ABSTRACT
Discrimination is sometimes the product of what the discriminator takes to be a fundamental distinction between classes of people. This chapter focuses on an important form of discrimination, which is often labeled statistical discrimination. The discussion deals with those forms of discrimination that are alleged by the discriminator to rest on a statistically valid empirical distinction, a correlation, between two (or more) people or categories. Statistical discrimination is thus defended by reference to the instrumental benefits of using a distinction that appears to have indicative or predictive value in pursuing a legitimate end existing independent of the discrimination itself. When the basis for the statistical discrimination is race, ethnicity, national origin, religion (sometimes), gender, or sexual orientation, however, the nature of the discussion changes, and the implicit presumption in favor of statistical discrimination in many domains shifts to a presumption in the opposite direction. There is nothing inherently problematic about this differential treatment.
