Validity of Holland’s theory in adolescence: evidence from a Slovak sample

Abstract

The underlying calculus assumption of Holland’s theory was tested in two samples of primary school students (N1 = 400 and N2 = 250) with an average age of 13.86 and 14.14 years, respectively. Both exploratory and confirmatory multidimensional scaling and Hubert and Arabie’s randomization test of hypothesized order relations were calculated. The circular spatial arrangement of RIASEC types was not confirmed in either of the two samples. The study discusses possible causes of the inappropriateness of using Holland’s model in an adolescent sample.

Full text

Vol.:(0123456789)
International Journal for Educational and Vocational Guidance (2020) 20:543–565
https://doi.org/10.1007/s10775-019-09416-0
1 3
Validity ofHolland’s theory inadolescence: evidence
fromaSlovak sample
MarcelMartončik1 · MonikaKačmárová1· EvaHruščová1·
IvanaMagáčováŽilková1· MichaelaKravcová1
Received: 26 September 2018 / Accepted: 29 November 2019 / Published online: 17 December 2019
© Springer Nature B.V. 2019
Abstract
The underlying calculus assumption of Holland’s theory was tested in two samples
of primary school students (N1 = 400 and N2 = 250) with an average age of 13.86 and
14.14years, respectively. Both exploratory and confirmatory multidimensional scal-
ing and Hubert and Arabie’s randomization test of hypothesized order relations were
calculated. The circular spatial arrangement of RIASEC types was not confirmed in
either of the two samples. The study discusses possible causes of the inappropriate-
ness of using Holland’s model in an adolescent sample.
Keywords Holland’s theory· Adolescence· RIASEC· Vocational interest· Self-
Directed Search
Résumé
Validité de la théorie de Holland à l’adolescence : Preuves provenant d’un
échantillon slovaque L’hypothèse sous-jacente de la théorie du choix vocationnel
de Holland a été testée sur deux échantillons d’élèves de l’école primaire (N1 = 400
and N2 = 250)dont l’ âge moyen est de 13.86 et 14.14 ans respectivement. Des analy-
ses exploratoires et confirmatoires multidimensionnelles, ainsi que le test de ran-
domisation de Hubert et Arabie ont été menées. La disposition spatiale circulaire du
modèle RIASEC n’a été prouvée dans aucun des deux échantillons. L’étude examine
les causes possibles de l’inadéquation de l’utilisation du modèle de Holland dans un
échantillon d’adolescents.
Zusammenfassung
Validität der Holland‘schen Theorie für die Jugend: Belege aus einer slowakis-
chen Stichprobe Die rechnerischen Annahmen, die der Holland‘schen Theorie zu-
grunde liegen, wurden in zwei Stichproben von Schülern (N1 = 400 und N2 = 250) mit
* Marcel Martončik
martoncik@protonmail.ch
1 Faculty ofArts, Institute ofPsychology, Presov University inPresov, Ul. 17. novembra 1,
08001Prešov, Slovakia

544
International Journal for Educational and Vocational Guidance (2020) 20:543–565
1 3
einem Durchschnittsalter von 13,86 bzw. 14,14 Jahren getestet. Sowohl explorative
als auch konfirmatorische multidimensionale Skalierung sowie Randomisierungst-
ests von hypothetisierten Ordnungsbeziehungen nach Hubert und Arabie wurden be-
rechnet. Die kreisförmige räumliche Anordnung der RIASEC-Typen wurde in keiner
der beiden Proben bestätigt. Die Studie diskutiert mögliche Ursachen für die Un-
angemessenheit der Verwendung des Holland‘schen Modells für Jugendliche.
Resumen
Validez de la teoria de Holand en la adolescencia: Evidencias a partir de una
muestra Eslovaca La asunción del calculo de la teoria de Holland se comprobó en
dos muestras de alumnos de educación primaria (N1 = 400 and N2 = 250) con una
edad promedio de 13.86 y 14,14 años, respectivamente. Se calcularon tanto el es-
calamiento multidimensional exploratorio y confirmatorio así como los tests de alea-
torización de las relaciones hipotéticas de orden de Huvert y Arabie. La corrección
circular espacial de los tipos RIASEC no se confirmó en ninguna de las dos muestras.
El estudio discute las posibles causas de inapropiación en el uso del modelo de Hol-
land en una muestra de adolescentes.
Introduction
Holland’s theory
Holland’s hexagonal model, also known by the acronym RIASEC, is probably the
best-known structural model of vocational interests (Holland, 1973, 1997) and is
a predominant model for interest questionnaires (Gottfredson & Holland, 1996,
in Holland & Messer, 2013; Savickas & Gottfredson, 1999). RIASEC question-
naires are also considered a useful way to assess personality differences, as voca-
tional interests are an important aspect of personality. The six letters in the acronym
RIASEC represent six personality and environment types, particularly: R stands for
the Realistic type, I for the Investigative type, A for the Artistic type, S for the Social
type, E for the Enterprising type and C for the Conventional type. The personal-
ity type as a personality disposition influences thinking, perception and action in a
specific way. Each of the six personality types possesses a different set of interests,
abilities, preferred activities and occupations, values, coping mechanisms, and self-
concept. For instance, “people who resemble the Social type are more likely to seek
out social occupations such as teaching, social work, or the ministry. They would be
expected to see themselves as friendly and social and to have more social competen-
cies (e.g., helping others with personal problems) than realistic competencies (e.g.,
using tools or understanding machines)” (Holland, 1997, p. 2).
Holland’s (1997) model is also often termed as the hexagonal model, as the per-
sonality types are spatially arranged into a hexagon, precisely in a way that they
do express interrelationships between the different RIASEC types. The types are
inversely proportional to the tightness of the theoretical relationships between
them. The closer the two types are located to each other, the greater is their mutual

545
1 3
International Journal for Educational and Vocational Guidance (2020) 20:543–565
resemblance. As noted by Tracey and Rounds (1993), the hexagonal model can be
identified with a circular model (circular ordering) as individual personality types
are arranged equally within both models. Two different structural assumptions
underlie Holland’s RIASEC model. The less restrictive calculus assumption repre-
sents the circular order of the RIASEC types. In the more restrictive circumplex
model, additional constraints are added to the calculus assumption: “the interpoint
distances are equal for types within adjacent categories, alternate categories, and
opposite categories” (Tracey & Rounds, p. 232).
The main idea behind Holland’s theory (its typology of personalities and envi-
ronments), but also its practical significance and benefit lies in the possibility of
matching individuals with the environment that is complementary to their abilities
and needs. It enables both identification and prediction of career or study choice,
satisfaction, stability, success, and competency, and social behavior (Holland, 1973,
1997). The possibility of matching individuals with an environment is used in career
intervention, in which adolescent students are frequent clients, as in this develop-
mental period a main concern is future education and occupation (Vondracek & Por-
feli, 2003).
Other well‑known models forthestructure ofvocational interests
Among the many alternative or complementary models to Holland’s RIASEC
model, the most widely-used are Gati’s (1979) hierarchical model, Prediger’s (1982)
dimensional model, the spherical model of Tracey and Rounds (1996) and the recent
SETPOINT model proposed by Su, Tay, Liao, Zhang, and Rounds (2019). Each of
these will be briefly described.
A different arrangement of RIASEC types was proposed by Gati (1979). Accord-
ing to him, “occupations can be characterized as collections of attributes or fea-
tures”, such as working indoors or working outdoors (p. 92). Therefore, occupa-
tions and occupational interests are divided into several subfields in a hierarchical
tree structure, based on their similarities in diverse attributes. RIASEC types are
arranged in a hierarchical model into three separate clusters merging types RI, AS
and EC together. Gati’s hierarchical model could be also viewed as a non-competing
simplified version of the Holland`s model, in which the relations between the pairs
RI, AS and EC are weaker than between the pairs IA, ES and CR (Hubert & Arabie,
1987). Gati’s and Holland’s model share together 75% of relational predictions (Ein-
arsdóttir, Rounds, Ægisdóttir, & Gerstein, 2002). Gati (1991) also proposed an alter-
native arrangement of clusters, in which “Artistic should also constitute a separate
cluster in Holland’s classification, and Social should join the cluster of Enterprising
and Conventional” (p. 316).
Prediger (1982) hypothesized the existence of two foundational dimensions,
which represent the relationships among the RIASEC types. He named two bipolar
dimensions as Data-Ideas tasks and People-Things tasks. Holland’s R type is repre-
sented by the Things tasks, S type by the People tasks, E and C types by the Data
and I and A types by the Idea tasks. Any individual interest or occupation can be
characterized by the presence of a different degree of the four tasks, where “one or

546
International Journal for Educational and Vocational Guidance (2020) 20:543–565
1 3
two of the tasks typically capture the “essence” of an occupation” (Prediger, 1982,
p. 261). For example, the work of a car mechanic is best described as a Things tasks.
The spherical model enriches the two-dimensional space of Holland’s circular
model by a third dimension, prestige, often also called as socioeconomic status,
level of training, difficulty or responsibility (Tracey & Rounds, 1996). In the spheri-
cal model, “the relations of RIASEC interest scales to each other vary as a function
of prestige” (Tracey & Rounds, 1996, p. 6). The highest differentiation among the
RIASEC types is supposed to be at a moderate level of prestige. One of the impor-
tant implications of the spherical model is the arbitrariness of the number of types
since the circular nature of vocational interests could be described by any num-
ber of interest types. Instead of the six RIASEC types, authors chose eight types,
which could better describe and explain the interconnection of the Prediger’s (1982)
dimensional and circular models.
The newest addition to the structural models is the eight-dimensional SETPOINT
interest model proposed by Su, Tay, Liao, Zhang, and Rounds (2019). It represents
an effort to synthesize all the previous findings within the area of interest research
and current changes in the labor market. In this model, interests are hierarchically
structured into eight dimensions: Health Science, Creative Expression, Technology,
People, Organization, Influence, Nature, and Things.
Validity ofHolland’s theory inadolescence
Adolescence is the developmental period from 12 to 20years of age (Shaffer &
Kipp, 2010). During this period, school, work, and leisure interests are closely inter-
related and play an important role in vocational development and identity formation
(Hofer, 2010; Vondracek & Skorikov, 1997). Adolescents acquire experiences and
knowledge about the world of work from hobbies, leisure interests, and school sub-
jects, and also from direct experience with part-time work, family and community
(Vondracek & Porfeli, 2003). This means that the structure of their knowledge of the
world of work is still evolving, as are their interests and goals.
The vast majority of studies, presented also in the specialized database on
Holland`s theory created by Foutch, McHugh, Bertoch, and Reardon (2013), have
used Hubert and Arabie’s (1987) randomization test of hypothesized order relations
(RTOR) for the verification of the circular arrangement of RIASEC types/quasi-
circumplex model. The quasi-circumplex model is less restrictive than circulant, in
which distances between adjacent types must be equal (Guttman, 1954). Two meta-
analyses of studies with samples of adolescents have previously been conducted on
this topic.
A meta-analysis conducted by Tracey and Rounds (1993) on 104 correlational
matrices confirmed a circular structure for those in 8–12th grades (Correspond-
ence index, CI = .68, SD = .16). Using three different structural meta-analytic
techniques, the authors confirmed the superiority of Holland’s model over Gati’s
even though Gati (1979, 1982, 1991) has provided some evidence for the bet-
ter fit of his hierarchical model. A problematic aspect of this meta-analysis is
that 14-year-olds were grouped together with 18-year-olds. Furthermore, no

547
1 3
International Journal for Educational and Vocational Guidance (2020) 20:543–565
verification of the structural hypothesis for subgroups of a shorter age period (e.g.
8–10th grade, 10–12th grade) were provided. In their later meta-analysis of 96
RIASEC matrices from 19 countries (with no age ranges indicated), they found
no universal cross-cultural support for Holland’s circular order model (Rounds &
Tracey, 1996). In addition, Holland’s RIASEC structure differed significantly in
15 out of 18 countries, excluding the U.S. ethnic sample. Gati’s partition and the
alternative partition model instead achieved the best fit.
Another meta-analysis conducted by Long and Tracey (2006) in a Chinese
population (29 correlational matrices) yielded different results. The authors found
inadequate model fit for samples of middle school students, leading them to ques-
tion the validity of Holland’s model and to confirm the superiority of Gati’s. The
results of studies using the RTOR method which have been published since the
Tracey and Rounds (1993) meta-analysis are shown in Table1, sorted by age of
the research samples. It can be seen there that the majority of studies disconfirm
the validity of Holland’s model (as can be seen from their low CI values or non-
significant results), particularly those including younger adolescents.
Another commonly used method for the verification of circular arrangement of
RIASEC types in addition to RTOR is multidimensional scaling (MDS). MDS is
used to spatially describe the relationships between RIASEC types in two-dimen-
sional space. The results of studies conducted in adolescent samples so far are
ambiguous, as was the case with the RTOR results. Only the data obtained by
Darcy and Tracey (2007) showed the support for the circular model, with met-
ric constrained MDS in a sample of Grade 8, 10 and 12. However, it is worth
mentioning that the authors did not confirm the circular model using confirma-
tory factor analysis. In contrast, a “RIASCE” arrangement was found in a sample
of Slovenian secondary school students by Boben, Hruševar-Bobek, Niklanovič,
and Lapajne (1993), where the CE and AS types were closer to each other than
expected and R was more distant from the others. The solution of Tien (1997)
did not resemble a circle, whilst ASEC types were arranged very close to each
other in the male subsample. Of the four solutions outlined by Du Toit and Bruin
(2002), only one approximated to a circle (for North West women), nonethe-
less, RIASEC ordering was not observed even in one of the four solutions. In the
remaining three subsamples, many types were close to each other, specifically:
CE and IR types for Eastern Cape province men; SCEA types for North West men
and CE and SA types for Eastern Cape province females. Similar results come
from the study of Armstrong, Hubert, and Rounds (2003). In MDS solution for
UNIACT measure in African American males, types C and E were arranged very
close to each other and for SII measure in African American females, Caucasian
American females and Hispanic American females, types R and I were arranged
very close to each other. For SII measure in a sample of Hispanic American
females, types C and E were arranged very close to each other. MDS solution in
a recent study of Sung, Cheng, and Wu (2015) only remotely resembled a circle
with IRASEC ordering and with types E and C arranged very close to each other.
Using confirmatory factor analysis, they also found poor model fit for two tested
Holland’s models, specifically equality constrained and inequality constrained
model.

548
International Journal for Educational and Vocational Guidance (2020) 20:543–565
1 3
Table 1 The results of the randomization tests of hypothesized order relations in previous research
Source NMean age or age interval Gender Country Scale Correspondence
Index
p value
Tracey and Ward (1998),
Study 1
134 Grade 4–5 b USA Inventory of Children’s
Activities
CIi = .33 pi = .13
CIc = .17 pc = .17
Tracey and Ward (1998),
Study 2
138 Grade 4–5 b USA Inventory of Children’s
Activities
Cii = .39 pi = .08
CIc = .39 pc = .06
Tracey and Ward (1998),
Study 1
607 Grade 6–8 b USA Inventory of Children’s
Activities
CIi = .57 pi = .02
CIc = .61 pc = .02
Tracey and Ward (1998),
Study 2
459 Grade 6–8 b USA Inventory of Children’s
Activities
CIi = .39 pi = .12
CIc = .44 pc = .10
Lent, Tracey, Brown,
Soresi, and Nota (2006)
621 11.58 b Italy Inventory of Children’s
Activities—Revised
CIi = .31 pi = .16
CIc = .25 pc = .21
Darcy and Tracey (2007) 1675 Grade 8 b USA UNIACT .58 p = .01
Sung, Cheng, and Wu
(2015)
1072 14–15 b Taiwan Situation-Based Career
Interest Assessment
.38 p < .05
Šverko and Babarovic
(2006)
318 15 b Croatia Self-Directed Search .31 p < .05
Iliescu, Ispas, Ilie, and
Ion (2013)
1519 15.2 b Romania Self-Directed Search .42 p = .05
Darcy and Tracey (2007) 1675 Grade 10 b USA UNIACT .61 p = .01
Flores, Spanierman,
Armstrong, and Velez
(2006)
487 16.28 F Mexico Strong Interest Inventory
(SII)
CIi = .56 pi = .07
CIs = .31 ps = .07
Flores, Spanierman,
Armstrong, and Velez
(2006)
487 16.28 M Mexico Strong Interest Inventory
(SII)
CIi = .63 pi = .02
CIs = .36 ps = .03
Lent, Tracey, Brown,
Soresi, and Nota (2006)
297 16.36 b Italy Inventory of Children’s
Activities—Revised
CIi = .40 pi = .10
CIc = .39 pc = .10

549
1 3
International Journal for Educational and Vocational Guidance (2020) 20:543–565
Table 1 (continued)
Source NMean age or age interval Gender Country Scale Correspondence
Index
p value
Tracey (2002a, b) 375 16.7 b USA Personal Globe Inventory .80 p = .02
Šverko and Babarovic
(2006); two data col-
lections from different
years
3421998, 3842002 16–17 b Croatia Self-Directed Search CI1998 = .57 p < .05
CI2002 = .67
Šverko, Babarović, &
Međugorac (2014)
528 15 b Croatia The Pictorial and
Descriptive Interest
Inventory
CI = .69 p = .033
Šverko, Babarović, &
Međugorac (2014)
528 15 b Croatia Personal Globe Inventory CI = .72 p = .017
Day, Rounds, and Swaney
(1998)
11,610 Grade 11–12 b USA UNIACT .61 - .75 p < .05
Tracey and Rounds
(1995)
116 High school students M USA The Vocational Prefer-
ence Inventory (VPI)
.47 p = .08
Tracey and Rounds
(1995)
159 High school students F USA The Vocational Prefer-
ence Inventory (VPI)
.22 p = .18
Ryan, Tracey, and
Rounds (1996)
370 High school students b USA The Vocational Prefer-
ence Inventory (VPI)
.79 p = .02
Tien (1997) 788 High school students M Taiwan Chinese Vocational Inter-
est Inventory
.22 Unknown
Tien (1997) 1073 High school students F Taiwan Chinese Vocational Inter-
est Inventory
.17 Unknown
Šverko and Babarovic
(2006); two data col-
lections from different
years
4221998, 4002002 18–19 b Croatia Self-Directed Search CI1998 = .72 p < .05
CI2002 = .86

550
International Journal for Educational and Vocational Guidance (2020) 20:543–565
1 3
Table 1 (continued)
Source NMean age or age interval Gender Country Scale Correspondence
Index
p value
Du Toit and Bruin (2002) 459 19 M South Africa/North West
Province
Self-Directed Search .49 p = .07
Du Toit and Bruin (2002) 144 19 M South Africa/Eastern
Cape Province
Self-Directed Search .35 p = .03
Guglielmi, Fraccaroli,
and Pombeni (2004)
534 19 b Italy Questionnaire sur les
Préférences Profession-
nelles
.83 p < .05
Nagy, Trautwein, and
Lüdtke (2010)
3851 19.57 b Germany Allgemeiner Interessen
Strukturtest
.79 p = .01
Du Toit and Bruin (2002) 573 20 F South Africa/North West
Province
Self-Directed Search .48 p = .08
Du Toit and Bruin (2002) 242 20 F South Africa/Eastern
Cape Province
Self-Directed Search .32 p = .03
F female, M male, b both genders
Subscript i = interests subscale, subscript c = competences subscale, subscript s = self-efficacy subscale

551
1 3
International Journal for Educational and Vocational Guidance (2020) 20:543–565
It can be concluded that studies in adolescent samples did not provide evidence
for the circular spatial arrangement of RIASEC types as hypothesized in Holland’s
theory. However, the fact that in many MDS solutions, some types are either very
close to each other or almost completely merged together can be considered as an
important finding. This means that six RIASEC types may not be the optimal num-
ber to represent all possible interests.
Besides the studies that used RTOR or MDS techniques, there are two factor ana-
lytic studies (Leung & Hou, 2005; Tuck & Keeling, 1980) in which the authors tried
to establish the existence of the six clearly distinguishable RIASEC factors. Fac-
tors identified in their research loaded on different RIASEC subscales, hence dis-
confirming the presence of the six RIASEC types. This is not surprising consider-
ing the hypothesized circumplex nature of vocational interests. There are two main
approaches to study personality traits, values or interests: the factor-list approach,
and the circumplex approach. The goal of the factor-list approach is to identify
“the minimum number of independent factors, or dimensions that characterize the
relations among variables” (Rounds & Tracey, 1993, p. 875). Therefore, in this
approach, factor analysis with a simple structure criterion is a prevailing technique.
However, its application within the framework of circumplex models of personal-
ity traits is not appropriate (Fabrigar, Visser, & Browne, 1997). The exact number
of factors that define variables is not important, of importance is “only the relative
relations among the variables” (Rounds & Tracey, 1993, p. 875). The identifica-
tion of the basic interest dimensions in factor analytic solutions should be impos-
sible because of the presence of confounding response-set variance, also known as
a general factor (Prediger, 1982; Tracey, Rounds, & Gurtman, 1996). The two main
assumptions of the circumplex models are the specific structure underlying con-
structs (interrelatedness of constructs) and the representation of relationships among
the constructs in a two-dimensional space, specifically in a circle (Fabrigar, Visser,
& Browne, 1997, see also Acton & Revelle, 2004, for discussion).
The goal ofthestudy
Many questionnaires based on Holland’s theory are used with students 11years and
older (e.g. in case of the SDS measure). As the results of validity studies conducted
on the samples of adolescents mentioned above are ambiguous, there is a strong
need for more empirical evidence for the construct validity of Holland’s assump-
tions and for the usefulness of his model in interpreting the interests of children in
young adolescence. This need for empirical evidence for construct validity has its
support also in the Standards for Educational and Psychological Testing (AERA,
APA, NCME, 2014, pp. 26–27) formulated as Standard 1.13: “If the rationale for
a test score interpretation for a given use depends on premises about the relation-
ships among test items or among parts of the test, evidence concerning the inter-
nal structure of the test should be provided”. However, in the previously published
validity studies mentioned above, such clear evidence is lacking. Therefore the goal
of this study was to confirm the underlying assumption of Holland’s theory (or the
RIASEC model), specifically the circular spatial arrangement of RIASEC types in

552
International Journal for Educational and Vocational Guidance (2020) 20:543–565
1 3
two samples of Slovak primary school students. Given the fact that the hierarchi-
cal model proposed by Gati (1979) could be superior in explaining the relationships
between RIASEC types, we would like to, similarly to (Einarsdóttir, Rounds, Ægis-
dóttir, & Gerstein, 2002 and Tracey & Rounds, 1993), compare the fit of both com-
peting models.
Method
Participants
Sample 1
The sample consisted of 400 students (205 males and 195 females) from Eastern
Slovakia finishing their primary education (generally focused junior high schools),
with an average age of 13.86 (SD = .73). Out of these, 182 students were in the
eighth grade and 218 in the ninth grade. Although the sample was not randomly
selected, it reflects the proportional size of the residence based on its population.
The sample was intentionally selected to capture the transitional phase between the
primary and secondary education and the processes that are naturally associated
with it. In the Slovak educational system, students after nine years of primary school
have to choose between three types of secondary schools: generally focused gram-
mar schools and two types of specifically career-oriented vocational schools. That
means the first serious vocational choice is done by the age of 15 when students
have to choose one secondary school from these three possible alternatives. The
eighth and ninth school year is, therefore, a period when students often seek help
from an educational counselor aimed at the vocational choice.
Sample 2
To verify our results, we include a second sample, of data originally collected for
another study (Martončik, Babjáková, Čupková, Köverová, & Kačmárová; 2019).
We decided not to merge the participants because questionnaires were administered
to Sample 2 three years later. Only the pretest scores from Sample 2 were used in
the present study. Sample 2 has similar characteristics as Sample 1 and consists of
250 students (128 male, 121 female, 1 not identified) from Eastern Slovakia, with
the average age of 14.14years (SD = .67). Out of these, 135 students were in eighth
grade and 115 in ninth.
Measures
Self-Directed Search, SDS, 5th Edition (Holland & Messer, 2013) is the most fre-
quently used interest questionnaire in history (Spokane & Holland, 1995) and it has
been translated into as many as 25 languages (Holland & Messer, 2013). SDS is
used to measure the degree of resemblance to each of the six RIASEC personality

553
1 3
International Journal for Educational and Vocational Guidance (2020) 20:543–565
types formulated by Holland in his theory of vocational choice (Holland, 1997). The
SDS Assessment Booklet consists of several parts: (a) occupational daydreams—
identifying occupational aspirations, (b) activities—identifying occupational inter-
ests (six scales with 14 items each), (c) competencies—evaluating one’s own abili-
ties (six scales with 14 items each), (d) Occupations—occupational preferences (six
scales with 14 items each), (e) self-estimates—evaluating one’s own abilities com-
pared to others (two sets of six evaluations with two corresponding to one personal-
ity type). The reliabilities of the RIASEC subscales expressed by the omega total
coefficient (Dunn, Baguley, & Brunsden, 2014) were as follows (values for Sample
2 are presented in parenthesis): for the Activities section: R = .88 (.89); I = .83 (.83);
A = .81 (.84); S = .83 (.85); E = .87 (.88); C = .83 (.88); for the Competencies sec-
tion: R = .80 (.84); I = .77 (.76); A = .74 (.77); S = .81 (.80); E = .84 (.87); C = .75
(.75) and for the Occupations section: R = .81 (.86); I = .84 (.86); A = .79 (.82);
S = .80 (.84); E = .88 (.89); C = .86 (.89). Pearson’s correlations between the two
self-estimates ratings were for the RIASEC subscales as follows: rR = .600 (.633);
rI = .390 (.457); rA = .522 (.487); rS = .368 (.423); rE = .466 (.452); rC = .594 (.546).
All correlations were significant at the .01 level. Construct evidence of validity for
the Slovak translation of the SDS are presented in Martončik etal. (2017). The SDS
was administered to students during class by paper and pencil.
Translation process
The permission to translate the SDS into the Slovak language was requested from
Psychological Assessment Resources, Inc. To create the Slovak version, the method
of back-translation was chosen. Both translators had no prior experience or knowl-
edge of SDS. The back translation was sent to the publisher for review. There were
several differences in the translation, especially in the Activities and Occupations
sections. As the four of occupations do not occur in Slovakia or are very rare, they
were substituted for more appropriate alternatives. All behavioral expressions con-
tained in the six RIASEC types of all subscales of the SDS should be fully relevant
for Slovak cultural environment. The first Slovak version was modified and trans-
lated back into English again and then sent for a review. This edition was approved
by the PAR publishing as final.
Data analysis
Overall, only .19% of item responses in Sample 1 and .32% of items in Sample 2
were missing from the datasets and the missing values were handled with the Expec-
tation–Maximization method. The reliability values expressed by the omega total
coefficient (Dunn, Baguley, & Brunsden, 2014) were calculated in jamovi 1.0 (The
jamovi project, 2019).
The assumption of circular spatial arrangement/relationships of RIASEC types
was verified employing the three most commonly used methods, specifically, by
confirmatory and exploratory multidimensional scaling and by the Hubert and Ara-
bie’s randomization test of hypothesized order relations, RTOR. In the confirmatory

554
International Journal for Educational and Vocational Guidance (2020) 20:543–565
1 3
analysis, constrained multidimensional scaling based on stress minimization using
majorization with spherical restrictions implemented in R (R 3.5.1, R Core Team,
2018) package smacof was computed (de Leeuw & Mair, 2009). In an explora-
tory analysis, unconstrained multidimensional scaling with the SPSS ALSCAL
algorithm and the Euclidean distance model was computed. Model fit was esti-
mated in terms of Kruskal’s stress formula, Shepard diagram, and permutation test.
The randomization test of hypothesized order relations, RTOR (Hubert & Arabie,
1987; Rounds, Tracey, & Hubert, 1992; Tracey, 1997) analyzed RIASEC correla-
tion matrices with the intention to test order predictions stated by the model. The
test yields two different indices of the model-data fit: the randomization p value and
the correspondence index (CI). Both indices were calculated using the RANDALL
function (Tracey, 1997), modified into the programming language R (R version
3.4.1; R Core Team, 2017) using the “permute” package (Simpson, R Core Team,
Bates, & Oksanen, 2016).
The datasets collected during the current study are stored at the OSF repository,
specifically, data from Sample 1 are available at: https ://osf.io/cyn3j and data from
Sample 2 can be found at: https ://osf.io/myx3z /.
Results
The circular RIASEC ordering was verified using the randomization test of hypoth-
esized order relations. The fit of the Holland’s (1997), Gati’s (1979), and Gati’s
(1979) alternative model was tested on the whole samples and also separately for
girls and boys within each sample. Correlational matrices for Sample 1 and Sample
2 are presented in Tables2 and 3.
Correspondence indices, p–values, the total number of predictions together with
predictions which were met for the specific model are presented in Table4.
Table 2 SDS summary scores
correlation matrix (N1 = 400,
N2 = 250)
*p < .05; **p < .01.
Sample 1 correlations are below the main diagonal, Sample 2 cor-
relations are above the main diagonal
R I A S E C
R – .078 − .024 − .199** .169** .139*
I .119* – .343** .395** .343** .341**
A− .129** .320** – .489** .305** .314**
S− .177** .469** .584** – .369** .323**
E .166** .318** .268** .424** – .742**
C .050 .412** .266** .454** .760** –

555
1 3
International Journal for Educational and Vocational Guidance (2020) 20:543–565
All the values (six samples, three alternative models) of Correspondence indices
were relatively small with corresponding p values not lower than the threshold .05,
which means that the assumption of the circular arrangement of RIASEC types was
not supported by the data. The results have replicated in both samples.
The circular spatial ordering of individual RIASEC types was verified using the
spherical SMACOF, a multidimensional scaling technique, in which the configu-
ration was restricted to be a circle. The circular spatial representations of the data
obtained from the Self-Directed Search Summary scores are presented in Figure1
together with the Shepard diagrams.
Table 3 SDS summary score correlation matrices separately by gender (female samples n1 = 195 and
n2 = 121 above the main diagonaland male samples n1 = 205 and n2 = 128 below)
*p < .05; **p < .01.
Sample 2 correlations are presented in parenthesis
R I A S E C
R – .363** (.274**) .398** (.391**) .314**(.251**) .233**(.274**) .243** (.293**)
I .083 (.039) – .336** (.270** .518** (.398**) .309** (.361**) .415** (.282**)
A .110 (.145) .350** (.429**) – .470** (.294**) .246** (.247**) .192** (.189**)
S .156* (.047) .519** (.466**) .508** (.517**) – .459** (.474**) .454** (.370**)
E .162* (.066) .329** (.328**) .400** (.480**) .553** (.471**) – .771** (.755**)
C .039 (.076) .409** (.404**) .371** (.504**) .523** (.414**) .760** (.734**) –
Table 4 Results of the randomization test of hypothesized order relations
Sample 2 results are presented in parentheses
Model N1 = 400 Female samples Male samples
(N2 = 250) n1 = 195 (n2 = 121) n1 = 205
(n2 = 128)
Holland (1997)
Predictions in model 72 72 72
Predictions which were met 48 (47) 51 (48) 42 (43)
Correspondence index .33 (.31) .41 (.34) .16 (.19)
p.05 (.06) .05 (.13) .10 (.08)
Gati (1979)
Predictions in model 36 36 36
Predictions which were met 27 (26) 30 (23) 22 (24)
Correspondence index .50 (.44) .66 (.30) .22 (.33)
p.06 (.06) .13 (.13) .13 (.26)
Gati’s (1979) alternative model
Predictions in model 44 44 44
Predictions which were met 32 (28) 36 (35) 34 (25)
Correspondence Index .45 (.27) .63 (.61) .54 (.13)
p.10 (.21) .05 (.06) .11 (.36)

556
International Journal for Educational and Vocational Guidance (2020) 20:543–565
1 3
Figure1 Circular scaling for the RIASEC interest structure (left) together with Shepard diagrams (right)

557
1 3
International Journal for Educational and Vocational Guidance (2020) 20:543–565
Figure2 Two-dimensional scaling solutions for the RIASEC structure

558
International Journal for Educational and Vocational Guidance (2020) 20:543–565
1 3
The stress value for the circular scaling was .221 in Sample 1 and .238 in Sample
2, which indicates a poor fit. Besides the Stress value, the fit of the MDS solution
was also evaluated using smacof’s permutation test (Borg & Mair, 2017) with 1000
replications, with a p value of .751 and permutation stress value equalling .234 in
Sample 1 and with a p value of .925 and permutation stress value equalling .184 in
Sample 2. Therefore, the proposed spatial ordering was not confirmed on the data.
Shepard diagrams in Figure1 presents the relationships between the proximities and
the distances of the point configuration (Mair, Borg, & Rusch, 2016). A high spread
can be interpreted as an indication of a poor fit.
As a next step, hexagonal or circular spatial ordering of RIASEC types was
explored using the unconstrained multidimensional scaling technique. Figure2 pre-
sents the two-dimensional configuration of data obtained from the Self-Directed
Search Summary scores in Samples 1 and 2 respectively. The stress value in Sam-
ple 1 was .092 and the RSQ was .953. The stress value in Sample 2 was .119 and
the RSQ was .930. Kruskal (as cited in Wagenaar & Padmos, 1971) suggested that
stress values less than .10 indicate adequate fit.
Both of the observed multidimensional scaling solutions did not approximate
hexagonal or circular shape, but they did conform to the hypothesized RIASEC
ordering. Type R was located further from other types than expected. At the same
time, type C was closer to type E and types I, A, and S were closer together.
Discussion
The main characteristics of Holland’s model include structuring of vocational inter-
ests into six factors and their spatial arrangement into a hexagon or circle. The
assumption of the circular spatial arrangement was verified using RTOR and MDS
techniques. The results of these statistical analyses are straightforward, indicating
that the structure of vocational interests of Slovak adolescents cannot be appropri-
ately explained by Holland’s and neither Gati’s model.
The results of MDS solutions could mean that adolescents do not consider inter-
ests in I, S, and A types and E and C types to be sufficiently distinct. Relatedness of
E and C types can also be found in MDS solutions of all previous research (Arm-
strong, Hubert, & Rounds, 2003; Boben, Hruševar-Bobek, Niklanovič, & Lapajne,
1993; Du Toit & Bruin, 2002; Sung, Cheng, & Wu, 2015; Tien, 1997). The cir-
cular arrangement of RIASEC types was not supported using the constrained or
unconstrained MDS or RTOR techniques. A lack of evidence supporting the circular
arrangement is consistent with the findings of other authors who studied the struc-
ture of vocational interests in primary school students using the RTOR technique
(e.g. Flores, Spanierman, Armstrong, & Velez, 2006; Iliescu, Ispas, Ilie, & Ion,
2013; Lent, Tracey, Brown, Soresi, & Nota, 2006; Mueller, as cited in Lent etal.,
2006; Šverko & Babarovic, 2006; Tracey & Ward, 1998 only in the male subsam-
ple). The correspondence indices in all of these studies were similarly low. Different
from the hexagonal/circular model is the distance of R type from other types and
the mutual closeness of EC and ISA types, which could be related to career-specific

559
1 3
International Journal for Educational and Vocational Guidance (2020) 20:543–565
gender stereotypes. In this regards, it is appropriate to cite Rounds and Zevon (1983,
p. 496) who argue, that “with respect to Hollands model, the empirical data gen-
erally conform to the RIASEC ordering; the shape of the configurations, however,
rarely approximates a hexagon.”
Differences in MDS solutions could be explained according to Haverkamp, Col-
lins, and Hansen (1994) by culture and ethnicity, which helps to form the structure
of interests. This means that in a collectivist culture, the Social, Enterprising, and
Conventional types are closely related, but in individualistic culture are not. Besides
the cultural reasons, there is also other hypothetical explanation for the absence of
validity of Holland`s theory (developmental reason). It is possible that at a young
age, the interests or the RIASEC personality types (consisted of interests, abilities,
preferred occupations, values, coping mechanisms and self-concept) are not yet sta-
ble and are constantly forming until the end of adolescence. As the adolescents get
older, stability of their interests is increasing and their interest structure resembles
the circular model more and more (Šverko & Babarovic, 2006; Tracey, 2002a). This
assumption is fully supported by the results of several cross-sectional or longitu-
dinal studies (Helwig, 2003; Šverko & Babarovic, 2006; Tracey, 2002a; Tracey &
Ward, 1998). As Tracey (2002a, p. 161) stated: “Interests and competence beliefs
undergo a good deal of change in both structure and level during these years.” A
gradual increase in stability of interests and their resemblance to circular arrange-
ment may, according to Tracey (2001), also be linked to the increased complexity
of information about occupations and interests that is associated with differentia-
tion within the six RIASEC types and their circular arrangement. Meaning that ado-
lescents may have a different outlook (simpler, not as elaborated) on occupations
and activities and skills associated with them than university students and based on
the results of our study we assume that this process could be influenced by gender-
specific upbringing and career specific gender stereotypes (Colley, 1998). Therefore,
the explanation of the structure of interests obtained in the Slovak sample could be
related to the findings of other authors studying leisure activities and interests of
adolescents. As a result of these different outlooks on interests and occupations,
both boys and girls are aware of and prefer those activities that are more accessible
to them, and which they feel more supported in; and on the other hand, they are
less aware of activities which may seem less attractive and therefore more remoted
from their needs and preferences. At the same time, “stereotypical notions of the
characteristics associated with various ‘occupational roles’ and their gender appro-
priateness, are prevalent within society, and these stereotypical beliefs influence the
perceived suitability of occupations” (Lightbody & Durndell, 1996, p. 135). In early
adolescence, the preferred occupations correspond to gender-specific leisure activi-
ties, with sports activities and computer use in boys and social and cultural activi-
ties in girls (Bruyn & Cillessen, 2008; Videnovic, Jelena, & Plut, 2010). Videno-
vic etal. (2010) mention more gender differences, such as gender-oriented school
activities, taking on household responsibilities (for girls) and gaining independence
and earning money (for boys). The remoteness of E and R type from other types
can be interpreted in such a way that the activities and occupations typical for these
types are perceived by adolescents as stereotypically male, while occupations typi-
cal for IAS as stereotypically female. At the same time, it could be caused not only

560
International Journal for Educational and Vocational Guidance (2020) 20:543–565
1 3
by preferences but also by the availability of the activities and interests at this age,
influenced by gender-specific upbringing, which could result, for example, into boys
perceiving IAS types less differentiated than other types. Similarly, Furlong and
Biggarta’s (1999) study reports 13–16years old boys and girls choosing gender-spe-
cific occupations, with the boys preferring careers such as Joiner, Sportsman, Com-
puting, Mechanic, etc. mostly falling under R type. On the other hand, girls prefer
careers such as Teacher, Hairdresser, Nurse, Social worker, etc. corresponding to S
and A type. Slovak students may, too, view certain professions as typically female
(e.g. teacher, psychologist) or typically male (e.g. electrician, farmer) corresponding
to Videnovic’s etal. (2010, p. 210) view that “significant differences in structuring
leisure time indicate that secondary school boys and girls actually belong to different
worlds (they socialize in different ways, gain independence differently, place differ-
ent level of importance on school and sports)”. It is also important to note the limits
of the study, which, as typical in behavioral sciences, usually include the non-rep-
resentative sample. We are aware of the absence of a priori sample size estimation
based on power analysis, but the sample size could not be increased due to financial
constraints. We also did not preregister our analysis plan as we were unaware of this
procedure at that time.
Conclusion andrecommendations
It may be concluded that Holland’s model does not adequately explain the structure
of vocational interests of Slovak primary school students. Therefore, in line with
the recommendation of Lent etal. (2006), caution should be exercised when using
interest questionnaires based on Holland’s model in primary school students. On the
basis of the results of this study and of all those mentioned in the theoretical back-
ground, we would like to highlight two main points:
(1) The results of the validity studies conducted in the sample of adolescents men-
tioned above are ambiguous, rather pointing to inappropriateness of Holland’s
theory in adolescence. We believe that attributing the invalidity of the theory to
cultural factors is not very likely to be correct as the studies with US samples
also produced ambiguous results. If we consider also the presence of publica-
tion bias (see Nelson, Simmons, & Simonsohn, 2018, for discussion), this issue
appears to be even more probable. Despite this fact, questionnaires based on
Holland’s theory, e.g. Self-Directed Search, are widely used in students with age
of 11years and older. More empirical evidence of construct validity is needed in
order that Holland’s RIASEC model can be correctly used for interpreting the
interests of children in young adolescence.
(2) Even though Holland’s model is a predominant model of constructing interest
questionnaires, an effort should be also made to verify its cross-cultural validity
of other promising structural models or measures. Based on existing findings, the
paramount alternative seems to be either the spherical model (Tracey & Rounds,
1996) or the eight-dimensional interest model SETPOINT recently proposed by
Su, Tay, Liao, Zhang, and Rounds (2019). Evidence that supports cross-cultural

561
1 3
International Journal for Educational and Vocational Guidance (2020) 20:543–565
validity of spherical model comes from many non-US adult samples, e.g., from
Switzerland and Burkina Faso (Atitsogbe, Moumoula, Rochat, Antonietti, &
Rossier, 2018), Germany (Etzel, Nagy, & Tracey, 2016), US high-school samples
(Tracey, 2002b) or non-US high-school samples e.g., from Turkey (Vardarlı,
Özyürek, Wilkins-Yel, & Tracey, 2017), Croatia (Šverko, 2008), China (Long,
Adams, & Tracey, 2005) or Ireland (Darcy, 2005). From the MDS solutions
presented in the theoretical background section, the presence of the six RIASEC
types can be questioned. Many RIASEC types in the MDS solutions are posi-
tioned almost next to each other, denoting the existence of a single factor rather
than two separate ones. Therefore, other number of factors should also be veri-
fied, e.g., eight factors proposed in the spherical model and measured by the
Personal Globe Inventory (Tracey, 2002a). The well-established number, six,
of the RIASEC types is only arbitrary and the circular arrangements of interests
“can be broken into any number of components and scales derived to represent
the interest circle” (Tracey & Rounds, 1995, p. 436). PGI incorporates prestige as
a third important dimension of vocational interest structure, “and thus provides
an instrument applicable to a wide range of prestige interests” (Tracey, 2002a,
p. 163).
As a reaction to the empirical evidence not supporting assumptions made from
Holland’s theory, Holland and Gottfredson (1992, p. 169) stated that they are “con-
cerned by the focus of research on the psychometric properties of inventories to the
exclusion of research on the manner in which scores are used and interpreted and the
effects of assessments on clients.” We agree, that the consequences of psychologi-
cal testing are an important part of the validity of measurement. However, construct
evidence is equally important. One of the main characteristics of the methods of sci-
ence is self-correction. If there is adequate, well-powered empirical evidence to test
a theory, and this evidence does not support the assumptions formulated by a theory,
the theory may need to be corrected. Interestingly, no studies with significant results
consider their results as a result of Type I error. On the other hands, nonsignificant
results are frequently considered as a consequence of problematic adaptation of the
scale or its invalidity. The researchers often do not dare to question the validity of
the assumptions of the theory itself.
Funding This work was supported by the Slovak scientific grant agency VEGA [project number VEGA
1/0610/16].
Compliance with ethical standards
Conict of interest The authors declare that they have no conflict of interest.
Informed consent Informed consent was obtained from all individual participants included in the study.

562
International Journal for Educational and Vocational Guidance (2020) 20:543–565
1 3
References
Acton, G. S., & Revelle, W. R. (2004). Evaluation of ten psychometric criteria for circumplex structure.
MPR-Online, 9, 1–27.
American Educational Research Association, American Psychological Association, National Council on
Measurement in Education, Joint Committee on Standards for Educational and Psychological Test-
ing (U.S.). (2014). Standards for educational and psychological testing. Washington, DC: American
Educational Research Association.
Armstrong, P. I., Hubert, L., & Rounds, J. (2003). Circular unidimensional scaling: A new look at group
differences in interest structure. Journal of Counseling Psychology, 50(3), 297–308. https ://doi.
org/10.1037/0022-0167.50.3.297.
Atitsogbe, K. A., Moumoula, I. A., Rochat, S., Antonietti, J.-P., & Rossier, J. (2018). Vocational interests
and career indecision in Switzerland and Burkina Faso: Cross-cultural similarities and differences.
Journal of Vocational Behavior, 107, 126–140. https ://doi.org/10.1016/j.jvb.2018.04.002.
Boben, D., Hruševar-Bobek, B., Niklanovič, S., & Lapajne, Z. (1993). Slovene adaptation of Holland’s
self-directed search. Psihološka obzorja, 3/4(2), 7–19.
Borg, I., & Mair, P. (2017). The choice of initial configurations in Multidimensional scaling: Local min-
ima, fit, and interpretability. Austrian Journal of Statistics, 46(2), 19. https ://doi.org/10.17713 /ajs.
v46i2 .561.
Bruyn, H. E., & Cillessen, A. (2008). Leisure activity preferences and perceived popularity in early ado-
lescence. Journal of Leisure Research, 40(3), 442–457.
Colley, A. (1998). Gender and subject choice in secondary education. In J. Radford (Ed.), Gender and
choice in education and occupation. London, UK: Routledge.
Darcy, M. U. A. (2005). Examination of the structure of Irish students’ vocational interests and com-
petence perceptions. Journal of Vocational Behavior, 67(2), 321–333. https ://doi.org/10.1016/j.
jvb.2004.08.007.
Darcy, M. U. A., & Tracey, T. J. G. (2007). Circumplex structure of Holland’s RIASEC inter-
ests across gender and time. Journal of Counseling Psychology, 54(1), 17–31. https ://doi.
org/10.1037/0022-0167.54.1.17.
Day, S. X., Rounds, J., & Swaney, K. (1998). The structure of vocational interests for diverse racial-ethnic
groups. Psychological Science, 9(1), 40–44. https ://doi.org/10.1111/1467-9280.00007 .
de Leeuw, J., & Mair, P. (2009). Multidimensional scaling using majorization: SMACOF in R. Journal of
Statistical Software. https ://doi.org/10.18637 /jss.v031.i03.
Du Toit, R., & Bruin, G. P. (2002). The structural validity of Holland’s R-I-A-S-E-C model of vocational
personality types for young black South African men and women. Journal of Career Assessment,
10(1), 62–77. https ://doi.org/10.1177/10690 72702 01000 1004.
Dunn, T. J., Baguley, T., & Brunsden, V. (2014). From alpha to omega: A practical solution to the per-
vasive problem of internal consistency estimation. British Journal of Psychology, 105(3), 399–412.
https ://doi.org/10.1111/bjop.12046 .
Einarsdóttir, S., Rounds, J., Ægisdóttir, S., & Gerstein, L. H. (2002). The structure of vocational interests
in Iceland: Examining Holland’s and Gati’s RIASEC models. European Journal of Psychological
Assessment, 18(1), 85–95. https ://doi.org/10.1027//1015-5759.18.1.85.
Etzel, J. M., Nagy, G., & Tracey, T. J. G. (2016). The spherical model of vocational interests in Germany.
Journal of Career Assessment, 24, 701–717. https ://doi.org/10.1177/10690 72715 61612 2.
Fabrigar, L. R., Visser, P. S., & Browne, M. W. (1997). Conceptual and methodological issues in testing
the circumplex structure of data in personality and social psychology. Personality and Social Psy-
chology Review, 1(3), 184–203. https ://doi.org/10.1207/s1532 7957p spr01 03_1.
Flores, L. Y., Spanierman, L. B., Armstrong, P. I., & Velez, A. D. (2006). Validity of the Strong Interest
Inventory and Skills Confidence Inventory with Mexican American high school students. Journal of
Career Assessment, 14(2), 183–202. https ://doi.org/10.1177/10690 72705 28376 0.
Foutch, H., McHugh, E. R., Bertoch, S. C., & Reardon, R. C. (2013). Creating and using a database
on Holland’s theory and practical tools. Journal of Career Assessment, 22(1), 188–202. https ://doi.
org/10.1177/10690 72713 49294 7.
Furlong, A., & Biggart, A. (1999). Framing ‘Choices’: A longitudinal study of occupational aspira-
tions among 13- to 16-year-olds. Journal of Education and Work, 12(1), 21–35. https ://doi.
org/10.1080/13639 08990 12010 2.

563
1 3
International Journal for Educational and Vocational Guidance (2020) 20:543–565
Gati, I. (1979). A hierarchical model for the structure of vocational interests. Journal of Vocational
Behavior, 15(1), 90–106. https ://doi.org/10.1016/0001-8791(79)90021 -6.
Gati, I. (1982). Testing models for the structure of vocational interests. Journal of Vocational Behavior,
21(2), 164–182. https ://doi.org/10.1016/0001-8791(82)90027 -6.
Gati, I. (1991). The structure of vocational interests. Psychological Bulletin, 109(2), 309–324. https ://doi.
org/10.1037/0033-2909.109.2.309.
Guglielmi, D., Fraccaroli, F., & Pombeni, M. L. (2004). Les intérêts professionnels selon le modèle
hexagonal de Holland. L’Orientation Scolaire et Professionnelle, 33(3), 409–427. https ://doi.
org/10.4000/osp.700.
Guttman, L. (1954). A new approach to factor analysis: The Radex. In P. F. Lazarsfeld (Ed.), Mathemati-
cal thinking in the social sciences (pp. 258–348). New York: Russell & Russell.
Haverkamp, B. E., Collins, R. C., & Hansen, J. C. (1994). Structure of interests of Asian-Amer-
ican college students. Journal of Counseling Psychology, 41(2), 256–264. https ://doi.
org/10.1037/0022-0167.41.2.256.
Helwig, A. A. (2003). The measurement of Holland types in a 10-year longitudinal study
of a sample of students. Journal of Employment Counseling, 40(1), 24–32. https ://doi.
org/10.1002/j.2161-1920.2003.tb008 53.x.
Hofer, M. (2010). Adolescents’ development of individual interests: A product of multiple goal regula-
tion? Educational Psychologist, 45(3), 149–166. https ://doi.org/10.1080/00461 520.2010.49346 9.
Holland, J. L. (1973). Making vocational choices: A theory of careers. Englewood Cliffs, NJ:
Prentice-Hall.
Holland, J. L. (1997). Making vocational choices: A theory of vocational personalities and work environ-
ments (3rd ed.). Odessa, FL: Psychological Assessment Resources Inc.
Holland, J. L., & Gottfredson, G. D. (1992). Studies of the hexagonal model: An evaluation (or, the per-
ils of stalking the perfect hexagon). Journal of Vocational Behavior, 40(2), 158–170. https ://doi.
org/10.1016/0001-8791(92)90063 -6.
Holland, J. L., & Messer, M. A. (2013). Self-Directed Search: Professional manual. Lutz, FL: Psycho-
logical Assessment Resources Inc.
Hubert, L., & Arabie, P. (1987). Evaluating order hypotheses within proximity matrices. Psychological
Bulletin, 102, 172–178. https ://doi.org/10.1037/0033-2909.102.1.172.
Iliescu, D., Ispas, D., Ilie, A., & Ion, A. (2013). The structure of vocational interests in Romania. Journal
of Counseling Psychology, 60(2), 294–302. https ://doi.org/10.1037/a0032 199.
Lent, R. W., Tracey, T. J. G., Brown, S. D., Soresi, S., & Nota, L. (2006). Development of inter-
ests and competency beliefs in Italian adolescents: An exploration of circumplex structure and
bidirectional relationships. Journal of Counseling Psychology, 53(2), 181–191. https ://doi.
org/10.1037/0022-0167.53.2.181.
Leung, S. A., & Hou, Z. (2005). The structure of vocational interests among Chinese students. Journal of
Career Development, 32(1), 74–90. https ://doi.org/10.1177/08948 45305 27703 6.
Lightbody, P., & Durndell, A. (1996). Gendered career choice: Is sex-stereotyping the cause or the conse-
quence? Educational Studies, 22(2), 133–146. https ://doi.org/10.1080/03055 69960 22020 1.
Long, L., Adams, R. S., & Tracey, T. J. G. (2005). Generalizability of interest structure to China: Appli-
cation of the Personal Globe Inventory. Journal of Vocational Behavior, 66(1), 66–80. https ://doi.
org/10.1016/j.jvb.2003.12.004.
Long, L., & Tracey, T. J. G. (2006). Structure of RIASEC scores in China: A structural meta-analysis.
Journal of Vocational Behavior, 68(1), 39–51. https ://doi.org/10.1016/j.jvb.2005.01.002.
Mair, P., Borg, I., & Rusch, T. (2016). Goodness-of-fit assessment in multidimensional scaling and
unfolding. Multivariate Behavioral Research, 51(6), 772–789. https ://doi.org/10.1080/00273
171.2016.12359 66.
Martončik, M., Babjáková, J., Čupková, L., Köverová, N., & Kačmárová, M. (2019). Simulation of voca-
tional activities: Experimental evidence of (no) changes in vocational interests. Manuscript submit-
ted for publication.
Martončik, M., Kačmárová, M., Hruščová, E., Žilková, I., Kravcová, M., & Vendel, Š. (2017). Dôkazy
reliability a validity slovenského prekladu dotazníka Self-Directed Search na vzorke adolescentov.
Psychológia a patopsychológia dieťaťa, 51(4), 269–283.
Nagy, G., Trautwein, U., & Lüdtke, O. (2010). The structure of vocational interests in Germany: Different
methodologies, different conclusions. Journal of Vocational Behavior, 76(2), 153–169. https ://doi.
org/10.1016/j.jvb.2007.07.002.

564
International Journal for Educational and Vocational Guidance (2020) 20:543–565
1 3
Nelson, L. D., Simmons, J., & Simonsohn, U. (2018). Psychology’s renaissance. Annual Review of Psy-
chology, 69(1), 511–534. https ://doi.org/10.1146/annur ev-psych -12221 6-01183 6.
Prediger, D. J. (1982). Dimensions underlying Holland’s hexagon: Missing link between interests and
occupations? Journal of Vocational Behavior, 21, 259–287.
Psychological Testing (U.S.). (2014). Standards for educational and psychological testing. Washington,
DC: American Educational Research Association.
R Core Team. (2017, 2018). R: A language and environment for statistical computing. Vienna: R Foun-
dation for Statistical Computing. Retrieved from https ://www.R-proje ct.org/.
Rounds, J., & Tracey, T. J. (1993). Prediger’s dimensional representation of Holland’s
RIASEC circumplex. Journal of Applied Psychology, 78(6), 875–890. https ://doi.
org/10.1037/0021-9010.78.6.875.
Rounds, J., & Tracey, T. J. (1996). Cross-cultural structural equivalence of RIASEC models and measures.
Journal of Counseling Psychology, 43(3), 310–329. https ://doi.org/10.1037/0022-0167.43.3.310.
Rounds, J. B., Tracey, T. J., & Hubert, L. (1992). Methods for evaluating vocational interest struc-
tural hypotheses. Journal of Vocational Behavior, 40, 239–259. https ://doi.org/10.1016/0001-
8791(92)90073 -9.
Rounds, J. B., & Zevon, M. A. (1983). Multidimensional scaling research in vocational psychology.
Applied Psychological Measurement, 7(4), 491–510. https ://doi.org/10.1177/01466 21683 00700 407.
Ryan, J. M., Tracey, T. J. G., & Rounds, J. (1996). Generalizability of Holland’s structure of vocational
interests across ethnicity, gender, and socioeconomic status. Journal of Counseling Psychology,
43(3), 330–337. https ://doi.org/10.1037/0022-0167.43.3.330.
Savickas, M. L., & Gottfredson, G. D. (1999). Holland’s theory (1959–1999): 40years of research and
application. Journal of Vocational Behavior, 55, 1–4. https ://doi.org/10.1006/jvbe.1999.1694.
Shaffer, D. R., & Kipp, K. (2010). Developmental psychology: Childhood and adolescence (9th ed.).
Belmont: Wadsworth.
Simpson, G. L., R Core Team, Bates, D. M., & Oksanen, J. (2016). permute: Functions for Generating
Restricted Permutations of Data. R Package Version 0.9-4.
Spokane, A. R., & Holland, J. L. (1995). The Self-Directed Search: A family of self-guided career inter-
ventions. Journal of Career Assessment, 4(3), 373–390. https ://doi.org/10.1177/10690 72795 00300
409.
Su, R., Tay, L., Liao, H.-Y., Zhang, Q., & Rounds, J. (2019). Toward a dimensional model of vocational
interests. Journal of Applied Psychology, 104(5), 690–714. https ://doi.org/10.1037/apl00 00373 .
Sung, Y.-T., Cheng, Y.-W., & Wu, J.-S. (2015). Constructing a situation-based career interest assessment
for junior high school students and examining their interest structure. Journal of Career Assessment,
24(2), 347–365. https ://doi.org/10.1177/10690 72715 58041 9.
Šverko, I. (2008). Spherical model of interests in Croatia. Journal of Vocational Behavior, 72, 14–24.
https ://doi.org/10.1016/j.jvb.2007.10.001.
Šverko, I., & Babarovic, T. (2006). The validity of Holland’s theory in Croatia. Journal of Career Assess-
ment, 14, 490–507. https ://doi.org/10.1177/10690 72706 28894 0.
Šverko, I., Babarović, T., & Međugorac, V. (2014). Pictorial assessment of interests: Development and
evaluation of Pictorial and Descriptive Interest Inventory. Journal of Vocational Behavior, 84(3),
356–366. https ://doi.org/10.1016/j.jvb.2014.02.008.
The jamovi project. (2019). jamovi (Version 1.0) [Computer Software]. Retrieved from https ://www.
jamov i.org.
Tien, H.-L. (1997). The vocational interest structure of Taiwanese high school students. Paper presented
at the Annual Meeting of the American Psychological Association 105, Chicago, IL, August 1997.
Tracey, T. J. G. (1997). RANDALL: A Microsoft FORTRAN program for a randomization test of
hypothesized order relations. Educational and Psychological Measurement, 57, 164–168. https ://
doi.org/10.1177/00131 64497 05700 1012.
Tracey, T. J. G. (2001). The development of structure of interests in children: Setting the stage. Journal of
Vocational Behavior, 59(1), 89–104. https ://doi.org/10.1006/jvbe.2000.1787.
Tracey, T. J. G. (2002a). Development of interests and competency beliefs: A 1-year longitudinal study of
fifth- to eighth-grade students using the ICA-R and structural equation modeling. Journal of Coun-
seling Psychology, 49(2), 148–163. https ://doi.org/10.1037/0022-0167.49.2.148.
Tracey, T. J. G. (2002b). Personal Globe Inventory: Measurement of the spherical model of interests
and competence beliefs. Journal of Vocational Behavior, 60(1), 113–172. https ://doi.org/10.1006/
jvbe.2001.1817.

565
1 3
International Journal for Educational and Vocational Guidance (2020) 20:543–565
Tracey, T. J., & Rounds, J. (1993). Evaluating Holland’s and Gati’s vocational interest models: A structural
meta-analysis. Psychological Bulletin, 113, 229–246. https ://doi.org/10.1037/0033-2909.113.2.229.
Tracey, T. J. G., & Rounds, J. (1995). The arbitrary nature of Holland’s RIASEC types: A con-
centric-circles structure. Journal of Counseling Psychology, 42, 431–439. https ://doi.
org/10.1037//0022-0167.42.4.431.
Tracey, T. J. G., & Rounds, J. (1996). The spherical representation of vocational interests. Journal of
Vocational Behavior, 48(1), 3–41. https ://doi.org/10.1006/jvbe.1996.0002.
Tracey, T. J. G., Rounds, J., & Gurtman, M. (1996). Examination of the general factor with the interper-
sonal circumplex structure: Application to the Inventory of Interpersonal Problems. Multivariate
Behavioral Research, 31(4), 441–466. https ://doi.org/10.1207/s1532 7906m br310 4_3.
Tracey, T. J. G., & Ward, C. C. (1998). The structure of children’s interests and competence perceptions.
Journal of Counseling Psychology, 45, 290–303. https ://doi.org/10.1037//0022-0167.45.3.290.
Tuck, B. F., & Keeling, B. (1980). Sex and cultural differences in the factorial structure of the Self-
Directed Search. Journal of Vocational Behavior, 16, 105–114. https ://doi.org/10.1016/0001-
8791(80)90042 -1.
Vardarlı, B., Özyürek, R., Wilkins-Yel, K. G., & Tracey, T. J. G. (2017). Examining the structure of voca-
tional interests in Turkey in the context of the personal globe model. International Journal for Edu-
cational and Vocational Guidance, 17(3), 347–359. https ://doi.org/10.1007/s1077 5-016-9338-6.
Videnovic, M., Jelena, P., & Plut, D. (2010). Young people’s leisure time: Gender differences.
Psihologija, 43(2), 199–214. https ://doi.org/10.2298/PSI10 02199 V.
Vondracek, F. W., & Porfeli, E. J. (2003). The world of work and careers. In G. R. Adams & M. D. Ber-
zonsky (Eds.), Blackwell handbook of adolescence (pp. 109–128). Malden, MA: Blackwell Publish-
ing Ltd.
Vondracek, F. W., & Skorikov, V. B. (1997). Leisure, school, and work activity preferences and their
role in vocational identity development. Career Development Quarterly, 45, 322–341. https ://doi.
org/10.1002/j.2161-0045.1997.tb005 37.x.
Wagenaar, W. A., & Padmos, P. (1971). Quantitative interpretation of stress in Kruskal’s multidimen-
sional scaling technique. British Journal of Mathematical and Statistical Psychology, 24, 101–110.
https ://doi.org/10.1111/j.2044-8317.1971.tb004 52.x.
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published
maps and institutional affiliations.