|
Article Excerpt
Japanese college students (N = 2,492) completed the translated Personal Globe Inventory (PGI: T. J. G. Tracey, 2002b) Occupational Title scales, and the structural validity of vocational interests was examined. Support was provided for the RIASEC (Realistic, Investigative, Artistic, Social, Enterprising, Conventional), octant, and spherical structure of the PGI Occupational Title scales. In addition, there were no structural differences found between men and women.
**********
With the greater ease of cultural communication and increasing globalization, there is an expanding interest in applying measures that have been generated on one culture to another culture. Such practices can be questionable given the possible different structural properties and underlying meanings of the measures across cultures (e.g., Ben-Porath, 1990; Berry, 1989; Buss & Royce, 1975; Irvine & Carroll, 1985; Prediger, 1994). Without equivalent structures, the explanation and comparison of the measure scores across cultures are meaningless. Given this reality, it is imperative that one of the first steps in examining cross-cultural applications of any measure is to establish the structural invariance. The purpose of the present study was to examine the structural validity of the Occupational Title scales of the Personal Globe Inventory (PGI; Tracey, 2002b) as applied with Japanese college students.
Holland's (1973, 1997) circular interest model of six interest types (Realistic, Investigative, Artistic, Social, Enterprising, and Conventional, known as RIASEC) has dominated in vocational interest assessment for more than 30 years, and it serves as an important starting point in examining cultural equivalence of interests. If the RIASEC scales are not related to each other in a similar manner across cultures (in this context being the circular ordering), then the scales are being interpreted differently in each culture and thus they have different meanings. To claim cross-cultural construct validity of a scale score, it is thus necessary to demonstrate a certain similar pattern of relations of the scale score with other scales or measures. If the RIASEC scale scores have different relations to other scale scores in one culture than they do in another, this is evidence against the proposition of construct equivalence.
The research has provided equivocal support of the cross-cultural equivalence of the RIASEC structure. Several studies have supported the validity of Holland's model of interest structure across cultures (e.g., Harrington & O'Shea, 1993; Holland, 1985; Yu & Alvi, 1996), whereas others have not (e.g., Farh, Leong, & Law, 1998; Fouad & Dancer, 1992; Glidden-Tracey & Parraga, 1996; Leong, Austin, Secaran, & Komarraju, 1998; Long & Tracey, in press). Hansen (1987) noted these mixed results in her early review of the literature. Rounds and Tracey (1996; Tracey & Rounds, 1993) conducted two meta-structural analyses on the studies based on Holland's theory and found that the RIASEC model fit the U.S. sample well, but was not as good a fit for either the international samples or the U.S. ethnic samples, which indicated that structural equivalence across culture cannot be assumed and that it needs to be examined in each application.
The PGI globe model (Tracey, 2002b) is a more general representation of the structure of interests than the RIASEC model. It includes not only the RIASEC types but also other types and an added dimension. The PGI globe model comprises three dimensions and 18 interest types, which are equally spaced around a sphere. Whereas the RIASEC circular model consists of the circular ordering of six types in two dimensions, often called people/things and data/ideas (Prediger, 1982; Prediger & Vansickle, 1992; Rounds & Tracey, 1993), the PGI globe model has three dimensions--the same two dimensions as the RIASEC model and the added dimension of prestige. Tracey and Rounds (1995) demonstrated that the RIASEC type model is an arbitrary representation of interests in the people/things and data/ideas space and that interests types can be validly formed by any equal slicing of the circle. As a result, they proposed an eight-type model, octants, as an alternative representation of interests. These octants have been refined and renamed in the PGI globe model and are presented along with the RIASEC types in Figure 1.
The addition of the third dimension of prestige means that interests would also vary with respect to prestige and that when interests were examined a spherical structure resulted. Putting the octant model and the third dimension together, Tracey and Rounds (1996) created scales to represent 18 points equally placed around the sphere and presented their model as a globe. This spherical model was reconceptualized and refined in the PGI, and it is this globe representation that is presented in Figure 2. On the equator of the globe are eight basic interest types instead of the six RIASEC interest types: Social Facilitating, Managing, Business Detail, Data Processing, Mechanical, Nature/Outdoors, Artistic, and Helping. The PGI globe also has five higher prestige interest types on the top hemisphere--Social Sciences, Influence, Business Systems, Financial Analysis, and Science--and five lower prestige interest types on the bottom hemisphere--Quality Control, Manual Work, Personal Service, Construction/Repair, and Basic Services.
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
The PGI globe model is a more general representation of interests than the RIASEC types because it incorporates greater differentiation of basic interests as well as the dimension of prestige. It can represent RIASEC scores but, in addition, it can yield more differentiated octant scales and 10 other scales with varying degrees of prestige. Structural support has been provided for this model on a sample of U.S. high school and college students (Tracey, 2002b; Tracey & Rounds, 1996). The most recent implementation of this globe model is the PGI instrument (Tracey, 2002b), which consists of three item types: activity preferences, competency estimates, and occupational preferences. Each item type has received structural support with respect to fitting the RIASEC, octant, and spherical structures across ages (high school and college students), gender, and U.S. ethnic groups.
Early examinations of the generalizability of the spherical model of the PGI to non-U.S. samples have yielded some support. Tracey, Watanabe, and Schneider (1997) found support for the octant and the spherical models on a sample of Japanese college students using an early version of the PGI (i.e., the Inventory of Occupational Preference [IOP]), which consisted of various occupational titles for which respondents endorsed their preferences. They also found that the octant model fit the Japanese sample better than did the RIASEC model. The PGI is an improved and condensed version of the IOP; however, it is not known if these aforementioned results would hold, given the changes to the PGI. Although the spherical model fit the Japanese data well, the spherical model fit the U.S. sample better than the Japanese sample (Tracey et al., 1997), suggesting the possibility of structural variance across cultures. With the improvement of item selection in the PGI instrument and the refinement of the scales, we hypothesized that the fit to the spherical model...
|
