Skip to Main Content
Table 6. Predictive Bias for Race/Ethnicity
  Holdout Data  Training and Holdout Combined 
  b β 95% CI p  b β 95% CI p 
Depression predicted 3.15 .25 .11 .39 .001  4.55 .72 .66 .78 < .001 
 race/ethnicity 0.01 .01 -.26 .29 .924  0.01 .01 -.10 .13 .848 
 predicted X race/ethnicity 0.82 .07 -.22 .35 .649  0.29 .05 -.08 .17 .460 
Anxiety predicted 2.23 .22 .09 .36 .001  3.96 .68 .61 .75 < .001 
 race/ethnicity -0.14 -.15 -.43 .12 .271  -0.03 -.03 -.15 .10 .652 
 predicted X race/ethnicity 1.38 .14 -.14 .41 .323  0.46 .08 -.05 .21 .243 
Post-Traumatic Stress predicted 2.88 .23 .08 .37 .002  4.42 .72 .66 .79 < .001 
 race/ethnicity 0.05 .19 -.08 .47 .173  0.02 .06 -.06 .18 .322 
 predicted X race/ethnicity 2.41 .19 -.10 .48 .195  0.57 .09 -.04 .23 .171 
Anger predicted 3.92 .23 .10 .36 .001  5.03 .70 .64 .76 < .001 
 race/ethnicity -0.05 -.06 -.33 .22 .691  -0.03 -.03 -.15 .09 .594 
 predicted X race/ethnicity 0.30 .02 -.25 .28 .897  0.18 .03 -.09 .14 .677 
  Holdout Data  Training and Holdout Combined 
  b β 95% CI p  b β 95% CI p 
Depression predicted 3.15 .25 .11 .39 .001  4.55 .72 .66 .78 < .001 
 race/ethnicity 0.01 .01 -.26 .29 .924  0.01 .01 -.10 .13 .848 
 predicted X race/ethnicity 0.82 .07 -.22 .35 .649  0.29 .05 -.08 .17 .460 
Anxiety predicted 2.23 .22 .09 .36 .001  3.96 .68 .61 .75 < .001 
 race/ethnicity -0.14 -.15 -.43 .12 .271  -0.03 -.03 -.15 .10 .652 
 predicted X race/ethnicity 1.38 .14 -.14 .41 .323  0.46 .08 -.05 .21 .243 
Post-Traumatic Stress predicted 2.88 .23 .08 .37 .002  4.42 .72 .66 .79 < .001 
 race/ethnicity 0.05 .19 -.08 .47 .173  0.02 .06 -.06 .18 .322 
 predicted X race/ethnicity 2.41 .19 -.10 .48 .195  0.57 .09 -.04 .23 .171 
Anger predicted 3.92 .23 .10 .36 .001  5.03 .70 .64 .76 < .001 
 race/ethnicity -0.05 -.06 -.33 .22 .691  -0.03 -.03 -.15 .09 .594 
 predicted X race/ethnicity 0.30 .02 -.25 .28 .897  0.18 .03 -.09 .14 .677 

Note.b refers to the unstandardized estimate; β refers to the standardized estimate. These analyses compared racial/ethnic majority (White and not Hispanic/Latino; nholdout = 147, ncombined = 446) vs. minority participants (either non-White or Hispanic/Latino; nholdout = 75, ncombined = 211); racial/ethnic majority/minority identity was contrast coded (racial/ethnic majority = -0.5; racial/ethnic minority = 0.5) so positive values for race/ethnicity indicate a higher intercept for racial/ethnic minority participants (relative to racial/ethnic majority participants) and positive values for the predicted by race/ethnicity interaction indicate a steeper slope for racial/ethnic minority participants (relative to racial/ethnic majority participants).

Close Modal

or Create an Account

Close Modal
Close Modal