Are the Japanese more depressed than the U.S. individuals? — An IRT-DIF study —

Abstract

This study analyzed ethnocultural differences in responses to items measuring depressive symptoms in demographically matched samples of Japanese and U.S. white collar workers (N = 368 for each). Using item response theory with a two-parameter logistic model (2PL-IRT), differential item functioning (DIF) of items comprising the Center for Epidemiologic Studies Depression Scale was assessed for Japanese and U.S. workers. The 2PL-IRT analysis revealed that the item characteristic curves (ICCs) were generally identical between (lack of) positive affect items and other negative symptom items among the U.S. workers, but not among the Japanese workers. Most of (lack of) positive affect items show­ed DIF between the groups: i.e., the Japanese tended to endorse such responses even at much lower depressive (latent trait) levels, as compared to the U.S. work­ers. Although some negative symptom items yielded significant DIF, averaged ICCs were comparable between the groups. Because of the sample-independent nature of IRT, these results provided more robust evidence to the Iwata hypothesis that Japanese respondents tend to inhibit positive affect expression.