Using Intersectional Implicit Association Measures Does Not Consistently Improve the Predictive Validity of the Implicit Association Test

Historically, studies of implicit associations have mostly measured one identity at a time, such as investigating associations based on race or gender. Here, we explore whether intersectional implicit associations (e.g., attitudes towards Black men) predict relevant behavior better than implicit associations based on a single social identity (e.g., attitudes towards Black people). Across four studies (total N > 8000) using multiple measures of implicit attitudes, we find that these intersectional implicit associations were never better predictors of intergroup behavior than associations based on a single identity. Our findings challenge the assumption that greater correspondence between attitude and behavioral target enhances predictive validity, suggesting that, in time-pressured judgment contexts, perceivers may default to single-identity representations even for intersectional targets. This work highlights boundary conditions under which intersectional implicit measures are likely to add value.