Table 3.
Contact, Role Modeling of Racial Bias, Climate, and Diversity Training in Medical School as Predictors of Explicit and Implicit Bias Against Black People in the Second Year of Medical Residency
| Explicit bias | Implicit bias | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Individual model (ICC = 1.1e-20 to 9.9e-15)a | Full model (ICC = 3.0e to 13; R2 = .547747)b | Individual model (ICC = 6.0e-16 to 1.4e-10)c | Full model (ICC = 7.0e to 16; R2 = .169533)d | |||||||||
| Predictor | β | SE | p | β | SE | p | β | SE | p | β | SE | p |
| Wave 2: quality of contact with Black peoplee | 0.08 | 0.02 | < .001 | 0.03 | 0.02 | .025 | –0.06 | 0.02 | .002 | –0.04 | 0.02 | .024 |
| Wave 2: quantity of contact with Black peoplef | 0.04 | 0.02 | .018 | 0.01 | 0.02 | .40 | −0.02 | 0.02 | .27 | −0.01 | 0.02 | .60 |
| Wave 2: perceived welcoming racial climate | 0.04 | 0.02 | .023 | 0.01 | 0.02 | .48 | −0.01 | 0.02 | .62 | 0.02 | 0.02 | .38 |
| Wave 2: observed role modeling of racial bias | −0.03 | 0.01 | .081 | −0.00 | 0.02 | .98 | 0.02 | 0.02 | .34 | 0.01 | 0.02 | .71 |
| Wave 2: hours of diversity training | 0.00 | 0.02 | .86 | −0.01 | 0.02 | .49 | −0.00 | 0.02 | .93 | 0.01 | 0.02 | .72 |
| Wave 1: quality of contact with Black peopleg | 0.05 | 0.01 | < .001 | 0.03 | 0.01 | .040 | –0.06 | 0.02 | < .001 | −0.03 | 0.02 | .074 |
| Wave 1: quantity of contact with Black peopleh | 0.05 | 0.02 | .003 | 0.02 | 0.02 | .37 | –0.06 | 0.02 | < .001 | –0.04 | 0.02 | .033 |
| Wave 2: bias | 0.33 | 0.03 | < .001 | 0.30 | 0.03 | < .001 | 0.24 | 0.02 | < .001 | 0.23 | 0.02 | < .001 |
Note: For explicit-bias models, positive values indicate warmer attitudes (less bias) because the dependent variable was explicit attitudes toward Black people and the model controlled for explicit attitudes toward White people. For implicit-bias models, positive values indicate more bias. All predictors and dependent variables have been standardized. All models include a random intercept by school and account for the effect of stratum as a categorical predictor. All values are based on pooled estimates across multiple imputations. Values in boldface are statistically significant. ICC = intraclass correlation coefficient.
aThese columns show results for models with individual predictors of explicit attitudes toward Black people during residency, controlling for explicit attitudes toward White people during residency, baseline explicit attitudes toward Black and White people, and demographics. bThese columns show results for one model with the five predictors of attitudes toward Black people during residency, controlling for explicit attitudes toward White people during residency, baseline and end-of-medical-school explicit attitudes toward Black and White people, and demographics. cThese columns show results for models with individual predictors of Implicit Association Test (IAT) D scores, which index pro-White implicit bias, controlling for baseline IAT D score and demographics. dThese columns show results for one model with the five predictors of IAT D score, controlling for baseline and end-of-medical-school IAT D score and demographics. eThe individual predictor model controlled for baseline quality of contact with Black people. fThe individual predictor model controlled for baseline quantity of contact with Black people. gThe individual predictor model did not control for quality of contact with Black people during medical school. hThe individual predictor model did not control for quantity of contact with Black people during medical school.