July 2011

IZA DP No. 5831: Bias in the Legal Profession: Self-Assessed versus Statistical Measures of Discrimination

published in: Journal of Legal Studies, 2014, 43(2), 323-357

Legal cases are generally won or lost on the basis of statistical discrimination measures, but it is workers' perceptions of discriminatory behavior that are important for understanding many labor-supply decisions. Workers who believe that they have been discriminated against are more likely to subsequently leave their employers and it is almost certainly workers' perceptions of discrimination that drive formal complaints to the EEOC. Yet the relationship between statistical and self-assessed measures of discrimination is far from obvious. We expand on the previous literature by using data from the After the JD (AJD) study to compare standard Blinder-Oaxaca measures of earnings discrimination to self-reported measures of (i) client discrimination; (ii) other work-related discrimination; and (iii) harassment. Overall, our results indicate that conventional measures of earnings discrimination are not closely linked to the racial and gender bias that new lawyers believe they have experience on the job. Statistical earnings discrimination is only occasionally related to increases in self-assessed bias and when it is the effects are very small. Moreover, statistical earnings discrimination does not explain the disparity in self-assessed bias across gender and racial groups.