Jeanrie, C., Laval University, Canada and Van de Vijver, F., Tilburg university, Netherlands
The California Psychological Inventory (CPI) is widely used and has been translated many times. Equivalence between the translated versions and the original, however, has not been thoroughly studied and thus leaves open the question of is generalizability to other culturally or linguistically different populations. Since the CPI is intended to measure folk (common sense) concepts , its author (Gough, 1956; 1995) consider the test as being cross-culturally valid. For this assumption to be valid, translated CPI item has to carry their original meaning, even for those scales that were developed through an empirical-keying strategy. In order to adapt the CPI-434 to the French Canadian population, the test was translated using the conceptual translation method (Jeanrie & Bertrand1999) which focuses primarily on the equivalence in the meaning item instead of on their linguistic equivalence. Consequently, it was hypothesized that this translation method could lead to the equivalence of the test item between two groups presenting linguistic and cultural differences. Subjects are 2000 students from either the English or French parts of Canada. Logistic regression analyses were performed on each item to assess the presence of differential item functioning between the two groups. Results confirm the hypothesis and show that very few item are biased. Moreover the few item showing DIF do not tend to be systematically biased against one specific group.