .508 .
Double Cream Correlation Coefficient .028 .039 .023 .126 -.034 -.062 -.016 1.000
Sig. (2-tailed) .834 .774 .864 .346 .802 .644 .904 .
Triple Cream Correlation Coefficient -.141 .216 -.172 -.103 -.107 .116 -.820 -.515 1.000
Sig. (2-tailed) .292 .103 .198 .442 .422 .387 .000 .000 .
N = 58
326
9.2 Conjoint survey (brie)
Table A 68 High objective knowledge and average importance brie

Self
confidence
brie brie COO price
Sig. (2-tailed) .
Sig. (2-tailed)
.410
Self confidence Correlation Coefficient -.014 1.000

Sig. (2-tailed) .910 .134 .
Average importance COO Correlation Coefficient -.167 .186 -.170 1.000
Sig. (2-tailed) .161 .118 .153 .
Sig. (2-tailed) .379 .040 .594 .458 .
Sig. (2-tailed) .095 .484 .407 .000 .000
N = 72

Objective knowledge brie Correlation Coefficient .056 -.036 -.018 .034 .049 -.048 -.116 .002 .159
Sig. (2-tailed) .640 .763 .882 .775 .685 .690 .330 .988 .182
Subjective knowledge brie Correlation Coefficient -.113 -.128 .203 .172 -.310 .051 -.117 -.014 .191
Sig. (2-tailed)
.346 .283 .088 .148 .008 .669 .328 .907 .108
Self confidence Correlation Coefficient .134 -.089 -.188 -.077 -.119 .077 .020 -.211 .110
Sig. (2-tailed) .263 .455 .115 .522 .319 .521 .868 .075 .359
Sig. (2-tailed) .
Sig. (2-tailed) .000 .142 .
Sig. (2-tailed) .575 .557 .064 .
Sig. (2-tailed) .789 .602 .778 .000 .
Sig. (2-tailed) .654 .271 .033 .000 .009 .
Full Cream Correlation Coefficient -.084 .067 .064 -.231 .030 .313 1.000
Sig. (2-tailed) .485 .575 .594 .051 .805 .007 .
Double Cream Correlation Coefficient .075 -.038 -.008 .254 .103 -.368 -.735 1.000 .351
Sig. (2-tailed) .530 .750 .947 .032 .388 .001 .000 . .002
Triple Cream Correlation Coefficient .114 -.072 -.126 .118 -.110 -.148 -.861 .351 1.000
Sig. (2-tailed) .340 .547 .290 .323 .357 .215 .000 .002 .
N = 72
327
Table A 70 Low objective knowledge and average importance brie
Self
Knowledge knowledge importance importance
confidence
brie brie COO price
Objective Knowledge brie Correlation Coefficient 1.000
Sig. (2-tailed) .
Sig. (2-tailed) .705 .611 .
Average importance COO Correlation Coefficient -.007 .257 .089 1.000
Sig. (2-tailed) .953 .032 .465 .
Average importance price Correlation Coefficient -.251 -.158 -.020 -.626 1.000
Sig. (2-tailed) .036 .191 .871 .000 .
Average importance fat Correlation Coefficient .265 -.022 -.008 -.122 -.550
Sig. (2-tailed) .026 .860 .950 .316 .000
N = 70

$28.95 $49.00 $69.95 Double Triple
Argentina Canada France Full Cream
per kilo per kilo per kilo Cream Cream
Objective Knowledge brie Correlation Coefficient .100 -.203 .035 .295 -.271 -.148 -.285 .123 .377
Sig. (2-tailed) .410 .092 .772 .013 .023 .221 .017 .311 .001
Subjective knowledge brie Correlation Coefficient .006 -.039 .090 .227 .018 -.260 .041 -.124 .125
Sig. (2-tailed) .958 .748 .460 .059 .884 .030 .737 .307 .301
Self confidence Correlation Coefficient -.103 .201 -.066 -.041 -.181 .161 .094 -.068 -.055
Sig. (2-tailed) .397 .095 .590 .735 .135 .183 .437 .573 .654
Sig. (2-tailed) .
Sig. (2-tailed) .000 .009 .
Sig. (2-tailed) .355 .155 .701 .
Sig. (2-tailed) .607 .466 .508 .000 .
Sig. (2-tailed) .185 .025 .274 .000 .177 .
Sig. (2-tailed) .452 .053 .053 .223 .785 .024 .
Double Cream Correlation Coefficient .178 -.165 .020 .020 .149 -.131 -.729 1.000 .012
Sig. (2-tailed) .139 .172 .870 .867 .219 .278 .000 . .924
Triple Cream Correlation Coefficient -.043 -.150 .289 .203 -.048 -.278 -.617 .012 1.000
Sig. (2-tailed) .722 .215 .015 .092 .692 .020 .000 .924 .
N = 70
328
Table A 72 High subjective knowledge and average importance brie
Self
confidence
brie brie COO price
Sig. (2-tailed) .
Sig. (2-tailed) .820 .140 .
Average importance COO Correlation Coefficient .016 -.067 .026 1.000
Sig. (2-tailed) .896 .583 .834 .
Sig. (2-tailed) .120 .221 .664 .002 .
Average importance fat Correlation Coefficient .141 .059 -.071 -.538 -.384
Sig. (2-tailed) .248 .629 .560 .000 .001
N = 69

Objective knowledge brie Correlation Coefficient -.184 -.050 .209 -.043 -.114 .172 -.183 .279 .078
Sig. (2-tailed) .130 .685 .085 .725 .350 .157 .132 .020 .526
Subjective knowledge brie Correlation Coefficient -.180 .009 .194 .050 -.157 .082 -.114 .113 .131
Sig. (2-tailed) .140 .944 .111 .683 .197 .503 .351 .353 .285
Self confidence Correlation Coefficient -.106 .012 .030 -.071 -.032 .022 .200 -.232 -.182
Sig. (2-tailed) .386 .923 .804 .562 .794 .857 .099 .055 .134
Sig. (2-tailed) .
Sig. (2-tailed) .000 .092 .
Sig. (2-tailed) .917 .600 .370 .
Sig. (2-tailed) .899 .152 .301 .000 .
Sig. (2-tailed) .914 .107 .086 .000 .051 .
Full Cream Correlation Coefficient -.297 .255 .113 -.050 -.174 .250 1.000
Sig. (2-tailed) .013 .034 .357 .682 .154 .038 .
Sig. (2-tailed) .142 .036 .781 .540 .127 .042 .000 .
Triple Cream Correlation Coefficient .346 -.209 -.217 .055 .051 -.141 -.835 .374 1.000
Sig. (2-tailed) .004 .085 .074 .651 .676 .249 .000 .002 .
N = 69
329
Table A 74 Low subjective knowledge and average importance brie
Self
confidence
brie brie COO price
Sig. (2-tailed) .
Self confidence Correlation Coefficient -.163 -.147 1.000
Sig. (2-tailed) .195 .242 .
Average importance COO Correlation Coefficient .198 .180 -.117 1.000
Sig. (2-tailed) .115 .152 .353 .
Sig. (2-tailed) .112 .906 .454 .000 .
Average importance fat Correlation Coefficient .064 -.161 .035 -.120 -.649
Sig. (2-tailed) .613 .201 .781 .340 .000
N = 65

Objective knowledge brie Correlation Coefficient .088 -.207 .132 .163 .100 -.202 -.302 .218 .272
Sig. (2-tailed) .487 .097 .295 .194 .426 .106 .014 .081 .028
Subjective knowledge brie Correlation Coefficient -.110 -.134 .229 .132 .013 -.045 -.015 -.024 .145
Sig. (2-tailed) .381 .288 .067 .294 .915 .722 .907 .848 .248
Self confidence Correlation Coefficient -.116 .271 -.186 -.062 -.251 .157 .100 -.115 -.039
Sig. (2-tailed) .359 .029 .139 .621 .044 .211 .429 .360 .760
Sig. (2-tailed) .
Sig. (2-tailed) .000 .010 .
Sig. (2-tailed) .972 .083 .030 .
Sig. (2-tailed) .069 .020 .800 .010 .
Sig. (2-tailed) .342 .001 .047 .000 .012 .
Full Cream Correlation Coefficient .012 .129 -.152 -.328 -.095 .402 1.000
Sig. (2-tailed) .925 .306 .226 .008 .452 .001 .
Double Cream Correlation Coefficient .175 -.053 -.095 .074 .254 -.239 -.792 1.000
Sig. (2-tailed) .162 .676 .453 .560 .041 .056 .000 .
Triple Cream Correlation Coefficient -.204 -.213 .413 .423 -.026 -.419 -.553 .023 1.000
Sig. (2-tailed) .103 .089 .001 .000 .835 .001 .000 .856 .
N = 65
330
Table A 76 High self-confidence and average importance brie
Self
confidence
brie brie COO price
Sig. (2-tailed) .
Sig. (2-tailed) .266 .154 .
Average importance COO Correlation Coefficient -.047 -.004 -.168 1.000
Sig. (2-tailed) .704 .976 .174 .
Sig. (2-tailed) .069 .052 .358 .000 .
Sig. (2-tailed) .023 .105 .508 .003 .000
N = 67

Objective knowledge brie Correlation Coefficient .078 -.327 .159 .164 -.051 -.168 -.250 .063 .356
Sig. (2-tailed) .533 .007 .199 .184 .684 .175 .041 .614 .003
Subjective knowledge Correlation Coefficient .021 -.312 .173 .112 .075 -.188 -.249 .109 .285
brie Sig. (2-tailed)
.868 .010 .161 .368 .546 .128 .042 .379 .019
Self confidence Correlation Coefficient .040 .035 -.254 -.019 -.084 .044 .155 -.191 -.078
Sig. (2-tailed) .750 .777 .038 .880 .501 .721 .209 .121 .528
Sig. (2-tailed) .
Sig. (2-tailed) .000 .029 .
Sig. (2-tailed) .131 .031 .488 .
Sig. (2-tailed) .185 .718 .502 .001 .
Sig. (2-tailed) .556 .012 .297 .000 .381 .
Full Cream Correlation Coefficient -.135 .224 -.022 -.145 .016 .183 1.000
Sig. (2-tailed) .277 .068 .861 .242 .898 .139 .
Double Cream Correlation Coefficient .096 -.121 -.014 .157 .081 -.228 -.762 1.000
Sig. (2-tailed) .440 .331 .910 .206 .515 .064 .000 .
Triple Cream Correlation Coefficient .150 -.246 .081 .119 -.150 -.057 -.768 .253 1.000
Sig. (2-tailed) .227 .045 .513 .339 .226 .649 .000 .039 .
N = 67
331
Table A 78 Low self-confidence and average importance brie
Self
confidence
brie brie COO price
Sig. (2-tailed) .
Sig. (2-tailed) .463 .911 .
Average importance COO Correlation Coefficient .163 .105 -.051 1.000
Sig. (2-tailed) .173 .383 .675 .
Sig. (2-tailed) .700 .405 .749 .000 .
Sig. (2-tailed) .814 .412 .895 .009 .000
N = 71

Objective knowledge brie Correlation Coefficient -.122 .067 .172 .016 .111 -.047 -.224 .219 .128
Sig. (2-tailed) .310 .577 .152 .896 .358 .700 .060 .067 .288
Subjective knowledge brie Correlation Coefficient -.010 -.002 .015 .146 -.069 -.115 -.293 .232 .251
Sig. (2-tailed) .933 .988 .898 .226 .570 .339 .013 .052 .035
Self confidence Correlation Coefficient -.119 .164 .043 .115 -.134 .000 .081 -.104 -.004
Sig. (2-tailed) .323 .171 .720 .340 .265 .999 .503 .386 .975
Sig. (2-tailed) .
Sig. (2-tailed) .000 .043 .
Sig. (2-tailed) .939 .265 .082 .
Sig. (2-tailed) .884 .597 .530 .002 .
Sig. (2-tailed) .661 .218 .124 .000 .062 .
Sig. (2-tailed) .157 .003 .104 .012 .455 .000 .
Sig. (2-tailed) .177 .014 .156 .025 .633 .006 .000 .
Triple Cream Correlation Coefficient .135 -.290 .118 .204 .081 -.314 -.726 .209 1.000
Sig. (2-tailed) .262 .014 .328 .088 .502 .008 .000 .080 .
N = 71
332
Appendix
10 Invitations to participate in survey
333
10.1 Electronic invitation to participate
Hello all!
Its Roberta Veale here from Business Studies (marketing) on the 2nd floor of
the City Campus. Im working on the final stage of my PhD study and invite
anyone over the age of 18 years to complete my survey booklet and, for
helping out, you receive a free bottle of wine.
The survey only takes about 15 minutes! If youd like to get one for yourself
(and if you take one for a friend or partner youll get 2 bottles!) drop me an
email. I really appreciate anybody taking the time to participate but please
only take a booklet if you absolutely intend to return it. (Thanks)
More info about my study below:
School of Commerce
When you return the questionnaire you will receive a free

bottle of Scarpantoni School Block red wine

Were doing some basic research to gain an understanding of how people make
purchasing decisions.
Contact Roberta Veale on Roberta.Veale@tafesasouth.org or Professor Pascale Quester on
pascale.quester@adelaide.edu.au to receive further information about this project.
Consent
By completing and returning the questionnaire, you have consented to participating in this
research.
How do I get a questionnaire?

Just reply to this email and Ill get one to you. This invitation is extended only to those over 18
years.
Info about the wine and how to get it
334
(electronic invitation cont)
School Block
Produced from a blend of Shiraz, Cabernet and Merlot, with each
variety individually processed and aged in oak for 12 months, the
School Block is soft and supple, with rich fruit and hints of toasty oak.
Enjoy with tomato based pasta sauces, succulent veal and lamb dishes
or sipping casually with cheese and crackers. Ideal drinking now or
cellaring for up to 8 years.
Gold-International Wine Challenge Silver-McLaren Vale Wine Show

Silver-Royal Perth Wine Show Silver-Royal Hobart Wine Show Bronze-Royal
Adelaide Wine Show Bronze-Royal Melbourne Wine Show
Bronze-New Zealand International Wine Show
How do I get my wine when Ive finished completing the survey?
You simply need to stop by the collection point on the second floor of the TAFESA
City Campus. This desk will be located on your right as you reach the top of stairs
leading up from the Atrium on the ground floor. Look towards the Eastern section of
the building and youll see an area clearly signposted to receive your completed
survey booklet and distribute your free wine to you.
The wine will be available for pick up between the hours of 5:30 pm and 7:30 pm on
the following dates:
Monday 30th October Valued at
Tuesday 31st October $15 per bottle!
Wednesday 1st November
Thursday 2nd November
Or by prior arrangement by contacting Roberta Veale on
0404833924
Cheers!
Roberta
335
10.2 Class room invitation to participate
(transparency shown by lecturers)
School of Commerce
PhD Research Project - Food Preferences
Complete the research questionnaire and

You will receive a free bottle of
Scarpantoni School Block red wine!
Why is the research being conducted
Its hoped the info may lead to production of better

products in the market place
How long does it take?

Only about 15 minutes, and, you are not identified,
so your answers remain totally confidential
Do I need to be a wine expert?

Absolutely not! We are interested in the opinions of a
Valued at
general cross section of the community. $15 per
bottle!
Your level of expertise is not important
How do I participate and receive my free wine?

Just complete the survey and when you return it to the drop off
point at the City Campus and you will be given your wine. Easy!
Information about where and when is provided in the survey

booklet.
Can I take a questionnaire for a friend?

Certainly, the more the merrier! One bottle per questionnaire
returned (BUT, all participants must be over the age of 18)
And, please only take a questionnaire if you absolutely intend to

complete it the data is a critical aspect of the PhD research.
336
Appendix
11 Conjoint survey questionnaire
337
School of Commerce
We would like to invite you to participate in this survey
(approx 15 min to complete)
When you return this questionnaire you will

receive a free bottle of Scarpantoni School Block red wine

Were doing some basic research to gain an understanding of how people make purchasing decisions.
Contact Roberta Veale on Roberta.Veale@tafesasouth.org or Professor Pascale Quester on
pascale.quester@adelaide.edu.au to receive further information about this project.
Consent
By completing and returning the questionnaire, you have consented to participating in this research.
Please detach this sheet and retain it for your later reference. This invitation is extended only to those over 18 years.
Lets get started!

Please turn the page and read the instructions carefully
Instructions for Completing the Questionnaire

Please work quickly and steadily, ensuring that you complete ALL questions in each section before
moving to the next. Its very important that you keep moving forward through the questionnaire, finish
each part, then move to the next dont go back! Also, please dont partially complete any section,
stop for a time, and return to complete it later. Once you commence the survey, please ensure you
can continue until all questions are answered. You can expect to spend about 15 minutes to complete
the entire questionnaire.
Its also critically important that you keep this questionnaire, and your answers, confidential, do not
show anyone else the questions or seek advice in providing answers. Were only interested in your
own personal responses.
This Questionnaire is comprised of 3 parts:

Part 1
This includes 4 sections; the first (a) asks you to state your level of agreement, or disagreement, with a number
of statements involving common social situations. Please remember that there are no right or wrong answers
here! Please try to answer honestly, with your answers reflecting what is (or would be) true for you in these
circumstances.
Sections (b) and (c) communicates your general experience and knowledge in relation to the two food products
being investigated, namely Brie cheese and Chardonnay wine. The last section (d) records demographic
details such as gender, age and so on.
338
Part 2
This part is comprised of 11 different descriptions of chardonnay wine and 11 different descriptions of brie
cheese, presented in a mixed order. Please consider each product example and grade it for quality. Then
please indicate whether or not you would consider buying the example as described if you were shopping for
this type of product.
These products come from different countries and youll notice there are no Australian examples included.
Whilst we realise that most Australians would prefer Aussie wine and cheese, sometimes it may not be possible
for you to buy Australian. So, wed like you to imagine a shopping situation where Australian products are not
available, for example when travelling overseas or when in a specialty store dealing exclusively with imported
products.
Also, we realise that you may not be particularly fond of chardonnay or brie; in fact, personally you may not like
these products at all! But the task is to assess each example and provide an opinion of quality and this can be
done even if the product is not for you. We often find ourselves in a situation where we need to buy something
(or we are asked to buy something) that is not our favourite type of product or even for our own use. For
example, you may have people coming over and want to provide a variety wines and cheeses for them, or your
partner or some other person has asked you to make a purchase for them. So please just imagine that youre
on a shopping mission and base your assessment on being in that kind of situation.
Remember, we want your first impressions so work quickly and steadily. There are no right or wrong answers,
we are only interested in your personal opinions. Again, please remember that you need to keep moving
forward - Do not go back and change answers!
Part 3
This part asks you some general knowledge questions about wine and cheese. The format of the questions is
multiple choice, allowing you to choose from the answers provided.
Please go to the next page and begin!
Part 1 (a)
yourself. Your answers will only appear in aggregate and average numbers.
Please (3) the box that best reflects your agreement Strongly Strongly
with the following statements. Disagree Agree
1 2 3 4 5 6 7 8 9

3.
5. I dont make a very favourable first impression on people.
6.
7.
wrong thing.
9.
laughed at.
10.
know.
Please turn the page and continue with Part 1 (b)
339
Part 1 (b)
In this part of the questionnaire, we would like to ask you some questions about your general knowledge
and experiences with purchasing Chardonnay.
and experience. Disagree Agree
1 2 3 4 5 6 7 8 9
average)
Please turn the page and continue with Part 1 (c)
Part 1 (c)
In this part of the survey, we would like to ask you some questions about your general knowledge
and experiences with purchasing Brie.
and experience about Brie. Disagree Agree
1 2 3 4 5 6 7 8 9
1. I feel confident about my knowledge of Brie.
2. I feel that I know how to judge the quality of Brie.
3. I do not feel very knowledgeable about Brie.
4. Among my circle of friends, Im one of the experts on Brie.
5. Compared to most buyers, I know less about Brie.
6. I know of most of the Brie cheeses in the shops.
7. When it comes to Brie, I really dont know a lot.
8. I can tell if a Brie is worth the price or not.
How often would you purchase Brie for yourself If you have a favourite brand of Brie, please list
9. 10.
or others? (on average) it below.
Please turn the page and continue with Part 1 (d)
340
Part 1 (d)
For the purpose of analysis, we would be grateful if you would provide some personal information
about yourself. Your anonymity is guaranteed, this data is only used for statistical purposes.

46 yrs to 55 yrs 4
Over 55 yrs 5
(gross) completed:
Please turn the page and begin the assessment of wine and cheese product examples
All Wine options are 2005 vintage and packaged in 750 ml bottles.
All Cheese prices are per kilo as would be seen in a deli or specialty cheese shop.
Part 2 - Chardonnay and Brie Products (11 descriptions of Chardonnay and 11 of Brie Cheese)
Consider these products and rate each on the scale where 1 = Low Quality and 10 = High Quality.
Please also indicate whether or not you would consider purchasing each product, in a shopping situation.
There are no right or wrong answers, we are only interested in your opinion of likely quality.
Assess each product individually, do not go back and change answers!
Chardonnay 253
Produced in France
Retail Price $16.00

1 2 3 4 5 6 7 8 9 10

341
Chardonnay 582
Retail Price $6.00
1 2 3 4 5 6 7 8 9 10

Chardonnay 481
Acidity High
Produced in Chile
Retail Price $16.00
1 2 3 4 5 6 7 8 9 10

342
Brie 810
Made with Full cream
Produced in France
1 2 3 4 5 6 7 8 9 10

Brie 139
Made with Triple cream
1 2 3 4 5 6 7 8 9 10

343
Brie 367
Produced in Canada
Retail Price $69.95
1 2 3 4 5 6 7 8 9 10

Chardonnay 696
Acidity High
Produced in France
Retail Price $6.00
1 2 3 4 5 6 7 8 9 10

344
Chardonnay 595
Retail Price $16.00
1 2 3 4 5 6 7 8 9 10

Chardonnay 924
Acidity Above average
Produced in Chile
Retail Price $53.00
1 2 3 4 5 6 7 8 9 10

345
Brie 266
Made with Double cream
Produced in Canada
1 2 3 4 5 6 7 8 9 10

Brie 709
Produced in France
Retail Price $49.95
1 2 3 4 5 6 7 8 9 10

346
Brie 380
Retail Price $69.95
1 2 3 4 5 6 7 8 9 10

Chardonnay 152
Retail Price $6.00
1 2 3 4 5 6 7 8 9 10

347
Chardonnay 823
Produced in France
Retail Price $53.00
1 2 3 4 5 6 7 8 9 10

Chardonnay 494
Produced in Chile
Retail Price $6.00
1 2 3 4 5 6 7 8 9 10

348
Brie 735
Produced in France
1 2 3 4 5 6 7 8 9 10

Brie 393
1 2 3 4 5 6 7 8 9 10

349
Brie 621
Made with Double
Produced in France
Retail Price $69.95
1 2 3 4 5 6 7 8 9 10

Chardonnay 950
Produced in France
Retail Price $16.00
1 2 3 4 5 6 7 8 9 10

350
Chardonnay 279
Acidity High
Retail Price $53.00
1 2 3 4 5 6 7 8 9 10

Brie 178
Produced in Canada
Retail Price $49.95
1 2 3 4 5 6 7 8 9 10

351
Brie 507
Made with Double cream
Retail Price $49.95
1 2 3 4 5 6 7 8 9 10

That concludes this part of the questionnaire

Please turn the page and begin the last part comprised of general knowledge questions.

wine. Please (3) the box that best reflects your knowledge and experience. If youre unsure about the answer
to any particular question, thats no problem just (3) the box indicating this.

352
7. Chardonnay, typically has an aging potential 8. Terms often linked with the taste of Chardonnay
of: are:
crushed in the year named, if a Vintage date combinations, it is important to remember that:
is given?
wines:
Colour 1 Improve the taste of a poor wine 1
Please turn the page and continue with Part 3 (b)
Part 3 (b) Brie

Please (3) the box that best reflects your knowledge and experience with Brie. If youre unsure about the
answer to any particular question, thats no problem just (3) the box indicating this.
3. Rennet is: 4. Brie:
A form of yeast used to make cheese. 1 Is eaten younger than Camembert 1
An enzyme extracted from stomach linings. 2 Is eaten older than Camembert 2
Neither of the above. 3 Matures in the same time as Camembert 3
5. Rind is: 6. Brie is nicest when made from milk produced:
An important influence on flavour development 1 In the spring and autumn 1
Not found on Brie 2 In the winter and summer 2
Never washed in making quality cheeses 3 In particularly wet summers 3
7. Brie: 8. Brie:
Never smells of mushrooms and yeast 1 Has slightly thicker mould than Camembert 1
Usually smells of mushrooms and yeast 2 Has slightly thinner mould than Camembert 2
Can sometimes smell like old socks 3 Has identical mould to Camembert 3
353
Part 3 (b) Brie (cont)
9. Triple Cream Brie: 10. Brie rind is:
Is made from milk with extra cream added later 2 Comprised of grey or white mould 2
Doesnt exist as a product 3 Sometimes orange 3
11. When choosing wine to drink with Brie: 12. Brie cheese:
Its critical to make sure its well chilled 1 Is a product where price usually influences quality 1
A chardonnay will suit, but never a red 2 Is a product where price is no indication of quality 2
Is a product where all brands cost pretty much the
A full bodied, earthy red is a sound choice 3 3
same, no matter where the cheese comes from
13. The inside of a Brie that is ready to eat is: 14. When you press the rind of a Brie it should:
That completes the questionnaire, thank you very much for your help!
Please see over for information about your free bottle of wine and how to receive it!
School Block
Produced from a blend of Shiraz, Cabernet and Merlot, with each variety individually processed
and aged in oak for 12 months, the School Block is soft and supple, with rich fruit and hints of
toasty oak. Enjoy with tomato based pasta sauces, succulent veal and lamb dishes or sipping
casually with cheese and crackers. Ideal drinking now or cellaring for up to 8 years.
Gold-International Wine Challenge Silver-McLaren Vale Wine Show

Silver-Royal Perth Wine Show Silver-Royal Hobart Wine Show Bronze-Royal Adelaide Wine Show
Bronze-Royal Melbourne Wine Show Bronze-New Zealand International Wine Show
How do I get my wine when Ive finished completing the survey?

You simply need to stop by the collection point on the second floor of the TAFESA City Campus. This desk will be located
on your right as you reach the top of stairs leading up from the atrium on the ground floor. Look towards the Eastern
section of the building and youll see an area clearly signposted to receive your completed survey booklet and
distribute your free wine to you.
The wine will be available for pick up between the hours of 5:30 pm and 7:30 pm on the following dates:
Monday 13th of November
Wednesday 15th of November
Thursday 16th of November
Or by prior arrangement by contacting Roberta Veale on 0404833924
Valued at
over $15!
354

Swarn Tantram

Uploaded by

Copyright:

Available Formats

Swarn Tantram

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Swarn Tantram

Uploaded by

Copyright:

Available Formats

Investigating the simultaneous influence of

intrinsic and extrinsic cues:

An examination of the interaction between

country of origin, price and selected sensory variables

A thesis submitted in fulfillment of

the requirements for the degree of

influences of objective knowledge, subjective knowledge and self-confidence on these

profile conjoint analysis experimental design to measure quality perception, followed by a

stages show findings that are remarkably similar in most respects.

strongest influence, with comparable expectations regarding the importance of country of

exhibited reasonably healthy degrees of self-confidence. The research provides important

and platform for a number of subsequent studies.

co-supervisor, Dr Amal Karunaratna, has also been a great source of knowledge,

encouragement and ideas. He was particularly instrumental in the development of the

management. Secondly, my work mates at the School of Commerce (University of

This research was made possible by funding by the

Wine and Grape Research Development Board.

TABLE OF CONTENTS iii

List OF FIGURES xxi

1.1 BACKGROUND TO THE RESEARCH 1

1.2 RESEARCH JUSTIFICATION 3

1.3 OBJECTIVES AND CONTRIBUTION OF RESEARCH 5

1.4 THESIS OUTLINE 6

2.1 CONSUMER USE OF INTRINSIC AND EXTRINSIC CUES 9

2.1.1 Extrinsic cues and sensory perceptions 11

2.2 COO AS AN EXTRINSIC CUE 12

2.2.1 Introduction to the COO construct 12

2.2.2 How country images are developed 15

2.2.3 Building country equity through international brands 17

2.2.4 Consumer bias against products from developing countries 18

2.2.5 The challenge of marketing 'hybrid' products 21

2.2.6 Effect of market specific attributes on use of CI 24

2.2.7 Effect of CI on commercial customers 28

2.2.8 Effect of consumer knowledge on use of CI 29

2.3 PRICE AS AN EXTRINSIC CUE 35

2.3.1 The role of price in consumer buying decision making 35

2.3.2 Effect of consumer knowledge on use of price 37

2.3.3 Price summary 37

2.4 CONSUMER EXPERTISE AND SELF-CONFIDENCE 38

2.4.1 Two dimensions of knowledge 38

2.4.1.1 Objective knowledge 40

2.4.1.2 Subjective knowledge 42

2.4.2 Consumer self-confidence 44

2.4.3 Consumer knowledge and self-confidence summary 45

2.5 GAPS IN THE EXISTING LITERATURE 46

2.5.1 Towards a conceptual framework 47

3 CAUSAL MODEL AND HYPOTHESIS DEVELOPMENT 49

3.2 RESEARCH FRAMEWORK 49

3.2.1 Testing the power of extrinsic cues 49

3.2.2 Testing the influence of consumer characteristics 50

3.2.3 Use of sensory experiments 51

3.2.4 Suitability of conjoint analysis experimental design 51

3.2.4.1 Formulating a conjoint analysis design 52

3.2.5 Expected quality vs perceived quality 54

3.2.6 Development of causal model and hypotheses 54

3.2.6.1 Causal model and hypotheses 54

3.2.7 Research paradigm 58

3.2.8 Justification for the model 59

3.2.9 Stimuli used 60