Social Policy: Affirmative Action

Beginning of article

Abstract

Empirical research, summarized here, clearly establishes that racial and ethnic minorities continue to experience substantial discrimination in employment. However, this discrimination is often subtle and unconscious. Because discriminatory practices are so intertwined with apparenth-neutral employment practices, affirmative action remains an important means of combating them. Properly designed, affirmative action can benefit employers and non-protected employees as well as the minorities directly covered.

**********

Acknowledgment

An earlier version of this paper was presented as testimony before the Committee on the Judiciary, California State Assembly, on May 4, 1995. It is based on research supported by the Rockefeller Ford, John D. and Catherine D. MacArthur, Nor-man, Russell Sage, and Public Welfare foundations. However, all findings and conclusions are those of the author.

This paper reviews research on racial/ethnic discrimination in employment conducted by my colleagues and myself at the Fair Employment Council of Greater Washington in Washington, D.C. and other non-profit, non-partisan research organizations. It makes five key points relevant to debate on the controversial issue of affirmative action:

* A substantial amount of racial/ethnic discrimination still operates in the American labor market today;

* "Reverse" discrimination against non-minorities occurs relatively rarely;

* Much of today's discrimination involves subtle cognitive and interpersonal processes;

* When properly implemented, affirmative action remains an important too for addressing these problems; and

* Affirmative action can represent a "win-win" development which benefits employers and white males as well as women and minorities.

This paper discusses each of these points in turn.

Despite Great Progress, Employment Discrimination Has Not Been Eliminated

The first question our research has addressed is: To what extent does racial/ethnic discrimination still operate in the American labor market in the 1990s?

During the past decades, literally thousands of research studies have been conducted on this question by labor economists and other social scientists.(1) The clear consensus of this literature is that a tremendous amount of progress has been made since the days, prior to passage of the federal Civil Rights Act of 1964. when "Jim Crow" laws, personal prejudice, and social custom firmly maintained widespread segregation of employment by race and ethnicity.

These studies also reach consensus (although not unanimously) that the job of erasing the present impacts of these past patterns is not finished. Specifically, racial/ethnic minorities:

* remain under-represented in higher-level occupations and over-represented in lower-level occupations; (2)

* often do not command the same wages as non-minorities for performing the same work; (3)

* often do not receive the same payoffs for acquiring educational credentials; (4)

* on average experience greater unemployment than equally-qualified nonminorities; (5) and

* experience higher rates of job dismissal. (6)

In the statistical studies underlying such conclusions, researchers are careful to consider differences between minority and non-minority workers in education, experience, skills, and other job-relevant qualifications, which account for some of the observed differences between minorities and non-minorities in labor market success. However, technical and data problems often limit researchers' ability to control such factors, so that estimates of the extent of remaining discrimination remain controversial.

To avoid such problems by examining discrimination directly, my colleagues and I employ a research method called "testing" or "auditing." We send out pairs of research assistants to apply for actual job openings listed in the "help wanted" section of newspapers or at a random sample of companies listed in the telephone "Yellow Pages." These research assistants are carefully matched in terms of age, appearance, and personality, and they carry resumes written by experts that credit them with equivalent education and experience. However, each pair of testers consists of persons who differ in their race or ethnicity--an African American tester paired with a white or an Hispanic tester paired with an Anglo. Thus, we set up a controlled experiment in which, if the two testing partners are treated differently, it is reasonable to attribute that treatment to the one way the testers differ: their race or ethnicity.

Since 1990, the Fair Employment Council of Greater Washington and The Urban Institute have run this experiment more than 2,000 times in Washington, D.C., Chicago, and San Diego. (7) The job vacancies tested typically have been for entry-level positions ranging in qualifications from less-than-high-school graduate to college graduate and drawn from a wide variety of industries. We have used both male and female testers and applied for jobs by mail and telephone as well as in-person.

Slightly less than 80 percent of the time that we have sent our testers to apply for jobs, employers did not appear to discriminate between our minority testers and their non-minority counterparts. That is, the equal credentials of the two testers met with equal success--both testers are offered a job, both are rejected, or there is a random alternation in which receives the offer. This finding is consistent with previous research, referred to earlier in this paper, that documents the substantial progress in eliminating discrimination that has occurred since the 1960s.

In the remaining tests--between 20 and 25 percent of our efforts--the outcomes are quite different. (8) With nearly one employer in four, the minority applicant is treated significantly worse than the nonminority. For example:

* Opportunities-to Interview. A vacancy for a receptionist in an optometrist's office in the Washington suburbs was advertised in a local newspaper. When an Hispanic tester called to apply for the job, she was put on hold, and then called by the wrong name (Carmen, when she had given her name as Juanita) and told that they were not taking any further applications. When her Anglo testing partner called thirteen minutes later, she was given an appointment for an interview the following morning.

* Job Offers …