Table 1.

Bias in implicit race attitude predicts trust disparity independent of explicit race attitude

Experiment: Trustworthiness ratings, n = 48
Dependent variable: Trust disparity
Final model: r2 = 0.329, P < 0.001
Independent predictors	IAT, EMS, IMS, MRS, SRS, LIB/CON, participant race
Factors	Standardized β	Significance in final model (P)	Change in r2
IAT	0.376	0.003	0.142
LIB/CON	0.430	<0.001	0.187

Experiment: Modified Trust Game, n = 43
Dependent variable: Offer disparity
Final model: r2 = 0.247, P < 0.003
Independent predictors	IAT, EMS, IMS, MRS, SRS, LIB/CON, participant race
Factors	Standardized β	Significance in final model (P)	Change in r2
IAT	0.358	0.014	0.166
EMS	0.289	0.044	0.081

Separate stepwise regression analyses (probability of F to enter, P = 0.05; probability of F to remove, P = 0.10) for disparity in ratings (study 1) and offers (study 2) found that IAT scores independently accounted for a significant portion of the variance in both, even when accounting for explicit race attitudes and participant race (n = 48 for the analysis of data from study 1 because two participants did not complete the explicit measures portion of the experiment). EMS, External Motivation to Avoid Prejudice Survey; IMS, Internal Motivation to Avoid Prejudice Survey; MRS, Modern Racism Scale; SRS, Symbolic Racism Scale; LIB/CON, political leaning scale (Liberal/Conservative); participant race, white/nonwhite.