DOCTYPE = ARTICLE

Cultural Validity of Holland’s Theory and

the Strong Interest Inventory for Five
Racial/Ethnic Groups
Nadya A. Fouad
University of Wisconsin-Milwaukee

Carolyn J. Mohler
Colorado State University

This study examined evidence for the use of the Strong Interest Inventory with
diverse populations, including examination of the structure of interests across five
racial/ethnic groups, finding similarity in perceptions of the world of work. The
study also studied differences in means for the General Occupational Themes and
Basic Interest Scales as well as assessed differences in consistency and differentia-
tion across five racial/ethnic groups for career clients. Overall, the pattern of
results suggests minimal differences based on racial/ethnic group membership but
more meaningful group differences based on gender.

Keywords: interests, Holland’s theory, cultural validity, race and interests

As the U.S. population continues to diversify, career counselors will increas-

ingly see changes in the demographic composition of the U.S. workforce as well
as increasing diversity of the clients seeking services. As an example of this trend,
individuals were able to endorse one or more racial/ethnic affiliation on the 2000
U.S. census, with individuals 18 or younger more likely to identify two or more
racial/ethnic minority affiliations. Data from the 2000 census indicate that
approximately two out of three individuals in the United States identified as
Caucasian (a 9% decrease from the 1990 census data) and provided evidence
that diversity exists within each U.S. state (Brewer & Suchan, 2001).
Furthermore, a committee of the National Research Council (NRC; 1999)
reported that the U.S. workforce has in fact experienced changes in diversity,
including age, sex, race, and education, in nearly all occupational groups.
Therefore, career counselors can expect to see more diversity in the types of
clients pursuing career counseling in the next decade.
It is also important for career counselors to recognize that the occupational
landscape remains unequally distributed among racial/ethnic groups, which may

JOURNAL OF CAREER ASSESSMENT, Vol. 12 No. 4, November 2004 423–439

DOI: 10.1177/1069072704267736
© 2004 Sage Publications

423

Downloaded from jca.sagepub.com at UCSF LIBRARY & CKM on April 25, 2015
424 JOURNAL OF CAREER ASSESSMENT / November 2004

have implications for the context and content of some occupations. For example,
the Bureau of Labor Statistics (BLS; 2002) reported that African Americans and
Hispanics are underrepresented in many professional occupations (e.g., execu-
tive or managerial specialties, engineers, natural scientists, physicians, dentists,
etc.) but overrepresented in other lower paying positions (e.g., dietician, social
worker, guard, postal worker, etc.). This unequal distribution is also evident in
median weekly salaries as African Americans earned 80% and Hispanics earned
68% of the median weekly wage that Caucasians did (BLS, 2002). It seems that
race and ethnicity may have an important impact on occupational choice.
The aforementioned data indicate that the occupational landscape in the U.S.
workforce is unequal, with racial and ethnic minorities being underrepresented
in several occupational groups. It is essential that career counselors be aware of
these trends and begin to identify strategies to promote occupational equity with
all clients.
One question that continues to surface among practitioners is whether inter-
est inventories are appropriate to use with racial and ethnic minority clients. This
is a critical concern as interest inventories are the most widely used assessment
tool in career counseling (Walsh & Betz, 2001). Recent research has focused on
several methods to help determine the utility of assessment inventories such as
the Strong Interest Inventory (SII) with individuals of different racial and ethnic
backgrounds and the application of Holland’s (1997) RIASEC (realistic, inves-
tigative, artistic, social, enterprising, conventional) model, on which many of the
interest inventories, including the SII, were based. The purpose of this article is
to provide a brief review of the literature examining the cross-cultural validity of
the SII and to provide support for the use of interest inventories with racially and
ethnically diverse populations. We will also present additional empirical evi-
dence on racial/ethnic differences of the vocational interests of career clients and
discuss the practical implications of this line of research.

CULTURAL VALIDITY OF
THE STRONG INTEREST INVENTORY

The SII was developed to assess vocational interests of professionals and to pre-
dict individuals’ future occupational satisfaction by assessing their similarity to
those professionals’ interests. An overarching framework of Holland’s (1997) the-
ory, with six General Occupational Themes (i.e., GOTs: realistic, investigative,
artistic, social, enterprising, and conventional) provides interpretive clarity to the
SII. When researchers examine the “cultural validity” of an assessment tool, they
tend to focus on whether the underlying theory adequately explains assessment
results for racial/ethnic minorities, or in this case, whether the results fit
Holland’s (1997) hexagonal model. This also has the added benefit of providing
additional construct validity and generalizability of Holland’s theory.

Downloaded from jca.sagepub.com at UCSF LIBRARY & CKM on April 25, 2015
 Fouad, Mohler / CULTURE VALIDITY OF HOLLAND’S THEORY 425

Empirical studies have focused on several methodologies to examine the cul-

tural validity of the SII. This area of research began by first exploring mean dif-
ferences in interest scores of racial/ethnic minority groups and has now moved on
to more statistically technical examinations including multidimensional scaling
techniques and calculating effect sizes to indicate the strength of an effect. The
majority of this research has suggested support for the use of the SII with diverse
clients and will be briefly discussed here to help provide a context for the present
study.

Examining Mean Differences of Interest Scores

Research that examines the mean differences of interest scores of racial/ethnic

minority groups has found that different groups seem to prefer different types of
vocational interests, and “stereotypical” interests for racial/ethnic groups have
emerged as a result. For example, African Americans scored lower on realistic
and investigative scales than Caucasians but higher on social, enterprising, and
conventional themes compared to Caucasians (Carter & Swanson, 1990). In
addition, Asian Americans have been shown to score higher on realistic,
investigative, and conventional scales when compared to Caucasians (Sue &
Kirk, 1972, 1973). Thus, for many years, career counselors and researchers
concluded that some groups tend to show preferences for certain types of areas
and occupations.
Although some researchers have found differences in interests among
racial/ethnic groups, Davison Aviles and Spokane (1999) for the most part did not
find significant differences on any of the six Holland themes among Hispanic,
African American, and Caucasian middle school students. However, they did
find that Hispanic students tended to express higher interests on the convention-
al theme when compared to Caucasian students. So, although some research has
provided support that racial/ethnic group minorities tend to differ from each
other in their preferences for occupational themes as measured by the SII, other
studies have not replicated these findings.

Using Effect Sizes to Compare Racial/Ethnic

Group Interests and Career Choice

The equivocal findings in studies of mean differences led some researchers to

examine ethnic group differences in interests and also occupational choice by
calculating effect sizes. Calculation of effect sizes provides a common metric
across studies to determine the strength of a particular relationship among vari-
ables and indicates the practical significance of the relationship. This has been
shown to be preferable to merely relying on statistical significance, which is heav-
ily influenced by sample size (Thompson, 2002). Calculation of effect sizes

Downloaded from jca.sagepub.com at UCSF LIBRARY & CKM on April 25, 2015
426 JOURNAL OF CAREER ASSESSMENT / November 2004

enables the comparison of the effects found in different studies. Effect sizes can
generally be measured in two ways. One way is as the standard difference
between two means (denoted by d). The other measurement includes a correla-
tion between an independent variable and individual scores on a dependent vari-
able; the correlation is called the “effect size correlation” (Rosnow & Rosenthal,
1996). In the latter method, effect sizes range from 0 to 1.0 and are often report-
ed as a correlation coefficient (i.e., r or r2). Cohen (1988) suggested using a guide-
line of small (0.1), medium (0.3), and large (0.5) to interpret practical signifi-
cance of effect sizes of the relationships between two variables. Vacha-Haase and
Nilsson (1998) recommended the use of proportion of variance, eta squared (η2),
in multivariate analyses, with .02, .06, and .14 as guides for estimating small,
medium, and large effect sizes.
Overall, it seems that effect size estimates indicate that racial/ethnic minority
group membership has a weak relationship with vocational interests and career
choice. For example, among middle school students, Davison Aviles and
Spokane (1999) found that the overall effect size for ethnicity across the six
Holland themes was very small (r = .06), meaning that although racial/ethnic
groups differed statistically in their vocational interests, this was not a big differ-
ence practically. Leung, Ivey, and Suzuki (1994) reported small effect sizes for
ethnic group differences in career choice among Caucasian and Asian American
students, finding as well that ethnicity and career choice were weakly related.
However, the researchers did find that Asian American students showed a higher
interest in investigative (r = .24) and typically male-dominated careers (r = .21).
Similarly, Park and Harrison (1995) found that Asian Americans have higher
interests in realistic, investigative, and conventional areas (r = .14, .11, and .32,
respectively) and less of an interest in social areas (r = .25) when compared to
Caucasian students. Some research indicates that Asian American students may
have different preferences for occupations than Caucasian students, although
other research indicates no difference in occupational choice.
Fouad, Harmon, and Hansen (1994) also reported few racial/ethnic group dif-
ferences in vocational interests among professionals who met the criteria for the
1994 revision of the SII. Minimal differences (i.e., effect sizes greater than r =
0.2) were found on the investigative scale, with Asian Americans scoring higher
than African American (r = .26), Hispanics (r = .21), and American Indians (r =
.29). Differences were also reported on the artistic scale, with American Indians
scoring lower than Asian Americans (r = .21) and Hispanics (r = .21). However,
it is worth noting that because they used the professionals who formed the crite-
rion sample for the revision of the SII, Fouad et al.’s sample consisted of those
who were successful in their occupation as well as satisfied in that occupation.
Thus, the professional sample used in this study may have a high degree of accul-
turation that may have implications for the findings, leading them to be more
similar to the mainstream Caucasian population in interests. Practitioners may
wish to think about incorporating discussions or assessing for level of accultura-
tion when working with clients from diverse backgrounds.

Downloaded from jca.sagepub.com at UCSF LIBRARY & CKM on April 25, 2015
 Fouad, Mohler / CULTURE VALIDITY OF HOLLAND’S THEORY 427

Structure of Interests for Racial/Ethnic Minorities

Researchers have also focused on the underlying structure of interests meas-

ured by the SII. Fouad, Harmon, and Borgen (1997) used multidimensional scal-
ing to determine the circular ordering of interests (R-I-A-S-E-C) of professionals.
Fouad et al. (1997) found that the circular structure was similar among African
American, Hispanic, Asian American, Native American, and Caucasian profes-
sionals, meaning that the overall pattern of interests is quite similar. Similarly,
Day and Rounds (1998) and Day, Rounds, and Swaney (1998), using the
American College Testing Program’s interest inventory (UNIACT), did not find
significant differences between American racial/ethnic groups in the structure of
interests. So, it seems that the structure of interests follows Holland’s theoretical
model for individuals in most racial/ethnic groups in the United States, and prac-
titioners can interpret SII results based on Holland’s theory for minority clients.
Taking this point a step further, Lattimore and Borgen (1999) found that based
on the six RIASEC themes, the SII was able to predict current occupation across
racial/ethnic groups. This research would suggest that the underlying structure of
vocational interests as well as the ability to predict occupations based on the SII
remains the same for most clients regardless of racial or ethnic background.
Finally, Fouad (2002) examined the structure of interests as well as within- and
between-group differences in interests of students and professionals of five differ-
ent racial/ethnic groups. By examining structure and calculating effect sizes,
Fouad determined that individuals identifying with different racial/ethnic groups
do not differ on vocational interests as measured by the SII. More specifically,
Fouad reported a small effect size for differences among racial/ethnic groups,
with no statistically significant interactions among the groups. These findings
seem to indicate that the SII can be used with diverse populations to help clients
make effective career decisions.
The aforementioned research seems to show very little differences across cul-
tures when using the SII. However, the research does not inform practitioners
whether there are racial/ethnic differences for career clients who may not be as
acculturated as the professional samples used earlier. The extant research has
also focused exclusively on one aspect of Holland’s (1997) theory, that of the
order of the themes on the hexagon. Although this research has been important
to understand whether the hexagon may be used across populations, it is also
important to examine other areas of Holland’s theory. For example, very little
work has focused on whether populations differ in the consistency or differentia-
tion of their profiles. The former concept, consistency, refers to how closely relat-
ed codes are found in individuals’ profiles. Higher consistency is thought to be
related to persistence, achievement, and job satisfaction because similar areas of
interests are more likely to be integrated (Holland, 1997). Differentiation, or the
difference between the highest and lowest themes, is also predicted to be related
to positive work outcomes.

Downloaded from jca.sagepub.com at UCSF LIBRARY & CKM on April 25, 2015
428 JOURNAL OF CAREER ASSESSMENT / November 2004

Some research has examined Holland’s secondary constructs but has shown
inconsistent results. Correlational studies have estimated the relationships
among job satisfaction and consistency and differentiation to be relatively small,
ranging from –.03 to .12 in one study (Carson & Mowsesian, 1993) to –.18 to .04
in an international sample (Leong, Austin, Sekaran, & Komarraju, 1998). Using
multiple regression, Leong et al (1998) found that consistency and differentiation
did not predict job satisfaction among workers in India, which may suggest that
these indices of Holland’s theory do not apply cross-culturally. Swanson and
Gore (2000) and Holland (1997) noted that little attention has been paid to con-
sistency and differentiation because of small homogeneous sample sizes but rec-
ommend future researchers continue to investigate these constructs with newer
operationalizations. Little work has focused on consistency and differentiation
across racial/ethnic groups. It is important to examine if these concepts are simi-
lar across populations to help determine the cultural validity of Holland’s model.
This study will add to the existing literature by providing additional evidence
for the use of the SII with diverse populations, including examinations of
General Occupational Themes and Basic Interest Scales, consistency, differenti-
ation, and structure across five racial/ethnic groups for career clients. The empir-
ical results of this study will help inform practitioners using interest inventories
with racial and ethnic minority clients by expanding information on consistency
and differentiation and the Basic Interest Scales, which can assist with avoca-
tional and leisure counseling.

METHODOLOGY

Participants

Participants were a sample of career clients and students who had completed
the SII in 2001 and who had self-identified as members of the following
racial/ethnic groups: African American/Black, Asian or Pacific Islander, Latino,
Latina/Hispanic, American Indian or Alaskan Native, and Caucasian. Those
individuals who marked more than one racial/ethnic category were not included;
we also excluded those who marked “other.” Data were drawn randomly from the
SII publishing and scoring company’s database until a sample of 750 individuals
was developed. The final sample of 750 of each group was drawn from a larger
total sample. African American participants were drawn from a sample of 10,915
cases, the 750 Asian American participants were drawn from a sample of 5,504
cases, the 750 Hispanics were drawn from a sample of 6,564 cases, the Native
Americans were drawn from a sample of 1,017 cases, and the Caucasians were
drawn randomly from 88,617 cases. All participant data were anonymous, with
no identifying information given other than an identification number. The indi-
viduals completing the SII were informed, “Responses on this form may be used
in the aggregate for research and inventory revalidation and renorming.”

Downloaded from jca.sagepub.com at UCSF LIBRARY & CKM on April 25, 2015
 Fouad, Mohler / CULTURE VALIDITY OF HOLLAND’S THEORY 429

The 750 cases in each group contained groups of both men and women, rang-
ing from 54% female (Caucasian sample) to 61% female (African American sam-
ple). Specifically, women were 431, 437, 423, 450, and 454 of the American
Indian, Asian American, Caucasian, Hispanic, and African American samples,
respectively. The median age of the groups was 20, with mean ages ranging from
23.41 (SD = 10.19) for Native American women and men, 28.0 (SD = 12.11) for
Caucasian women and men, 22.78 (SD = 8.37) for Asian Americans, 25.19 (SD =
10.11) for African American men and women, and 23.03 (SD = 8.63) for
Hispanic/Latino women and men. Participants were asked to indicate the edu-
cational level they had completed; the mean for the entire group was 3.2, where
2 represented high school graduate, 3 represented trade/technical training, and 4
represented some college. Educational levels for specific groups were 2.52
(American Indian), 3.9 (Caucasian), 3.34 (Asian American), 3.25 (African
American), and 2.95 (Hispanic/Latino).
Participants were also asked about their employment status; they were also
able to respond that they were enrolled as a student, engaged as a homemaker,
or retired. Of the 3,750 participants, 63% indicated that they were only enrolled
as students, though percentages ranged from 59% (Caucasians) to 81% (Asian
Americans), with 67% of African Americans, 75% of Hispanics, and 76% of
American Indians indicating they were enrolled as students. A small number of
the groups indicated they were retired or were homemakers, whereas 28%, 25%,
23%, 20%, and 13% of Caucasians, African Americans, Hispanics, American
Indians, and Asian Americans, respectively, indicated they were unemployed and
looking for work. As a whole, the participants averaged working 20 to 25 hours a
week; those who were working were somewhat satisfied with their employment.
Consistent with the relatively young age of the group and their student status, the
majority had worked an average of between 3.5 and 5 years, though the group
also included a few people who had worked as many as 37 years.

Instrumentation

Participants completed the SII (Harmon, Hansen, Borgen, & Hammer, 1994).
The 317 items of the SII are designed to assess an individual’s preferences for a
variety of occupations, schools subjects, and activities. For most items, individu-
als are asked if they like, dislike, or are indifferent to each item. An individual’s
likes and dislikes are then used to compare that individual’s interests to the inter-
ests of professionals in some 100 occupations. The SII includes four types of
scales. The General Occupational Themes and Basic Interest Scales were used
in this study; the former are designed to measure the six Holland (1997) RIASEC
themes and the latter are used to measure how much an individual likes a vari-
ety of clusters of interests (e.g., mathematics, athletics, helping people). Two
additional types of scales include the Personal Styles Scales and Occupational
Scales.

Downloaded from jca.sagepub.com at UCSF LIBRARY & CKM on April 25, 2015
430 JOURNAL OF CAREER ASSESSMENT / November 2004

Psychometric information for the GOTs and BISs are provided in the manu-
al (Harmon et al., 1994). Internal consistency estimates for the six General
Occupational Themes range from .90 for the social GOT to .94 for the artistic
GOT. Internal consistency estimates for the BISs range from .74 (agriculture) to
.94 (mechanical activities). Test-retest reliabilities of the GOT over 3 to 6 months
for a sample of college students ranged from .77 (enterprising) to .91 (artistic);
test-retest reliabilities for this same group on the BISs ranged from .68 (military
activities) to .91 (religious activities). Test-retest reliabilities for employed adults
were higher, ranging in the .80s and .90s. The authors of the manual have pro-
vided substantial evidence of criterion related validity.

RESULTS

Means and standard deviations for all groups on the General Occupational
Themes and Basic Interest Scales are provided in Tables 1 through 3. The analy-
ses focused on four areas to assess cultural validity of Holland’s theory and
racial/ethnic group differences in the SII. The first analysis indicated whether the
groups differed structurally on the General Occupational Themes. Those groups
that were similar in pattern of relationships on the GOT (and thus appeared to
perceive the world of work similarly) were then used in the subsequent analyses:
mean differences on the GOT and BIS across groups, racial/ethnic and gender
differences in differentiation, and finally, racial/ethnic and gender differences in
consistency.

Structural Analyses

The structure of interests for the 10 groups (men and women in each
racial/ethnic group) was examined to determine if the structure underlying the
GOT across groups is similar. The correlation matrices for the 6 RIASEC themes
were examined to determine if they are related as predicted by the theory; that is,
if they are ordered around a circle such that realistic is next to investigative and
conventional, investigative next to artistic, and so on. The theory also suggests
that the correlations between adjacent pairs (e.g., RI, IA, AS) are higher than the
correlations between themes one apart (e.g., RA, IS, AE), which in turn are
greater than opposite themes (RS, IE, AC).
This predicted circular model and the predicted relationships among themes
are tested using the randomization test of hypothesized order (Hubert & Arabie,
1987). This test evaluates whether the relationships between themes for each
groups is random or if it fits the predicted pattern. In effect, the RANDALL pro-
gram (Tracey, 1997) tests the predicted order of correlations against a random
order of correlations. The program evaluates the number of predicted matches
that were met (e.g., that the IA correlation is higher than the IS correlation); the

Downloaded from jca.sagepub.com at UCSF LIBRARY & CKM on April 25, 2015
 Fouad, Mohler / CULTURE VALIDITY OF HOLLAND’S THEORY 431

Table 1
Means and Standard Deviations on General
Occupational Theme by Gender and Racial/Ethnic Group

Female Participants Male Participants

Ethnicity M SD n M SD n
Realistic
Native American 42.44 8.97 431 51.38 9.98 319
Asian American 42.08 7.75 437 49.96 9.24 313
African American 41.07 7.65 454 47.08 9.23 296
Caucasian 42.65 7.91 423 50.86 9.93 327
Hispanic 41.10 7.87 450 49.79 9.60 300
Investigative
Native American 41.44 9.83 431 41.71 10.26 319
Asian American 44.98 10.03 437 48.09 9.85 313
African American 42.46 9.44 454 43.56 10.13 296
Caucasian 42.66 9.14 423 46.02 9.96 327
Hispanic 41.70 9.50 450 44.80 10.40 300
Artistic
Native American 45.96 10.85 431 40.78 10.13 319
Asian American 49.56 9.80 437 45.20 10.12 313
African American 48.11 9.62 454 43.20 9.08 296
Caucasian 50.65 10.27 423 44.69 10.33 327
Hispanic 46.74 10.12 450 44.49 9.57 300
Social
Native American 50.23 11.11 431 43.79 11.45 319
Asian American 50.51 10.47 437 47.12 10.83 313
African American 54.08 10.86 454 48.50 10.61 296
Caucasian 51.62 10.21 423 46.52 10.09 327
Hispanic 50.56 10.28 450 47.47 11.00 300
Enterprising
Native American 47.46 9.91 431 47.07 11.26 319
Asian American 51.46 10.38 437 51.92 10.61 313
African American 53.24 10.66 454 51.67 10.64 296
Caucasian 50.37 10.27 423 50.22 10.82 327
Hispanic 49.73 10.02 450 50.11 10.48 300
Conventional
Native American 49.97 11.06 431 45.19 9.32 319
Asian American 50.24 9.77 437 50.98 9.56 313
African American 53.02 11.01 454 49.53 9.48 296
Caucasian 48.82 10.56 423 46.98 8.82 327
Hispanic 50.77 10.17 450 49.33 9.48 300

Downloaded from jca.sagepub.com at UCSF LIBRARY & CKM on April 25, 2015
432 JOURNAL OF CAREER ASSESSMENT / November 2004

Table 2
Female Means and Standard Deviations for
Basic Interest Scales for Racial/Ethnic Groups

Native Asian African

American American American Caucasian Hispanic
M SD M SD M SD M SD M SD
Agriculture 45.05 8.81 41.77 7.46 40.38 6.92 45.49 9.14 41.69 7.49
Applied arts 44.40 10.80 46.72 10.07 43.86 9.92 47.55 10.39 43.57 10.26
Art 48.17 10.64 51.62 9.41 47.74 9.47 51.79 10.10 48.64 10.19
Athletics 47.54 9.34 46.24 8.43 49.34 8.71 47.18 9.63 47.49 8.97
Computer 49.11 11.82 48.80 10.83 51.99 11.47 46.67 10.92 49.43 11.51
Culinary arts 50.35 10.22 52.45 8.71 52.50 9.12 52.77 9.78 51.41 8.94
Data management 45.49 9.47 46.76 8.86 47.93 9.59 45.23 9.85 45.78 9.26
Law/politics 43.20 9.79 45.06 9.75 47.17 9.89 45.07 9.69 44.52 9.71
Mathematics 43.62 8.96 46.45 8.94 45.20 8.92 43.58 9.00 43.87 8.53
Mechanical
activities 43.47 8.45 44.01 7.78 43.66 7.83 43.02 7.52 42.92 7.64
Medical science 44.77 9.99 47.64 10.24 46.26 10.34 46.09 9.81 45.43 10.15
Medical service 51.62 11.66 51.21 10.90 52.95 11.80 51.10 11.07 51.56 11.99
Merchandising 47.37 9.50 51.39 9.72 52.64 10.01 50.30 10.19 49.74 9.79
Military activities 48.38 9.27 47.78 8.52 49.13 9.60 46.74 7.60 48.51 9.78
Music/drama 47.18 11.02 51.12 9.71 50.48 9.67 51.71 10.26 48.66 10.41
Nature 43.26 10.65 43.12 10.36 37.99 10.21 45.59 10.58 40.22 9.87
Office practices 53.31 11.01 52.19 10.04 56.20 11.15 51.68 10.68 54.49 10.57
Organizational
management 44.98 9.91 48.86 9.62 50.79 9.44 47.33 9.82 47.54 9.62
Public speaking 44.01 9.46 46.45 9.62 49.66 9.53 46.17 9.50 45.07 9.20
Religious activities 48.15 9.83 49.25 10.04 54.67 10.15 49.07 10.10 48.16 9.69
Sales 50.08 9.36 52.44 10.53 53.60 10.32 50.14 10.01 50.97 9.87
Science 42.86 9.33 44.80 9.28 43.36 9.04 43.52 8.85 42.56 9.15
Social service 51.11 11.59 52.82 9.84 55.12 10.36 54.41 9.96 52.53 10.32
Teaching 46.48 11.17 46.35 10.77 47.57 11.41 48.89 10.74 46.16 11.09
Writing 43.94 10.51 46.04 10.24 46.98 10.31 48.28 10.85 43.30 10.13

test also provides a correspondence index (CI). The CI is a summary of the pre-
dicted matches met from which those matches not predicted are subtracted and
the entire sum is divided by 72 (the total possible number of predictions from six
themes). Possible CI scores range from –1 to 1; 0 indicates random pattern of cor-
relations (and poor fit of Holland’s model). Higher scores (e.g., greater than .50)
indicate a greater fit to Holland’s model. Probability value less than .05 was used
in this study to indicate that CI and thus predicted matches were significantly
greater than random.

Downloaded from jca.sagepub.com at UCSF LIBRARY & CKM on April 25, 2015
 Fouad, Mohler / CULTURE VALIDITY OF HOLLAND’S THEORY 433

Table 3
Male Means and Standard Deviations for
Basic Interest Scales for Racial/Ethnic Groups

Native Asian African

American American American Caucasian Hispanic
M SD M SD M SD M SD M SD
Agriculture 48.00 8.66 43.53 7.49 42.50 6.84 48.07 9.37 44.58 7.67
Applied arts 43.39 10.57 46.00 10.29 43.84 9.89 46.14 10.26 45.66 9.88
Art 43.44 10.38 46.52 10.09 44.09 9.39 45.44 10.32 45.80 9.95
Athletics 53.32 9.94 53.60 9.39 56.74 9.00 55.50 10.03 55.20 8.98
Computer activities 48.89 11.09 54.34 9.90 53.02 10.55 49.17 10.06 51.82 10.78
Culinary arts 43.68 11.02 47.18 10.01 46.71 9.68 47.85 9.64 47.07 9.73
Data management 43.52 8.36 50.44 9.72 48.39 8.83 47.43 9.69 47.55 8.97
Law/politics 43.26 10.61 48.58 10.43 47.57 10.33 49.46 10.92 47.44 10.27
Mathematics 43.97 8.60 50.97 9.20 46.81 9.04 47.92 9.69 46.99 8.96
Mechanical
activities 52.48 10.11 52.43 9.49 50.31 9.60 51.50 9.95 51.91 9.83
Medical science 43.26 10.06 48.14 9.89 45.20 10.16 46.90 9.67 46.19 10.20
Medical service 46.44 11.03 49.41 10.23 48.70 10.35 47.41 9.18 49.34 11.19
Merchandising 45.95 10.58 51.27 10.18 50.74 9.74 49.28 10.42 49.48 9.84
Military activities 54.02 12.33 52.87 10.87 51.66 11.29 53.22 11.79 54.97 12.61
Music/drama 42.69 9.76 47.24 10.11 45.80 8.96 46.87 9.67 46.71 9.68
Nature 42.52 10.46 41.69 10.01 38.10 10.33 45.06 10.44 41.95 9.94
Office practices 46.74 8.63 50.39 8.94 50.43 9.36 46.67 7.73 50.07 9.13
Organizational
management 44.46 11.00 50.51 10.07 49.95 10.15 48.18 10.27 48.37 10.08
Public speaking 43.66 10.55 47.79 10.52 48.91 9.71 48.90 10.30 47.66 9.95
Religious activities 46.46 9.83 49.08 10.31 51.74 10.18 48.43 10.21 47.01 9.93
Sales 52.14 10.15 53.69 10.85 54.58 10.53 52.91 11.09 52.88 10.53
Science 44.95 9.85 49.46 9.62 45.71 9.59 47.35 9.97 47.28 9.93
Social service 43.31 11.33 47.60 10.23 48.08 10.45 47.39 9.75 47.97 10.69
Teaching 42.04 11.67 45.43 10.88 45.29 10.95 46.18 11.33 45.89 11.97
Writing 39.46 9.88 43.34 10.02 41.94 9.51 44.21 10.57 42.71 9.74

The number of predicted matches met, CIs, and associated probability levels
for each of the groups on the randomization test are presented in Table 4. Results
indicate that CIs ranged from .57 (p = .02) for Native American men to .94 (p =
.02) for the Asian American women. The mean CI was .77. Thus, as indicated by
Table 4, all CIs are high, with p values equal to or less than .05. These results
indicate that the pattern of relationships among the RIASEC themes for all of the
groups were in the predicted circular order.

Downloaded from jca.sagepub.com at UCSF LIBRARY & CKM on April 25, 2015
434 JOURNAL OF CAREER ASSESSMENT / November 2004

Table 4
Correspondence Indices, Number of Predicted
Matches, and Probability Levels for Randomized Test

Predicted Matches Met C-Index Probability Value

Native American
Women 62/72 .72 .03
Men 56/72 .57 .02
African American
Women 62/72 .72 .02
Men 60/72 .69 .02
Asian American
Women 70/72 .94 .02
Men 65/72 .81 .02
Caucasian
Women 64/72 .79 .03
Men 66/72 .83 .02
Hispanic
Women 61/72 .71 .02
Men 67/72 .89 .02

Mean Differences on GOT

Differences across groups were examined for the GOT using a multivariate
analysis of variance (MANOVA). In these analyses, the six orthogonal GOT
scales served as the dependent variables and gender and race/ethnicity as the
independent variables.
Results of the MANOVA indicated that all main and interaction effects were
significant; however, this may have been due to the very large sample sizes (all
were greater than 300). Thus, as recommended by Thompson (2002) and Vacha-
Haase and Nilsson (1998), we calculated effect sizes, using guidelines discussed
earlier. MANOVA results indicated a large effect size for sex differences among
the GOT themes, F(6, 3,735) = 314.15, p < .001, η2 = .34. A small effect size was
found for differences among ethnic groups, F(6, 37,35) = 20.01, p < .001, η2 =
.03). Effect sizes for interaction effects were all below .01.
The largest contributor to the large effect size for differences between men
and women was the realistic GOT (F = 754.62, p < .001, η2 = .17), with a mean
of 49.81 for men and 41.87 for women. Smaller effect sizes were found for artis-
tic (F = 185.77, p < .001, η2 = .05), with women scoring higher (48.19) than men
(43.67), and social GOT (F = 177.79, p < .001, η2 = .05), with women again scor-
ing higher (51.42) than men (46.64).
Follow-up analyses of the ethnicity main effect indicated that the contributors
to the small effect size were the investigative GOT (F = 25.83, p < .001, η2 = .03)

Downloaded from jca.sagepub.com at UCSF LIBRARY & CKM on April 25, 2015
 Fouad, Mohler / CULTURE VALIDITY OF HOLLAND’S THEORY 435

and enterprising GOT (F = 26.43, p < .001, η2 = .03). Asian Americans scored
higher (46.3) than the Native American (41.56) group on the investigative GOT,
and African Americans (52.63) scored higher than Native Americans (47.3) on
the enterprising GOT.

Mean Differences on BIS

Because of item overlap on some of the BISs, a univariate analysis of variance

(ANOVA) was used to assess gender by race differences on each BIS.
Examination of the effect sizes indicates that there were few meaningful eth-
nic group differences in the BIS. A medium effect size difference was found on
the agriculture BIS (F = 65.5, p < .001, η2 = .07); small-medium effect size dif-
ferences were found on the nature BIS (F = 48.75, p < .001, η2 = .05) and reli-
gious activities BIS (F = 40.28, p < .001, η2 = .04). On the agriculture BIS,
African Americans scored the lowest on this scale (41.44), whereas Caucasians
and Native Americans scored 46.78 and 46.53, respectively. On the nature BIS,
African Americans were also the lowest (38.04), whereas Caucasians were high-
est (45.3). On the religious activities BIS, the African American mean was high-
est of the groups (53.21), whereas the Hispanic and Native American groups were
lowest (47.58 and 47.3, respectively).
Similar to the pattern found on the GOT, there were more meaningful gen-
der differences on the BIS than ethnic group differences. Large effect size differ-
ences were found between men and women on the following two scales: athlet-
ics (F = 572.04, p < .001, η2 = .13) and mechanical activities (F = 827.45, p <
.001, η2 = .18), with men scoring higher (54.87 and 51.73, respectively) than
women (47.56 and 43.42, respectively). Medium or small-medium effect size
gender differences were found on the following five BISs: military activities (F =
237.23, p < .001, η2 = .06, men’s X = 53.35, women’s X = 48.11), art (F = 186.89,
p < .001, η2 = .05, men’s X = 45.06, women’s X = 49.59), culinary arts (F =
284.47, p < .001, η2 = .07, men’s X = 46.5, women’s X = 51.9), social service
(F = 332.88, p < .001, η2 = .08, men’s X = 46.87, women’s X = 53.2), and office
services (F = 204.4, p < .001, η2 = .05, men’s X = 48.86, women’s X = 53.57).

Differentiation

The third type of analysis calculated group differences on the construct of dif-
ferentiation, or the difference between an individual’s highest and lowest GOTs.
Differentiation was calculated using the measure recommended by Sackett and
Hansen (1995), which takes into account the entire spread of scores. They used
a standard deviation of all six scores instead of the difference between the two
highest codes suggested by Monahan (1987); Sackett and Hansen’s approach
incorporates the difference across the entire hexagon rather than for just two

Downloaded from jca.sagepub.com at UCSF LIBRARY & CKM on April 25, 2015
436 JOURNAL OF CAREER ASSESSMENT / November 2004

Table 5
Means and Standard Deviations for Differentiation

Female Participants Male Participants

Ethnicity M SD n M SD n
Native American 8.28 3.16 431 7.69 2.83 319
Asian American 8.49 2.70 437 7.88 2.81 313
African American 9.18 2.91 454 7.46 2.76 296
Caucasian 9.01 2.92 423 8.35 2.64 327
Hispanic 8.41 3.02 450 7.66 2.77 300

themes. Mean differentiation scores for the 10 groups in this study are presented
in Table 5.
Differentiation differences for gender and race/ethnicity were calculated using
analysis of variance. Results indicated a small effect size for gender (F = 82.82, p <
.001, η2 = .02, men’s X = 7.8, women’s X = 8.68). Ethnic group differences were
negligible.

Consistency

The final analysis conducted was to assess consistency of themes, or how close-
ly related an individual’s high themes are. Consistency was calculated using
Strahan’s (1987) formula, which assigns a weight of 6, 2, 1 to the three codes
(indicating distance between code 1 and 2 is more important than between code
2 and 3) and then a 2 to adjacent codes, a 1 to codes that are one theme apart,
and a 0 to codes that are opposite each other on the hexagon. Thus, a Holland
code of RSA would be C = 6(0) + 2(1) + 1(1), or 3, whereas a code of RIC would
be C = 6(2) + 2(2) + 1(1), or 17. Strahan then recommends a rank transforma-
tion of Y to eliminate gaps in the intervals caused by the weighting, thus the con-
sistency index of C ranges from 1 to 10 and would be 4 for the first example of
RSA and 10 for the second example of RIC. Mean consistency scores for the 10
groups are presented in Table 6. There were no meaningful gender or racial/eth-
nic group differences in consistency; effect sizes were less than .003.

DISCUSSION

The primary goal of this study was to provide additional evidence for the use
of the SII with diverse populations, including examinations of the structure of
and differences in means for the GOTs and BISs as well as assessing differences
in consistency and differentiation across five racial/ethnic groups for career

Downloaded from jca.sagepub.com at UCSF LIBRARY & CKM on April 25, 2015
 Fouad, Mohler / CULTURE VALIDITY OF HOLLAND’S THEORY 437

Table 6
Means and Standard Deviations for Consistency

Female Participants Male Participants

Ethnicity M SD n M SD n
Native American 6.70 2.59 431 6.71 2.83 319
Asian American 7.07 2.59 437 6.80 2.71 313
African American 7.32 2.45 454 6.89 2.87 296
Caucasian 6.90 2.56 423 6.57 2.80 327
Hispanic 6.63 2.57 450 6.76 2.79 300

clients. Overall, the pattern of results suggests minimal differences based on

racial/ethnic group membership but more meaningful group differences based
on gender, though within-group differences seemed to be greater than between-
group differences for all groups.
By calculating correspondence indices, the first analysis indicated that the pat-
tern of relationships among the RIASEC themes were in the predicted circular
order across groups, confirming results of previous research that racial/ethnic
groups do not differ structurally on the GOTs (e.g., Day & Rounds, 1998; Fouad,
2002; Fouad et al. 1997) and suggesting that Holland’s model fits well for most
racial/ethnic groups in the United States. This finding can be interpreted in that
U.S. racial/ethnic minority groups seem to perceive the world of work in similar
manner as do Caucasians. However, there appear to be more meaningful gender
differences than ethnic differences on the GOTs and BISs.
Differences across groups were examined and indicated a large effect for gen-
der differences on some of the GOTs in stereotypical ways. Specifically, men
were more likely to endorse the realistic GOT than women. There were also
meaningful gender differences in BISs, as demonstrated by large effect size
between men and women on athletics and mechanical activities, with men scor-
ing higher than women on both. This finding replicates previous work suggest-
ing that men and women seem to prefer different occupational interests.
A small effect size was reported for ethnicity, meaning that ethnic group mem-
bership does not seem to predict a particular pattern on the GOTs. Similarly,
small to medium effect sizes indicate few meaningful ethnic group differences in
the BISs. This suggests that Holland’s theory also holds up well across the types
of occupations and interests preferred by members of racial/ethnic minority
groups.
Group differences in differentiation and consistency were also examined and
found little meaningful differences based on ethnicity. This can be interpreted as
meaning that Holland’s model seems to hold up well when comparing ethnic
groups to Caucasian samples, and similar predictions can be made about results
that are consistent or differentiated. A small effect size for gender was reported on

Downloaded from jca.sagepub.com at UCSF LIBRARY & CKM on April 25, 2015
438 JOURNAL OF CAREER ASSESSMENT / November 2004

differentiation differences but not for consistency, meaning that on average,

women tended to have more highly differentiated profiles.
The results of this study have important implications for the use of the SII and
more specifically, Holland’s theory with racial/ethnic minority groups in the
United States. First, the study was able to provide documentation that Holland’s
secondary constructs of consistency and differentiation may be used when inter-
preting SII data for members of racial/ethnic minority groups in the United
States. Second, the pattern of results suggests meaningful differences based on
gender but negligible differences based on ethnicity. Although there appear to be
some interpretable differences within specific racial/ethnic groups, they do not
appear to be pervasive patterns that would contradict the application of Holland’s
theory to minority group members. Rather than focusing on interpretation of SII
results based on ethnic group membership, it is perhaps more efficacious to
address the role of interests in decision making for racial/ethnic minority clients.
It is important for practitioners to be cognizant of commonly occurring prefer-
ences and interests based on ethnic group membership as well as levels of accul-
turation and perceptions of opportunity relating to the world of work.

REFERENCES

Brewer, C. A., & Suchan, T. A. (2001). Mapping Census 2000: The geography of U.S. diversity.
Washington, DC: Government Printing Office.
Bureau of Labor Statistics. (2002). 2002 occupational employment statistics. Retrieved June 2002
from http: //www.bls.gov
Carson, A. D., & Mowsesian, R. (1993). Moderators of the prediction of job satisfaction from con-
gruence: A test of Holland’s theory. Journal of Career Assessment, 1, 130-144.
Carter, R. T., & Swanson, J. L. (1990). The validity of the Strong Interest Inventory with Black
Americans: A review of the literature. Journal of Vocational Behavior, 36, 195-209.
Cohen, J. (1988). Statistical power analyses for the behavioral sciences (2nd ed.). Hillsdale, NJ:
Lawrence Erlbaum.
Davison Aviles, R. M., & Spokane, A. R. (1999). The vocational interests of Hispanic, African
American, and Caucasian middle school students. Measurement and Evaluation in Counseling
and Development, 32, 138-148.
Day, S. X., & Rounds, J. (1998). Universality of vocational interest structure among racial and eth-
nic minorities. American Psychologist, 53, 728-736.
Day, S. X., Rounds, J., & Swaney, K. (1998). The structure of vocational interests for diverse racial-
ethnic groups. Psychological Science, 9, 40-44.
Fouad, N. A. (2002). Cross-cultural differences in vocational interests: Between-group differences
on the Strong Interest Inventory. Journal of Counseling Psychology, 49, 282-289.
Fouad, N. A., Harmon, L. W., & Borgen, F. H. (1997). Structure of interests in employed male and
female members of U.S. racial-ethnic minority and nonminority groups. Journal of Counseling
Psychology, 44, 339-345.
Fouad, N. A., Harmon, L. W., & Hansen, J. C. (1994). Cross-cultural use of the Strong. In L. W.
Harmon, J. C. Hanson, F. H. Borgen, & A. C. Hammer (Eds.), Strong Interest Inventory:
Applications and technical guide (pp. 255-280), Palo Alto, CA: Consulting Psychologists Press.
Harmon, L. W., Hansen, J. C., Borgen, F. H., & Hammer, A. L. (Eds.). (1994). Strong Interest
Inventory: Applications and technical guide. Palo Alto, CA: Consulting Psychologists Press.

Downloaded from jca.sagepub.com at UCSF LIBRARY & CKM on April 25, 2015
 Fouad, Mohler / CULTURE VALIDITY OF HOLLAND’S THEORY 439

Holland, J. L. (1997). Making vocational choices: A theory of vocational personalities and work envi-
ronments (3rd ed.). Odessa, FL: Psychological Assessment Resources.
Hubert, L., & Arabie, P. (1987). Evaluating order hypotheses within proximity matrices.
Psychological Bulletin, 102, 172-178.
Lattimore, R. R., & Borgen, F. H. (1999). Validity of the 1994 Strong Interest Inventory with racial
and ethnic minority groups in the United States. Journal of Counseling Psychology, 46, 185-195.
Leong, F. T. L., Austin, J. T., Sekaran, U., & Komarraju, M. (1998). An evaluation of the cross-
cultural validity of Holland’s theory: Career choices by workers in India. Journal of Vocational
Behavior, 52, 441-455.
Leung, S. A., Ivey, D., & Suzuki, L. (1994). Factors affecting the career aspirations of Asian
Americans. Journal of Counseling and Development, 72, 404-410.
Monahan, C. J. (1987). Construct validation of a modified differentiation index. Journal of
Vocational Behavior, 30, 217-226.
National Research Council. (1999). Changing nature of work: Implications for occupational analy-
sis. Washington, DC: National Academy of Science.
Park, S. E., & Harrison, A. A. (1995). Career-related interests and values, perceived control, and
acculturation of Asian-American and Caucasian-American college students. Journal of Applied
Social Psychology, 25, 1184-1203.
Rosnow, R. L., & Rosenthal, R. (1996). Computing contrasts, effect sizes, and counternulls on
other people’s published data: General procedures for research consumers. Psychological
Methods, 1, 331-340.
Sackett, S. A., & Hansen, J. C. (1995). Vocational outcomes of college freshman with flat profiles
on the Strong Interest Inventory. Measurement and Evaluation in Counseling and Development,
28, 9-24.
Strahan, R. F. (1987). Measures of consistency for Holland-type codes. Journal of Vocational
Behavior, 31, 37-44.
Sue, D. W., & Kirk, B. A. (1972). Differential characteristics of Japanese-American and Chinese-
American college students. Journal of Counseling Psychology, 20, 142-148.
Sue, D. W., & Kirk, B. A. (1973). Psychological characteristics of Chinese-American students.
Journal of Counseling Psychology, 9, 471-478.
Swanson, J. L., & Gore, P. A., Jr. (2000). Advances in vocational psychology theory and research.
In S. D. Brown & R. W. Lent (Eds.), Handbook of counseling psychology (3rd ed., pp. 233-269).
New York: John Wiley.
Thompson, B. (2002). “Statistical,” “practical,” and “clinical”: How many kinds of significance do
counselors need to consider? Journal of Counseling & Development, 80, 64-71.
Tracey, T. G. (1997). RANDALL: A Microsoft Fortran program for a randomization test of hypoth-
esized order relations. Educational and Psychological Measurement, 57, 164-168.
Vacha-Haase, T., & Nilsson, J. E. (1998). Statistical significance reporting: Current trends and uses
in MECD. Measurement and Evaluation in Counseling and Development, 31, 46-57.
Walsh, W. B., & Betz, N. E. (2001). Tests and assessment (4th ed.). Englewood Cliffs, NJ: Prentice
Hall.

Downloaded from jca.sagepub.com at UCSF LIBRARY & CKM on April 25, 2015