What do implicit measures of bias actually measure?

Abstract

Borne out of the limitations posed by self-reports, social psychologists developed implicit measures capable of assessing unconscious bias (e.g., the IAT). Scepticism towards the IAT has grown in recent years, however, with studies revealing the weak relationship between explicit and implicit measures and the apparent disconnect between implicit bias and behaviour. This has led researchers to call for innovative ways to measure the key processes underpinning implicit bias. The aim of the current study was to develop a novel battery of behavioural measures capable of assessing implicit racial bias. In a within-participants design, 257 participants completed a battery of socio-cognitive measures that were adapted to feature a race-based component. We pre-registered the prediction that participants would show more imitative tendencies, higher empathy, better perspective-taking and emotion recognition towards people of their own race. Moreover, it was predicted that these indicators of own-race bias would be related to implicit racial bias. Findings indicate that participants exhibited better emotion recognition but poorer perspective taking and empathic concern for ingroup relative to outgroup members. None of the measured socio-cognitive mechanisms correlated with IAT scores. These findings are discussed in relation to the construct, discriminant and predictive validity of the IAT.