Blacks are also more likely to mispredict their credit worthiness. We produce two
types of false negatives. The first is the probability that the respondent believes she has
bad credit when in fact the credit score is within one standard deviation of the mean
credit score for all respondents. The second is the probability that the respondent
believes he has bad credit but he does not fall into the “bad credit” bucket defined by
Freddie Mac. The first false negative is 0.11 for whites and 0.19 for blacks. The second
false negative is 0.14 for whites but 0.23 for blacks. Both racial differences in false
negatives are statistically significant.3 The unconditional probabilities of self-rated bad
credit are also higher for blacks than they are for whites, although many of the selfrated
bad credit respondents indeed have bad credit. The racial gap in self-rated good
credit, moreover, is far larger than the racial gap in the self-rated bad credit.