Measurement equivalence of English and Spanish versions of the Campbell Interest and Skill Survey

Gabriela Guerrero, University of Texas at El Paso

Abstract

This study illustrates an application of the item response theory (IRT)-based differential functioning of items and tests (DFIT; Raju, van der Linden, & Fleer, 1995) framework to assess the fidelity of a Spanish translation of the Campbell Interest and Skill Survey (CISS; Campbell, Hyne, & Nilsen, 1992) across three culturally- and linguistically-different groups: (a) English-speaking Anglo-Americans, (b) English-dominant Hispanic-Americans, and (c) Spanish-speaking Mexican nationals. A two-parameter logistic dichotomous model (Birnbaum, 1968; Hambleton & Swaminathan, 1985; Hulin, Drasgow, & Parsons, 1983; Lord, 1980) and the DFIT framework were used to evaluate differential functioning (DF) at both the test and item levels. DFIT provided two indices of DF at the item level [i.e., a noncompensatory differential item functioning (NCDIF) and a compensatory differential item functioning (CDIF)], and one index at the test or scale level [i.e., the differential test functioning (DTF)]. Across the three two-group comparisons, an average of 7.6% of the items were identified as DF using the CDIF index and an average of 40% were identified as DF using the NCDIF index. The difference in the number of DF found using the CDIF and the NCDIF indices highlights the benefits of allowing items to compensate for each other's DF. ^ Using Hofstede's (1980, 1991) uncertainty avoidance dimension, predictions were made regarding which items would exhibit DF. The scales that were expected to show DF (i.e., Organizing and Creating orientations) did not show a greater proportion of DF than the remaining scales, suggesting no support for using cultural differences to hypothesize DF items. Finally, as expected, the number of items identified as functioning differentially using the CDIF index was greater in the different-language comparisons (i.e., Anglo versus Mexican and Hispanic versus Mexican) than in the same-language comparison (i.e., Anglo versus Hispanic). Post hoc examinations of several NCDIF items provide information about the probable sources of DF (i.e., translation errors, differences in cultural relevance across groups, or both). Benefits of using the DFIT framework over traditional DIF assessment methods in future measurement equivalence studies are discussed. ^

Subject Area

Language, Modern|Education, Guidance and Counseling|Psychology, Psychometrics

Recommended Citation

Guerrero, Gabriela, "Measurement equivalence of English and Spanish versions of the Campbell Interest and Skill Survey" (2001). ETD Collection for University of Texas, El Paso. AAI3035097.
http://digitalcommons.utep.edu/dissertations/AAI3035097

Share

COinS