Race, Sex, and Performance Ratings in the Federal Service

Abstract

The Civil Service Reform Act of 1978 required all federal agencies to develop performance evaluation systems to for periodic appraisals of job performance of and to use the results...as a basis for training, rewarding, reassigning, promoting, demoting, retaining, and separating employees (Dresang, 1991, 132). Thus, federal performance appraisals are intended both to provide unbiased measures of job performance and to strongly influence federal career patterns. If they achieve either of these purposes, they can provide important clues to racial and sexual inequality in the federal service. If performance appraisals accurately measure productivity on the job, they can help us assess the commonly heard argument that affirmative action has led to reverse discrimination against white men and has too often placed less qualified women and minorities in positions where they are unlikely to succeed. If the federal service has promoted large numbers of women and minorities past better-prepared white men, then women and minorities should be receiving lower performance evaluations than white men at the same grade levels. Alternatively, if women and minorities have to work harder and be better to earn as much as white men, then women and minorities may receive higher ratings than white men at the same grades. If performance appraisals have a major impact on promotions, demotions, and retentions, then racial and sexual differences in ratings could help explain why women and minorities remain substantially underrepresented at the top levels of the federal civil service and earn substantially less than white males even when they have comparable levels of education and experience (Taylor, 1979; Grandjean, 1981; Borjas, 1982; DiPrete, 1989). Critics (Filer, 1983; Polachek, 1976) fault such studies for failing to capture differences in important unmeasured characteristics (e.g., effort, intelligence, quality of education, personality) that result in real productivity differences among apparently comparable workers. If performance appraisals objectively measure productivity and if women and minorities really are less productive than comparably educated and experienced white men, then (legitimately) lower ratings could help explain why women and minorities earn less than apparently comparable white men. Alternatively, if discriminate in their performance appraisals, those (illegitimately) lower ratings may be stunting women's and minorities' career growth. For performance ratings to be playing an important (proper or improper) role in explaining racial and sexual pay inequality, however, white men must be receiving higher performance evaluations than others with similar characteristics. In this article, I investigate whether performance ratings vary substantially with race and sex. I first review some arguments for doubting the accuracy of performance appraisals, and some of the previous research on racial and sexual differences in performance evaluations I then consider which employee and position characteristics we need to control in looking for race and sex differences. Using a 1 percent sample of federal personnel records for 1990 through 1995, I look for race and sex differences in ratings (1) overall, (2) among employees at the same grade levels, (3) among employees at the same grades in the same agencies, and (4) among comparably educated and experienced employees in comparable positions. How Accurate Are Performance Appraisals? Are performance appraisals meaningful enough that racial and sexual difference in ratings would suggest anything about differences in performance? In a study of the research on performance appraisals and merit pay commissioned by the U.S. Office of Personnel Management (OPM), the National Research Council concludes that supervisors can give reliable ratings on employee performance under controlled conditions and with carefully developed rating scales. …