Interests as predictors of performance: An omitted and underappreciated variable

Recent meta-analyses indicated that interest congruence can predict performance both at work and in school. Given these findings, the exclusion of interests from predictive models of performance has several potential consequences. In this study, we demonstrate that excluding vocational interests from models of performance has implications for the validity of the model as well as for understanding predictive bias in selection testing. Using a sample of 1449 students from a large university, we examined the validity and incremental validity of vocational interests for predicting academic performance above and beyond ACT scores, high school GPA, and other noncognitive predictors. Building on recent research, we also demonstrate that including vocational interests in the prediction model eliminates the predictive bias observed for ACT and HSGPA across men and women but not across racial/ethnic subgroups. The implications of these results for understanding performance and future research needs in the area of vocational interests are discussed.