Item response theory and confirmatory factor analysis: Two approaches for testing the psychometric equivalence of translated tests

Translating psychological tests from one language and culture to other lang uages and cultures is a common practice in cross-cultural research. However , researchers should not assume that the translation is perfectly equivalen t to the original source language version; instead of that it is necessary to carry out DIF analysis to test the equivalence of the different versions of the questionnaire. Confirmatory Factor Analysis (CFA) and item Response Theory (IRT) models are alternative procedures for the detection of DIE Th e objective of this study is to compare the utility of CFA and IRT models f or testing the psychometric equivalence of translated tests. The English ve rsion of the PSDQ was administered to 986 Australian teenagers, and a trans lated Spanish version of the questionnaire was administered to 986 Spanish teenagers. CFA with nonzero Intent means were carried out to test the equiv alence of both versions of the questionnaire; and in the context of IRT Sam ejima's Graded Response Model was used with the same purpose. The results o f this study show that CFA with nonzero latent means allows to identify uni form DIF as well as nonuniform DIF, so the use of AFC is comparable to the use of IRT models.