Structure of vocational interests in Serbia: Evaluation of the spherical model

https://doi.org/10.1016/j.jvb.2007.12.004Get rights and content

Abstract

To explore the structure of vocational interests in Serbia, 1063 participants of various age, education and gender completed the Serbian version of the Personal Globe Inventory [PGI, Tracey, T. (2002). Personal Globe Inventory: Measurement of the spherical model of interests and competence beliefs. Journal of Vocational Behavior, 6, 113–172]. The fit of Tracey’s octagonal and spherical models to the structure of correlations between Tracey’s eight and 18 type scales was evaluated, as was the fit of Holland’s hexagonal, Gati’s, Rounds–Tracey’s and Liu–Rounds’ models to the structure of correlations between RIASEC scales. Randomization test of hypothetical orders, multidimensional scaling with fixed coordinates and Myors’ test were used to assess the fit of models to the data. The results showed good fit of Tracey’s octagonal and spherical models to the data. Holland’s hexagonal model also fit the data very well, but the fit of Gati’s, Rounds–Tracey’s and Liu–Rounds’ models of relations between RIASEC types was either nonsignificant or very low.

Introduction

Since 1970s the model of relations between vocational interests proposed by John Holland (Holland, 1959, Holland, 1994) has been the dominant model in the field of vocational interests. The model proposed that there are six types of interests—Realistic (R), Investigative (I), Artistic (A), Social (S), Enterprising (E) and Conventional (C), commonly referred to as RIASEC types, that are positioned at nodes of an equilateral hexagon. Types that are closer on the hexagon are more similar, while types farther away are less similar. This also means that correlations between each two types correspond to their spatial distances. Neighboring types have highest positive correlations, while correlations between opposite types are the lowest or highest negative. This makes it possible to create and test a number of specific structural hypotheses claiming that correlations between some types are higher than correlations between some other types. For example, correlation between R and I types, which are neighboring types, should be higher than correlation between R and A types. If sets of activities and vocations that should “go together” according to theory do not really do so in reality, than validity of any conclusion based on the theory is questionable. Testing the described structural hypotheses is important whenever the theory or an instrument for measuring types is applied on a new population.

Several methods have been proposed and used for testing Holland’s model structural hypotheses. Hubert and Arabie (1987) randomization test of hypothetical orders, especially as implemented by Tracey’s RANDALL program (Tracey, 1997b) has become a very popular test for this purpose and has been used by many researchers.

Researchers in this area have also used multidimensional scaling with or without restrictions, a method proposed by Myors, 1996, Myors, 1998, which is based on calculating Spearman’s correlation coefficients between obtained and theoretical rankings between types and circular unidimensional scaling (CUS).

The results of these studies have not been uniform across countries. If we use, the above described, correspondence coefficient as a measure of model fit—relatively high levels of fit of the model to the data were found on some US samples, Japan, Iceland, Israel, South Korea, Serbia and Croatia (Einarsdóttir et al., 2002, Feldman and Meir, 1976, Fouad and Mohler, 2004, Hedrih, 2006, Hedrih and Šverko, 2007, Rounds and Tracey, 1996, Šverko and Babarović, 2006, Tak, 2004, Tracey et al., 1997, Yang et al., 2005). Lower levels of fit were found in a much larger number of countries—India (Leong, Austin, Sekaran, & Komarraju, 1998), both mainland China and Hong Kong, although there were a few Chinese samples with high levels of fit (Long & Tracey, 2006), Bolivia (Glidden-Tracey & Parraga, 1996), South Africa (Du Toit & De Bruin, 2002), Spain—Basque Country (Elosua, 2007), Portugal, Pakistan, Indonesia, Brazil, Paraguay, France, Columbia, Mexico, Australia and some other countries (Rounds & Tracey, 1996). The order of obtained levels of statistical significance of the randomization test of hypothetical orders in these studies also very closely match the described order.

As for the studies of vocational interests in Serbia, Hedrih (2006) conducted a study on a sample of 362 using a RIASEC-based instruments. The results showed high levels of fit of the data to the predictions of the Holland’s hexagon. The reported significance level of the randomization test of hypothetical orders was 0.017, and the correspondance coefficient was 0.791 for the Holland’s hexagon, indicating, a relatively high level of fit. These results were later compared with those obtained by Šverko in Croatia (Šverko & Babarović, 2006) and the obtained levels of fit in Serbia were very similar to the levels of fit found in Croatia on an equivalent sample (Hedrih & Šverko, 2007).

As an alternative to Holland’s hexagonal ordering of types Gati (1979) proposed a hierarchical ordering of types with three clusters of vocational interests (RI, AS and EC). Rounds and Tracey (1996) proposed a somewhat different placement of interest types into clusters—RI, A, SEC, and Liu and Rounds (2003, according to Long & Tracey, 2006), based on the results of their research on Chinese population, proposed an octagonal structure of RIASEC interests in which the structure of vocational interests is represented by an equilateral octagon in which there is a free unnamed node between R and C types and also between I and A types.

Prediger, 1982, Prediger, 1999, Prediger, 2000 proposed that there are two dimensions of vocational interests underlying the hexagon, that define the plane in which the hexagon lies. He named these two dimensions people-things and ideas-data. People-things dimension passes through nodes where S and R types are, while the ideas-data dimension is orthogonal to it, cutting the hexagon in half at a point equally distant from types A and I on one side (ideas side of the dimension) and E and C on the other.

Tracey and Rounds (1995) explored the structure of vocational interests by using preference responses to a variety of occupational titles and examined their placement along Holland’s circle and found occupational titles to be uniformly distributed along Holland’s circle. They then reasoned that if Holland’s types were truly distinct types, titles would cluster around six nodes corresponding to Holland’s types. As this was not the case, they concluded that any number of types distributed along the circle would be equally good as Holland’s six types. Tracey (1997a) later showed that the finding about uniformity of the distribution was also the case with other types of items commonly used for measuring vocational interests—activity preferences and competence self-evaluations.

Tracey and Rounds (1995), and Tracey (1997a) proposed that there is a third dimension of vocational interests, alongside two dimensions proposed by Prediger, which they named prestige and proposed a spherical model of interests which states that the structure of interests consitutes a sphere rather then a circle. The plane in which Holland’s RIASEC types are located is positioned at the equator of the vocational interests sphere. For measuring interests according to this model Tracey (2002) constructed the Personal Globe Inventory (PGI) and a corresponding vocational interests typology which, in its current version, has 18 types of vocational interests. Eight out of those 18 types are equidistantly distributed along the equator of the sphere, in the Holland’s plane, to form an equilateral octagon. These types are called basic interest types. Five other types are positioned in the upper hemisphere of the model and represent higher prestige types, and five are positioned in the lower hemisphere comprising lower prestige types. In relation to Prediger’s dimensions, there is one Tracey’s type positioned at + and −22.5 degrees of each of the four poles of these two dimensions. One higher and one lower prestige type is positioned at the top and the bottom of the sphere, respectively. Each of the other four higher and lower prestige types are positioned in one of the quadrants of their hemisphere.

PGI measures interests using three types of items—activity preferences, activity competence self-evaluations and occupation preferences. Apart from Tracey’s 18 types, PGI provides scores for Holland’s RIASEC types and basic dimensions of the spherical model, as well as on four types positioned in the basic interests plane at poles of each of Prediger’s dimensions. We will use the term “Holland’s plane” or the “basic interest plane” to describe the equatorial plane of the spherical model on which Holland’s RIASEC and Tracey’s basic interest types are located. RIASEC type measures are obtained either as linear combinations of Tracey’s types from the basic interest plane or are considered equivalent to certain Tracey’s types. Due to this, actual positions of RIASEC types on the spherical model, as obtained on PGI, are not equidistant, but the deviations are rather small and they do not compromise the order of correlations between types predicted by Holland’s hexagonal model.

Furthermore, unlike Holland’s model, neighboring types of the spherical model are not placed ideally equidistantly from each other in all directions. While neighboring types on the same height are equidistant, these distances are not equal to distances between neighboring types on different heights.

Apart from the initial confirming results obtained in the US and Japan (Tracey, 1997a, Tracey, 2002, Tracey et al., 1997), the spherical model was so far evaluated in Ireland (Darcy, 2005) and China (Long, Adams, & Tracey, 2005). At the time our study began in Serbia, a similar evaluation study was underway in Croatia, conducted by Iva Šverko (Šverko, 2007). These studies have all reported results similar to those obtained on Tracey (2002) original US sample, confirming the spherical model.

The goal of this study was the evaluation of the spherical model in Serbia, and apart from that, as we also obtained measures of RIASEC types, we evaluated Holland’s hexagonal and three alternative model of relations between RIASEC types.

Section snippets

Participants

Participants were people of working age from several regions in Serbia conveniently recruited by interviewers. Data were collected from 1063 participants of various age, education and gender. Mean age of participants was between 33 and 34 years with the oldest participant being 65 and the youngest 18 years old. 476 participants were male and 587 were female. Most of the participants had a 4 years secondary school education. Participants with the lowest education level had finished trade school

PGI psychometric characteristics

I first examined the internal consistency of PGI scales. Results showed that all PGI scales had very good internal consistency with Cronbach’s α coefficient ranging from .85 for one of the basic interest types to .97 for basic dimensions. When each type of items was analyzed separately the occupational preference items Social facilitating scale had a reliability of .68, but reliabilities of all the others were higher.

Next, I analyzed the relations between measures of vocational interest types

Acknowledgments

I thank Terence J.G. Tracey for the permission to create the Serbian version of PGI and for many useful suggestions. I also thank Iva Sverko for making available the Croatian version of PGI and help with the creation of the bilingual Serbian and Croatian web version of the same instrument.

This research was partly supported by a research grant from the Ministry of Science and Environment Protection of the Republic of Serbia within the framework of the 149062D research project.

References (33)

  • D.J. Prediger

    Dimensions underlying Holland’s hexagon missing link between interests and occupations?

    Journal of Vocational Behavior

    (1982)
  • D.J. Prediger

    Holland’s hexagon is alive and well—though somewhat out of shape: Response to Tinsley

    Journal of Vocational Behavior

    (2000)
  • M. Savickas et al.

    Convergent and discriminant validity of five interest inventories

    Journal of Vocational Behavior

    (2002)
  • T. Tracey

    Personal Globe Inventory: Measurement of the spherical model of interests and competence beliefs

    Journal of Vocational Behavior

    (2002)
  • D.T. Campbell et al.

    Convergent and Discriminant Validation by the MultitraitMultimethod Matrix

    Psychological Bulletin

    (1959)
  • R. Du Toit et al.

    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

    (2002)
  • Cited by (28)

    • Evaluation of the dimensions of the spherical model of vocational interests in the long and short version of the Personal Globe Inventory

      2019, Journal of Vocational Behavior
      Citation Excerpt :

      The same should be true for the other two dimensions. Based on the results of former validation studies (e.g., Hedrih, 2008; Tracey, 2002), we expected all of these relationships to be positive and of considerable magnitude. Corresponding PGI dimensions obtained from different item types will be positively associated.

    • Vocational interests and career indecision in Switzerland and Burkina Faso: Cross-cultural similarities and differences

      2018, Journal of Vocational Behavior
      Citation Excerpt :

      Overall, these tests yielded similar CI values for both countries and genders. The CI values found in Switzerland for the 6-type, 8-type, and 18-type structures, and those found in Burkina Faso for the 6- and 8-type structures were similar to those found in several other countries and cultures including the US (Tracey, 2002), China (Long et al., 2005), Japan (Long et al., 2006), Serbia (Hedrih, 2008), Croatia (Šverko, 2008), and Germany (Etzel et al., 2015). However, the CIs found for the 18-type structure across both Burkinabe gender and overall sample were low compared to those found in the above-mentioned cultures.

    • Pictorial assessment of interests: Development and evaluation of Pictorial and Descriptive Interest Inventory

      2014, Journal of Vocational Behavior
      Citation Excerpt :

      PGI measures 18 vocational interests by using three different methods: liking of activities, competence in activities and liking of occupations. PGI gained wide cross-cultural validity support, for example in the US (Tracey, 2002), Ireland (Darcy, 2005), China (Long, Adams, & Tracey, 2005), Japan (Long, Watanabe, & Tracey, 2006), Croatia (Šverko, 2008), Serbia (Hedrih, 2008), and the Caribbean (Wilkins, Ramkissoon, & Tracey, 2013). In this study we applied just eight interest scales of a medium prestige (measured only with liking of activities subtest), which can also provide results in the RIASEC framework.

    View all citing articles on Scopus
    View full text