Anthrozoös

A multidisciplinary journal of the interactions of people and animals

ISSN: 0892-7936 (Print) 1753-0377 (Online) Journal homepage: http://www.tandfonline.com/loi/rfan20

Psychometric Evaluation of the Lexington

Attachment to Pets Scale (Laps)

Timothy P. Johnson, Thomas F. Garrity & Lorann Stallones

To cite this article: Timothy P. Johnson, Thomas F. Garrity & Lorann Stallones (1992)
Psychometric Evaluation of the Lexington Attachment to Pets Scale (Laps), Anthrozoös, 5:3,
160-175

To link to this article: http://dx.doi.org/10.2752/089279392787011395

Published online: 27 Apr 2015.

Submit your article to this journal

Article views: 4

View related articles

Citing articles: 40 View citing articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=rfan20

Download by: [University of Windsor] Date: 25 September 2015, At: 23:31

 REVIEWS AND RESEARCH
REPORTS
PSYCHOMETRIC EVALUATION OF INTRODUCTION
THE LEXINGTON ATTACHMENT Eight years ago, a review of the literature
TO PETS SCALE (LAPS) concerned with the role that pet attach-
ment may play in the maintenance of hu-
man health concluded that most of the
Timothy P.Johnson,1 available evidence was lacking in scien-
Thomas F.Garrity,2 tific rigor (Marx 1984). Since then, the re-
and Lorann Stallones3 search community has taken an active in-
Downloaded by [University of Windsor] at 23:31 25 September 2015

terest in the hypothesized link between

human-animal interaction and both physi-
Abstract. This paper reports on the devel- cal and psychological well-being (Culliton
opment and psychometric evaluation of a 1987; Department of Health and Human
scale for assessing emotional attachment Services 1988). Evidence has been re-
of individuals to their pets. Previous attach- ported that pet ownership may be benefi-
ment scales have suffered variously from cial in both younger (Poresky et al. 1987)
low internal consistency and reliance on and older populations (Akiyama,
small or nonrepresentative samples for Holtzman, and Britz 1986–87; Bolin 1987;
their development. Telephone interviews Lago, Connell, and Knight 1983; Ory and
of a random, representative sample of 412 Goldberg 1983).
pet owners in Fayette County, Kentucky, Studies by our research group have
were completed in September 1990; a found that emotional attachment to com-
69.5 percent response rate was achieved. panion animals was associated with de-
From a preliminary set of 42 questions, a creased reports of depressive symptoms
final 23-question instrument, the Lexington among elderly respondents (Garrity et al.
Attachment to Pets Scale (LAPS), was de- 1989). Using a similar methodology, we
veloped, having excellent psychometric also studied a national probability sample
properties. The scale is suitable for use of adults aged 21–64. We again observed
with dog and cat owners. Data on internal a significant relationship between pet at-
consistency, factor structure, and item re- tachment and depressive symptoms,
sponse theory (IRT) modeling are pre- but only among those aged 35–44, and
sented, along with correlations between this association was reduced to
the LAPS and several domains of variables nonsignificance when other respondent
known to relate to pet attachment. characteristics were controlled (Stallones
et al. 1990). The research conducted to
date suggests that, although some evi-
1
University of Illinois, P.O. Box 6905, Chicago, IL 60680.
2
University of Kentucky, Lexington, KY. dence does support the hypothesis that
3
Colorado State University, Fort Collins, CO. pets contribute to emotional health, the
relationship between pet attachment and
This research was supported by a grant from the Delta
Society and the American Pet Products Manufacturers well-being is complex.
Association to the authors.

160 ANTHROZOÖS, Volume V, Number 3 Johnson, Garrity, and Stallones

 Research undertaken since Marx’s development of a new pet attachment
(1984) review reflects a marked instru ment, to be referred to as the
improvement in the quality and rigor of Lexington Attachment to Pets Scale (LAPS),
the methods used. However, additional that has psychometric properties superior
efforts are needed to more adequately to our previous efforts.
measure the concept of pet attachment.
After examining available instruments, METHODS
Wilson, Netting, and New (1987)
concluded that few measures existed for
Data Collection
which adequate evaluations of reliability
and validity have been conducted. Data were collected via a telephone sur-
Similarly, Poresky (1988) argued that vey conducted in Fayette County, Kentucky,
“there is a clear need to improve the utility a metropolitan area with a population of
of the measures used to assess the human- approximately 250,000. The survey sample
Downloaded by [University of Windsor] at 23:31 25 September 2015

animal relationship.” During this period, was generated using random-digit dialing,
several important measures were a methodology that gives all residential tele-
developed to address many of these phone numbers in the geographic area of
concerns. Prominent among these were interest an equal probability of selection
the Companion Animal Bonding Scale (Waksberg 1978). Each number contacted
(Poresky et al. 1987), the Companion was screened to determine if there was a
Animal Semantic Differential (Poresky et pet in the household. Within those house-
al. 1988), the Pet Attitude Inventory holds with a pet, an adult respondent (18+)
(Wilson, Netting, and New 1987), the Pet was selected using a random respondent
Attitude Scale (Holcomb, Williams, and selection procedure. Call-back appoint-
Richards 1985), and the Pet Relationship ments were made whenever the selected
Scale (Lago et al. 1988). Most of the respondent was unavailable. In addition,
developmental work these scales were selected individuals refusing to be inter-
based upon, however, involved viewed were called back at a later date and
nonrandom convenience samples. Our asked to reconsider participation. In all,
group has reported the development of a 412 interviews were completed, and the
six-item index of pet attachment that was survey’s response rate (completed inter-
evaluated using data from a national views/total eligible) was 69.5%.
sample of the elderly (Stallones et al. All interviewing was conducted by
1988). This measure was found to have a professional interviewers employed by the
moderate level of internal consistency. University of Kentucky Survey Research
During a subsequent study, we developed Center. Interviewing was 100 percent
an eight-item scale that improved upon supervised, and a sample of each
our earlier measure (Stallones et al. 1990). interviewer’s work was monitored for
Although these efforts were promising, accuracy. Completed questionnaires were
given that the work was based upon edited prior to data entry. All interviewing
national random probability samples, we was conducted during September 1990.
have been concerned that the available
measures remain imperfect indicators of
Instrument Development
the degree of affection that may exist
between individuals and their companion Pet attachment items were developed
animals. This article reports on the based upon our review of a number of

The Lexington Attachment to Pets Scale ANTHROZOÖS, Volume V, Number 3 161

 sources. We adapted a number of items The sequence in which items were asked
from the two previous attachment scales was randomized in an effort to minimize
developed by our research team. The work order effects (Groves 1989).
of other investigators, including the Com- In addition to the candidate items for the
panion Animal Bonding Scale (Poresky et attachment scale, the questionnaire also
al. 1987), the Pet Attitude Inventory (Wil- contained information regarding types of
son, Netting, and New 1987), and the Pet pets in the household, favorite type of pet,
Attitude Scale (Templer et al. 1981), were and care of favorite pet. These items were
also examined. In developing candidate positioned in the survey instrument prior to
items we made an effort to include ques- the candidate LAPS items. The final section
tions that might be representative of weak of the instrument contained several baseline
attachment to pets. Previous analyses have demographic items, including age, race,
reported an insufficient number of items sex, education, marital status, and income.
that measure weak attachment (Stallones Finally, interviewers were asked to record
Downloaded by [University of Windsor] at 23:31 25 September 2015

et al. 1990), suggesting that the full range their own opinion of how attached the re-
of human-animal bonds may not have spondent was to his or her pet.
been adequately assessed.
The development of items for this scale
Analysis Methods
was also guided by theoretical consider-
ations. Research into the association be- Items to be included in the LAPS were se-
tween human social support and health lected using both traditional item analysis
suggests that emotional ties is the dimen- procedures (Kline 1986) and item response
sion of support that is most closely related theory (IRT) models (Hambelton and
to health (House and Kahn 1985). Several Swaminathan 1985). The traditional proce-
researchers have suggested that a similar dures utilized included calculation of al-
mechanism may underlie human-pet rela- pha internal consistency coefficients
tionships (Garrity et al. 1989; Lago et al. (Cronbach 1951), examination of
1988; Ory and Goldberg 1983). Although itemtotal-scale correlations, and factor
a variety of candidate items were devel- analysis techniques to examine the scale’s
oped for this study, we emphasized ques- dimensionality. For these analyses all vari-
tions that appeared to be indicative of re- ables were coded in a continuous manner.
spondent affection for the companion ani- The response categories for each item
mal. Our theoretical approach makes the were coded as follows: 0=strongly dis-
assumption that, as with human relations, agree; 1=somewhat disagree; 2=somewhat
it is this aspect of owner-pet relations that agree; and 3=strongly agree. This coding
is most closely related to well-being. scheme was reversed for items worded in
A total of 42 attachment items were a negative direction (items h and u in
included in the instrument. These items Table 1).
were introduced by the following The IRT model utilized was the two-
statement: parameter binary logistic IRT model. Our
decision to use this particular model was
guided by the fact that it had been
I’d like to ask you whether you agree or disagree
successfully utilized in previous efforts to
with some very brief statements about your
favorite pet. For each statement, please tell me apply IRT methods for similar applications
whether you strongly agree, somewhat agree, (Duncan-Jones, Grayson, and Moran
somewhat disagree, or strongly disagree. 1986; Reiser 1989; Schaeffer 1988).1 This

162 ANTHROZOÖS, Volume V, Number 3 Johnson, Garrity, and Stallones

Downloaded by [University of Windsor] at 23:31 25 September 2015

procedure models responses to individual point at which each respondent has a 50–
items as being a function of item 50 chance of item endorsement. By
discrimination and threshold. Discrimina- ranking item thresholds, we will be able to
tion is the extent to which each item is observe how adequately the items in the
associated with the underlying trait of LAPS cover the range of affective ties that
interest, pet attachment in this case. Item pet owners feel for their animals. This work
discrimination is also referred to as item was accomplished using the BILOG 3
slope and can be considered the reliability program (Mislevy and Bock 1990).
with which each item measures the scale’s The IRT analysis required that the four
underlying characteristic (Schaeffer 1988). response categories of each scale item be
Higher discrimination is indicative of collapsed into binary values. In examining
greater reliability. Threshold is the item statistics, it was found that a plurality
probability that a given individual will of respondents endorsed the response
endorse each item. It is considered the category reflective of highest attachment

The Lexington Attachment to Pets Scale ANTHROZOÖS, Volume V, Number 3 163

Downloaded by [University of Windsor] at 23:31 25 September 2015

for a majority of the items (see Table 2). As disagreed. Otherwise, each item was
a consequence, each item was recoded for recoded as 0, reflecting an absence of
the IRT analysis such that each indicated strong attachment. It is important to
whether or not the respondent’s answer emphasize that these attachment items
represented strong attachment to the pet. were dichotomized for IRT analysis only.
This was accomplished by recoding to a The two-parameter IRT model was also
value of 1 items worded in a positive used to examine the structure of the LAPS
direction if the respondent strongly agreed across two groups of pet owners (dog
with them and items worded in a negative versus cat owners). In analyzing pet
direction if the respondent strongly attachment scales using heterogeneous

164 ANTHROZOÖS, Volume V, Number 3 Johnson, Garrity, and Stallones

 samples of pet owners, the assumption is RESULTS
commonly made that the scale is equally
efficient in discriminating levels of The final sample for this study was 59%
attachment for all pet owners. By female, and 94% white. The average age
conducting separate analyses for various was 43.0 years, and ages ranged from 18
groups, we will be able to determine if to 83 years. The typical respondent had 12
there are large differences in the ability of years of education (range 5 to 25 years)
individual scale items to discriminate and reported annual family income of be-
differently among dog and cat owners tween $20,000 and $30,000. In regard to
(other types of pet owners were not marital status, 62% were married, 12%
considered in this part of the analysis had never been married, 4% were
because of small numbers). Large cohabitating, 13% were separated or di-
differences in the characteristics of various vorced, and 8% were widowed; 87% re-
questions across type of pet ownership ported having grown up with pets; 63% of
Downloaded by [University of Windsor] at 23:31 25 September 2015

would, of course, make it necessary to the respondents indicated that dogs were
eliminate them from our final scale. their favorite pets, and 28% said that cats
Finally, preliminary exploration of the were their favorite. Smaller percentages
construct validity of the LAPS using other said other types of pets were their favorite
available data was examined. As part of (4%), and 5% were unable to specify
this effort, responses to the LAPS were which of their pets was their favorite. An
correlated with interviewer assessments of average of 16 minutes was required for
each respondent’s attachment to his or her interviewers to complete each question-
favorite pet. After each survey was naire.
completed, interviewers were asked to Internal consistency of the 42
rate their subjective assessment of each attachment questions was estimated using
respondent’s attachment to his or her coefficient alpha. The coefficient for these
favorite pet as being “very attached, items was 0.937, indicating a high degree
somewhat attached, not very attached, or of internal consistency. The purpose of this
not at all attached.” Although interviewer study was to develop a scale with a more
assessments were undoubtedly based in practical (i.e., smaller) number of items.
large part on the respondent’s answers to Therefore the corrected item-total
the LAPS items, they also reflect an correlations for each of the 42 questions
additional assessment of the respondent’s were examined, and 24 questions were
relationship with the favorite pet. In found to have values greater than 0.50.
addition, oneway analysis of variance These 24 questions were subsequently
(ANOVA) was used to examine included in a two-parameter IRT analysis
relationships between the LAPS and three to determine the extent to which each
categories of respondent characteristics for question provided a reasonable “fit” to the
evidence of construct validity: respondent latent concept of “pet attachment.” A good
demographics, social network ties, and fit is indicated by low (nonsignificant)
other characteristics related to respondent- chi-square values. Although the likelihood
pet interaction. Variables included within ratio chi-square for the overall model
each of these three categories have been suggested an acceptable fit (G2=102.5,
previously reported to be associated with df=93, P=.235), one question had a
pet attachment (Kidd and Kidd 1989; significant chi-square value. The analysis
Serpell 1981; Stallones et al. 1990). was repeated with this item eliminated,

The Lexington Attachment to Pets Scale ANTHROZOÖS, Volume V, Number 3 165

Downloaded by [University of Windsor] at 23:31 25 September 2015

resulting in an acceptable fit both for the question are provided in Table 2. The scale
model as a whole (G 2 =72.2, df=88, had a range of values from 0 to 69. The
P= .739), and for each of the remaining 23 mean value for the 322 respondents who
questions included in the model. answered all questions was 47.99
The alpha coefficient for these 23 (standard deviation=12.65, median
questions (0.928) was close to the value=50, modal value=54). Based upon
coefficient for the entire set of questions, the findings of these analyses, we further
indicating that the number of items in the investigated the properties of the 23-item
scale could be reduced by nearly half scale, which will subsequently be referred
without a noticeable loss in reliability. The to as the LAPS.
wording of each question is shown in Table The parameters obtained from the final
1, and frequency distributions for each IRT analysis are presented in Table 3. The

166 ANTHROZOÖS, Volume V, Number 3 Johnson, Garrity, and Stallones

Downloaded by [University of Windsor] at 23:31 25 September 2015

units in which these parameters are pre- pet attachment is more adequately cov-
sented are to a large extent arbitrary. A ered by scale items at the high attachment
common practice is to center the mean end of the spectrum (i.e., those items with
threshold value to 0 and the estimated positive threshold values). Although the
variance to 1. Scales calibrated in this remaining nine items in the scale could be
manner have a theoretical threshold range endorsed by individuals with less than av-
from +3 to -3. The thresholds shown in erage levels of pet attachment, these items
Table 3 indicate that 14 of the 23 items in have a more restricted range. In general,
the model are indicators of above-average these findings suggest that the LAPS is
(i.e., strong) pet attachment. These data more successful in measuring strong at-
also indicate that the theoretical range of tachment than weak attachment.

The Lexington Attachment to Pets Scale ANTHROZOÖS, Volume V, Number 3 167

Downloaded by [University of Windsor] at 23:31 25 September 2015

Next, the IRT model was examined results are presented in Table 4. There
separately for respondents who indicated appears to be a reasonable amount of
their favorite pet was a dog and those congruence between the parameters
whose favorite pet was a cat. As estimated for dog versus cat owners, as the
mentioned earlier, this procedure was correlation between the 23-item
undertaken as a first step in assessing thresholds is 0.94. The distribution of the
whether or not there are differences in item threshold values is quite similar,
how individual responses in the LAPS are suggesting the LAPS covers a similar range
affected by types of pets owned. These of affective ties for each group. The largest

168 ANTHROZOÖS, Volume V, Number 3 Johnson, Garrity, and Stallones

Downloaded by [University of Windsor] at 23:31 25 September 2015

difference in item threshold was for item d identified. Examination of the items
(“I believe my pet is my best friend”), loading most highly on each of these
which had an estimated value for cat factors suggested that the first represents a
owners that was more than twice that of “general attachment” dimension. Items
dog owners. This suggests that this item is included in factor 2 indicate the pet in
an indicator of stronger attachment among question occupies a more central position
cat owners. Table 4 also indicates that in the respondent’s life. This factor has thus
most of the items in the LAPS inventory (16 been labeled “people substituting.” The
of 23) are somewhat more reliable third factor contained items that appear to
indicators of pet attachment among cat be mostly concerned with the pet’s status
owners. within the respondent’s household, and
A principal-components analysis was has been designated as the “animal rights/
conducted to examine the structure of the animal welfare” factor. Cronbach’s alpha
LAPS. Results of this analysis are presented coefficients for these three factors were
in Table 5. Using a varimax rotation 0.90, 0.85, and 0.80, respectively.
procedure, three orthogonal factors were The associations between the LAPS and

The Lexington Attachment to Pets Scale ANTHROZOÖS, Volume V, Number 3 169

Downloaded by [University of Windsor] at 23:31 25 September 2015

a number of respondent characteristics with less education and lower income also
were also examined. Table 6 shows the reported stronger attachment.
average pet attachment for five The relationship between pet attach-
demographic items. All were found to be ment and personal social networks was
significantly related to the LAPS. Female next evaluated (Table 7). Several of these
and black pet owners tended to be more variables were also found to vary
highly attached to their favorite pets, as significantly with attachment to pets.
were older respondents. In addition, those Respondents from smaller households

170 ANTHROZOÖS, Volume V, Number 3 Johnson, Garrity, and Stallones

Downloaded by [University of Windsor] at 23:31 25 September 2015

(those with one to two members) and attachment, although this association was
those in households with no children were only borderline significant (P=.08). Those
found to be more attached to their who indicated that a dog was their favorite
animals. Marital status was also related to pet were found to have higher attachment
degree of pet attachment. Those never than those who reported a cat to be their
married, divorced, and who were favorite pet. No relationship was observed
cohabiting expressed the highest between pet care and attachment.
attachment, and those who were married Additional questions (not shown) asked
expressed the least. Inverse relationships about responsibility for feeding and for
between the LAPS and (1) numbers of cleaning up after the pet. No association
persons the respondent felt close to, (2) the was found between these and the LAPS.
number of groups and organizations the Finally, average LAPS scores for various
respondent was active in, and (3) the interviewer ratings of respondent
number of friends and relatives the attachment to pets revealed that a strong
respondent saw monthly were also association existed between respondent
observed. The first two of these variables and interviewer assessments of respondent
were significantly related to the LAPS. attachment to a favorite pet. The
The relationship of the LAPS with other correlation coefficient between these two
pet-related variables is presented in Table variables was 0.64.
8. Respondents who had grown up with The association between these
pets reported greater degrees of demographic, social network, and

The Lexington Attachment to Pets Scale ANTHROZOÖS, Volume V, Number 3 171

 pet-related variables and each of the three scale, as it is strongly correlated with sub-
LAPS factors reported earlier was also jective ratings of respondent pet attach-
examined (Appendices available from first ment that were made by interviewers. Ide-
author). With few exceptions, a pattern of ally, of course, these ratings would have
results very similar to that reported for the been made by individuals unaware of the
total LAPS scale (Tables 6–8) was observed respondent’s answers to the LAPS items—
in these analyses. something that would not have been prac-
tical in this study. Future research will
need to correlate the LAPS with one or
DISCUSSION
more of the other available attachment
Reviews of instruments designed to mea- scales before its construct validity can be
sure various aspects of social support have more firmly established.
concluded that these instruments are far Evidence for the construct validity of the
from adequate indicators of what they pur- LAPS also comes from its association with
Downloaded by [University of Windsor] at 23:31 25 September 2015

port to represent (House and Kahn 1985; other respondent characteristics previously
Leavy 1983; Payne and Jones 1987). Com- demonstrated to be related to pet attach-
pared with the development of measures ment. In particular, gender, education, in-
of affective ties with companion animals, come, marital status, the presence of chil-
however, the social support literature ap- dren in the household, having grown up
pears well established. Indeed, research with pets, and being primarily responsible
into the link between pet attachment and for pets have all been significantly corre-
personal health would appear today to be lated with pet attachment in other studies
where social support research was in the (Kidd and Kidd 1989; Lago et al. 1988;
mid-1970s—with abundant anecdotal in- Stallones et al. 1990). Perhaps most inter-
formation but insufficient empirical sup- esting, though, was the association between
port. the LAPS and indicators of social relation-
The purpose of the research presented ships. For both primary and secondary so-
in this article has been to develop and cial relationships, respondents reporting
evaluate a measure of pet attachment that fewer ties were found to have greater at-
is reliable and for which some preliminary tachment to their pets. To the extent that pets
evidence of validity can be provided. might be expected to play a more central
Previous efforts, reviewed earlier, suffered role in the lives of individuals with smaller
either from less than ideal internal social networks, these results represent evi-
consistency, as measured by coefficient dence of construct validity. They also sug-
alpha, or from reliance on small or gest an important direction for future re-
nonrepresentative samples. The measure search: the degree to which pet attachment
reported here has excellent reliability and is a complement to or a substitute for social
was constructed using data collected from relationships. The answer is most likely some
a random probability sample of pet owners of each. Our investigations, however, do
in a medium-sized metropolitan area. suggest that pet attachment is greater among
Validation of any instrument is an ongo- those with fewer social ties, a finding that
ing process, but the LAPS appears to have appears to lend support to the “substitute”
content validity in that the items all appear hypothesis. Further work is needed in this
to represent some level of attachment to a area.
companion animal. Some initial evidence The IRT analysis demonstrated that the
also exists for the construct validity of this LAPS items do not cover the entire

172 ANTHROZOÖS, Volume V, Number 3 Johnson, Garrity, and Stallones

 potential range of levels of attachment to attachment. The Item thresholds for factor
pets. We were more successful in 3 were somewhat more randomly
developing items that measure strong distributed. These findings suggest that
attachment. This is particularly frustrating items from each factor (or at least factors 1
given the fact that items reflecting lower and 2) are necessary to adequately
levels of attachment were found, on represent the overall construct. We
average, to be more reliable indicators of therefore conclude that attempts to
attachment, as evidenced by higher develop a unidimensional measure of pet
discrimination parameters. Yet, the LAPS attachment would not be appropriate.
items do provide a reasonable distribution Several limitations of this research
of attachment levels, and can be used as a should be acknowledged. First, the state-
benchmark for further improvements. ment used to introduce the LAPS items
Although some differences were asks respondents to answer questions re-
observed, it was also useful to determine garding their “favorite” pet. The term “fa-
Downloaded by [University of Windsor] at 23:31 25 September 2015

that the IRT parameters for the 23 LAPS vorite” was included to provide respon-
items were reasonably stable when dents who had more than one pet with a
examined separately for dog versus cat frame of reference for responding. We
owners. Had estimates varied used this approach because our own per-
substantially, it would have called into spective assumes the most salient, or “fa-
question the utility of the instrument for vorite” pet, is most important for the re-
measuring attachment to different types of search purposes we have identified. In
pets. Although other research (Miller and retrospect, directing respondents to an-
Lago 1990) suggests differences in how swer regarding their “favorite” may have a
persons interact with pets of varying social desirability effect, influencing them
species, we nonetheless conclude that the to answer in a more positive manner than
LAPS is a useful indicator of pet they might otherwise. This may in part ac-
attachment for both dogs and cats. count for the inability of the LAPS items to
Because so few other types of pets were assess weak attitudes.
reported to be the respondent’s favorite in In addition, the LAPS has to date been
this study, examination of other groups evaluated using responses collected only
could not be conducted. Questions of via telephone interviews. Although other
applicability to other types of pets must research suggests that attitude scales are
therefore await further investigation. largely robust to variations in data collec-
Most previous attempts to develop a pet tion (van Tilburg and de Leeuw 1991), the
attachment measure have reported the utility of this scale in self-administered and
scale to contain more than one dimension in-person interviews has yet to be estab-
(Poresky et al. 1987; Stallones et al. 1988; lished. Finally, although a random prob-
Templer et al. 1981). The LAPS was found ability sample was used to develop this
to have three well-defined factors: general instrument, the survey was conducted in a
attachment, people substituting, and metropolitan county. Baseline information
animal rights/animal welfare. Not regarding responses from rural inhabitants
surprisingly, examination of the item is therefore unavailable.
thresholds in Table 3 reveals that the items In conclusion, this study has presented
in factor 1 represent lower levels of pet details on the development and
attachment and the items in factor 2 are psychometric evaluation of a measure of
representative of higher levels of pet attachment referred to as the Lexington

The Lexington Attachment to Pets Scale ANTHROZOÖS, Volume V, Number 3 173

 Attachment to Pet Scale (LAPS). As stated response theory: Principles and applications.
earlier, attempts to address the question of Boston: Kluwer-Nijhoff.
Holcomb, R., C.R.Williams, and P.S.Richards.
how emotional attachment to companion 1985. The elements of attachment: Relationship
animals may be linked to personal well- maintenance and intimacy. Journal of the Delta
being have only recently begun. Although Society 2: 28–33.
continued psychometric evaluation of the House, J.S., and R.L.Kahn. 1985. Measures and
LAPS will be necessary, it is our hope that concepts of social support. In Social support and
health, ed. S.Cohen and S.L.Syme. Orlando, FL:
this instrument will contribute to the
Academic Press.
development of research in this field. Kidd, A.H., and R.M.Kidd. 1989. Factors in adults’
attitudes toward pets. Psychological Reports 65:
NOTE 903–10.
Kline, P. 1986. A handbook of test construction:
1. Consideration was also given to utilizing one- Introduction to psychometric design. New York:
and three-parameter IRT models. The one- Methuen.
parameter model was not selected because
Downloaded by [University of Windsor] at 23:31 25 September 2015

Lago, D., C.M.Connell, and B.Knight. 1983. A

it makes the assumption that all items have companion animal program. In Mental Health
equal discriminatory power. Examination of and Aging, ed. M.A.Smyer and M.Gatz. Beverly
these and previously collected data strongly Hills: Sage Publications.
suggested this assumption was not valid. The Lago, D., R.Kafer, M.Delaney, and C.Connell. 1988.
three-parameter model adds a parameter for Assessment of favorable attitudes toward pets:
“guessing,” a factor that we considered to be Development and preliminary validation of the
a negligible consideration for the self-report Self-Report Pet Relationship Scales. Anthrozoös
indicators used in this study. 1: 240–54.
Leavy, R.L. 1983. Social support and psychological
disorder: A review, Journal of Community
REFERENCES
Psychology 11: 3–21.
Akiyama, H., J.M.Holtzman, and W.E.Britz. 1986– Marx, M.B. 1984. The salubrious endearment:
87. Pet ownership and health status during Review of New Perspectives on Our Lives with
bereavement. Omega 17: 187–93. Companion Animals by A.H.Katcher and A.M.
Bolin, S.E. 1987. The effects of companion animals Beck (Eds.). Contemporary Psychology 29:
during conjugal bereavement. Anthrozoös 1: 902–3.
26–35. Miller, M., and D.Lago. 1990. Observed pet-owner
Cronbach, L.J. 1951. Coefficient alpha and the in-home interactions: Species differences and
internal structure of tests. Psychometrika 16: association with the Pet Relationship Scale.
297–334. Anthrozoös 4: 49–54.
Culliton, B.J. 1987. Take two pets and call me in Mislevy, R.J., and R.D.Bock. 1990. BILOG 3: Item
the morning. Science 237: 1560–61. analysis and test scoring with binary logistic
Department of Health and Human Services. 1988. models, 2nd ed. Mooresville, IN: Scientific
Health benefits of pets: Summary of working Software.
group. U.S. Government Printing Office: 2988– Ory, M.G., and E.L.Goldberg. 1983. Pet possession
216–107. and life satisfaction in elderly women. In New
Duncan-Jones, P., D.A.Grayson, and P.A.P.Moran. perspectives on our lives with companion
1986. The utility of latent trait models in animals, ed. A.H.Katcher and A.M.Beck.
psychiatric epidemiology. Psychological Philadelphia: University of Pennsylvania Press.
Medicine 16: 391–405. Payne, R.L., and J.G.Jones. 1987. Measurement and
Garrity, T.F., L.Stallones, M.B.Marx, and T.P. methodological issues in social support. Stress
Johnson. 1989. Pet ownership and attachment and health: Issues in research methodology, ed.
as protective factors in the health of the elderly. S.V.Kasl and C.L.Cooper. Chichester: Wiley.
Anthrozoös 3: 35–44. Poresky, R.H. 1988. Analyzing human-animal
Groves, R.M. 1989. Survey errors and survey costs. relationship measures. Anthrozoös 2: 236–44.
New York: Wiley. Poresky, R.H., C.Hendrix, J.E.Mosier, and M.L.
Hambelton, R.K., and H.Swaminathan. 1985. Item Samuelson. 1987. The Companion Animal

174 ANTHROZOÖS, Volume V, Number 3 Johnson, Garrity, and Stallones

 Bonding Scale: Internal reliability and construct Stallones, L., M.B.Marx, T.F.Garrity, and T.P. Johnson.
validity. Psychological Reports 60: 743–6. 1988. Attachment to companion animals
Poresky, R.H., C.Hendrix, J.E.Mosier, and M.L. among older pet owners. Anthrozoös 2: 118–
Samuelson. 1988. The companion animal 24.
semantic differential: Long and short form Stallones, L., M.B.Marx, T.F.Garrity, and T.P.
reliability and validity. Educational and Johnson. 1990. Pet ownership and attachment
Psychological Measurement 48: 255–60. in relation to the health of U.S. adults, 21–64
Reiser, M. 1989. An application of the item- years of age. Anthrozoös 4: 100–12.
response model to psychiatric epidemiology. Templer, D.J., C.A.Salter, S.Dickey, R.Baldwin, and
Sociological Methods & Research 18: 66–103. D.M.Veleber. 1981. The construction of a pet
Schaeffer, N.C. 1988. An application of item attitude scale. The Psychological Record 31:
response theory to the measurement of 343–48.
depression. In Sociological Methodology 1988, Van Tilburg, T., and E.de Leeuw. 1991. Stability of
Volume 18, ed. C.C.Clogg. San Francisco: scale quality under various data collection
Jossey-Bass. procedures: A mode comparison on the ‘De
Serpell, J.A. 1981. Childhood pets and the influence Jong-Gierveld Loneliness Scale. International
Downloaded by [University of Windsor] at 23:31 25 September 2015

on adults’ attitudes. Psychological Reports 49: Journal of Public Opinion Research 3: 69–85.
118–24. Waksberg, J. 1978. Sampling methods for
Stallones, L., T.P.Johnson, T.F.Garrity, and M.B. Marx. randomdigit dialing. Journal of the American
1990. Quality of attachment to companion Statistical Association 73: 40–7.
animals among U.S. adults, 21 to 64 years of Wilson, C.C., F.E.Netting, and J.C.New. 1987. The
age. Anthrozoös 3: 171-76. Pet Attitude Inventory. Anthrozoös 1: 76–84.

The Lexington Attachment to Pets Scale ANTHROZOÖS, Volume V, Number 3 175