Statistical Methods in Discrimination Litigation

Article excerpt

D. H. Kaye and Mikel Aickin (eds.). New York: Marcel Dekker, 1986. x + 218 pp. $49.75.

Kaye and Aickin persuaded six other authors to join them in writing about statistical methods in discrimination cases. The resulting collection of diverse articles provides interesting glimpses, impressions, and previews of what goes on in such cases. The articles tend to be somewhat fragmentary, raising key legal considerations or sources of statistical bias without resolving the issues. Readers may feel acquainted with some aspects of the legal cases, but not comfortable with all dimensions. The book is not a comprehensive treatment of its subject, since certain types of legal cases, such as those involving discrimination in promotion or termination, are omitted.

The articles vary in difficulty but generally assume a good knowledge of statistical methods. Although several articles examine the relevant legal and statistical issues in depth, the majority are surveys. The book should be useful to both statisticians interested in learning more about the subject and statisticians testifying as experts or serving as analysts for actual legal cases. The survey style of the articles will appeal to the first group, whereas the second group will find the bibliographies extensive and up-to-date. Diligent readers can obtain additional understanding of the subject from the original legal cases and the statistical literature cited.

The articles appear as chapters and address four topic areas. Chapters 1 and 9 provide overviews about the use and misapplication of statistical methods in discrimination cases. Kaye's article (Chap. 1), "The Place of Statistics in Establishing Unconstitutional Acts of Discrimination," is a well-balanced legal discussion that delineates different types of discrimination resulting from violations of equal protection. Citing specific legal cases involving capital sentences, jury selection, and city zoning laws, Kaye shows how statistical evidence is used to establish or refute allegations of discrimination.

Aickin's article (Chap. 9), "Issues and Methods in Discrimination Statistics," focuses on appropriate statistical methods for discrimination litigation. An important point of this article is that relatively little new statistical methodology has been developed explicitly for legal cases. In a detailed and technical discussion, Aickin examines how biased conclusions may result when standard regression methods, chi-squared tests, and structural-equation models are applied to data. The author's use of hypothetical examples to illustrate possible biases is unsatisfactory, however, and evades the pressing question of how much these biases actually matter in specific legal cases.

Chapters 3 and 4 provide informative discussions about the legal issues in discrimination cases. George Rutherglen's article (Chap. 3), "Claims of Employment Discrimination under Title VII of the Civil Rights Act of 1964," outlines how statistical evidence addresses claims of disparate treatment or impact in individual and class-action lawsuits. The numerous Supreme Court cases used for illustration help focus and enrich the article. Rutherglen carefully shows the interplay between statistical evidence and legal reasoning on issues of adverse impact, business justification, and affirmative action. The article is well written and incisive in its ideas.

Elaine Shoben's article (Chap. 4), "Defining the Relevant Population in Employment Discrimination Cases," is a tight, compact discussion about applicant flow and labor-market statistics used in cases of hiring discrimination. Specifically, Shoben examines the sensitivity of the statistical results to alternate definitions of the target population. A district court case alleging sex discrimination in hiring women truck drivers shows nicely how different methods of counting the work force or relevant labor pool influence the statistical results.

Chapters 2 and 8 are well-written discussions of validation and jury-selection procedures. …