On the relationship between indirect measures of Black versus White racial attitudes and discriminatory outcomes: An adversarial collaboration using a sample of White Americans

The idea that racial prejudice contributes to discrimination not only deliberately but also in a more automatic fashion has been one of the most prominent topics in social psychological research in the past 30 years. Much of the evidence for theories of automatic prejudice stems from the use of indirect measures of implicit attitudes, yet meta-analyses give differing estimates regarding the predictive validity of such measures. The present adversarial collaboration provides a test of the relationships between prominent measures of implicit racial attitudes and discriminatory behavior using a set of established lab-based paradigms among a sample of White Americans (N = 2,114). Using structural equation models that can account for measurement error, frequentist and Bayesian multiverse analyses confirmed that White Americans' performance on indirect measures correlate modestly with these behavioral outcomes, and explain unique variance (similar to 2.5%) beyond direct, self-report measures of racial attitudes. At the same time, self-report measures exhibited greater predictive and incremental validity than indirect measures (explaining similar to 45% of the variance) despite behavioral measures of discrimination displaying weak internal reliability. Results provided some support for greater predictive and incremental validity for indirect measures among participants scoring relatively low on measures of executive function and motivation to control prejudice. These results lend themselves to both relatively optimistic and pessimistic interpretations concerning scientific and practical significance. All collaborators agree that the best path forward is collaborative and focused on the generalizability of implicit racial attitudes to high-accountability organizational settings.