Reverse Discrimination by the Numbers.

Racial discrimination can be defined as unequal (unfavorable) treatment of an individual on account of his or her racial or ethnic group membership.(1) Discrimination involves denying to someone a privilege or reward that he would have otherwise received except for the fact of his group membership.(2) The opposite of racial discrimination is equal treatment of individuals according to a single standard.(3) In the usual accounts of racial discrimination, it is members of majority ethnic groups who discriminate against members of minority groups. It describes behavior that was typical in the Jim Crow era of the American South and elsewhere.(4)

This clear and precise definition of racial discrimination allows for the possibility of "reverse discrimination" or the operation of "racial preferences." The possibility of reverse discrimination appears paradoxical. Standard accounts of inter-group relations assume that all groups operate to a certain extent to favor their own members (of the in-group) over others (members of the out-group). In the case of reverse discrimination, however, members of the majority group are said to discriminate against other members of the majority group in favor of members of the minority group. (Alternatively, it is possible that members of the minority group discriminate against members of the majority group in favor of members of the minority group.)(5)

Over the last twenty-five years, charges of reverse discrimination resulting from the operation of racial preference programs in employment and government contracting have become quite common. They are prevalent in the world of education--we illustrate this study with our analysis of how reverse discrimination in the form of racial favoritism has colored admissions decisions at the University of Virginia. In California, such charges supplied the basis for the passage of Proposition 209; in the state of Washington they led to Initiative 200. In addition, they have been the subject of several critical court cases such as Regents of the University of California v. Bakke (1978), Podberesky v. Kirwan (1995), and Hopwood v. University of Texas (1996) in higher education, Wygant v. Jackson Board of Education (1986) in employment, and City of Richmond v. J.A. Croson Co. (1989) and Adarand Constructors, Inc. v. Pena (1995) in contracting.(6)

Under certain circumstances, racial discrimination is easy to detect because it is official state policy written into statutes and defined in legal cases. Often, however, this adverse treatment is hidden from public view, either because the policy is illegal, because it is considered to be shameful, because it is unpopular, or for any combination of these reasons. Reverse discrimination can be equally difficult to detect conclusively.

In these circumstances, statistical evidence can play an important role in describing racial and ethnic disparities, and in helping to ascertain whether disparities in outcomes between members of different racial and ethnic groups are likely due to discrimination or to other causes. Although it is sometimes forgotten, racial discrimination, when it exists, is only one of many complex causes of social behavior. The influence of these external causes needs to be statistically controlled in order to determine whether or not racial discrimination is present.

The statistical study of racial discrimination employs the same procedures used in studying any such presumed causal relationship. There exists a well-developed set of procedures that provide substantial assistance in ascertaining whether or not such causal interpretations of statistical data are plausible. We draw freely on this literature here.(7)

Statistical Determination of Causality