Gender Differences in Item Endorsement of An Anxiety-Depression Scale by Chilean Youth: An Analysis with IRT-Differential Item Functioning

Background and Purpose: Recent work in the U.S. has suggested that males and females differ in the types of depression items they endorse. However, this work on gender differences in the endorsement of anxiety and depression items is not only in its infancy, it is also mainly relegated to U.S. populations. To address this gap in knowledge, we conducted a study to examine whether adolescent males and females may differentially endorse anxiety and depression items with a large sample of Chilean adolescents. Methods: We used cross-sectional data from the Santiago Longitudinal Study (SLS), a NIDA-study of community-dwelling adolescents in Santiago, Chile. The sample consists of 925 participants (Mean age = 14, SD=1.3, 49% females) of mid-to-low socioeconomic status. Youth completed a two-hour interviewer-administered questionnaire with measures that assessed mental health problems, drug use, and relationship with parents, among others. Interviews were conducted in Spanish in a private office at the University of Chile by Chilean psychologists trained in the administration of standardized instruments. The measurement examined in this study was the 13 items of the Youth Self-Report (YSR) anxious/depressed problem scale (Achenbach & Rescorla, 2001). A three-step approach was utilized. First, exploratory factor analysis was conducted to examine the potential factors underlying the anxiety-depressed scale. We found two potential factors. Second, we used confirmatory factor analysis to test if the two factor structure is better than the one factor structure. Results were in favor of the two factor structure. Third, using a two factor structure, we fit a two-parameter logistic (2PL) IRT DIF model assuming the mean DIF is 0. Results: Results revealed differential item functioning for six items including “I cry a lot” (according to ETS criteria, C-level item, p<0.005 and β=1.31) and “I feel too guilty” (A-level item, p<0.05 and β=-0.625). These results imply that adolescent females with the same latent level of depression as adolescent males were more likely to endorse the item “I cry a lot” by a factor of 3.6 odds ratio. On the other hand, males with the same latent level of anxiety as females were about fifty percent more likely to endorse the item “I feel too guilty'.   Conclusions and Implications: The first finding from this research suggests that a two-factor depression and anxiety structure best explains Chilean adolescents' manifestation of these constructs. With regards to the results of the DIF analysis, our second set of findings indicate that among this population females were more likely to endorse the item about crying and males were more likely to endorse the item about feeling guilty. Together, the results suggest that some anxiety and depression items have different meanings for Chilean adolescent females and males when controlling for latent anxiety and depression. Several cultural aspects of the Chilean culture may explain these findings. These are discussed in relation to implications for the assessment of depression and anxiety with international youth.