The preceding article by Dr. Richard M. Cohn' concerning the use of statistics in Title VII employment discrimination cases makes three basic points. First, Cohn rejects the methods used to assess disproportionate differences between groups on tests, such as ability tests. He finds fault both with the approach of the Uniform Guidelines on Employee Selection Procedures and with the method based on finding statistical significance that I have advocated. Second, he also rejects the approach courts have adopted for evaluating the relative exclusion of groups defined by race, sex, or national origin in the employer's work force. He argues that this method of comparing the composition of the employer's work force to the relevant labor pool is an inaccurate probe of whether discrimination has occurred. Finally, Cohn offers an alternative
method for analyzing employment records which he believes should more accurately detect discrimination.
This reply contends that Cohn's arguments are based on a misunderstanding of the legally relevant questions under Title VII and thus that he proposes the use of particular quantitative methods that provide answers to the wrong questions. The exceptions that this reply takes with each of the three sections of Cohn's article are not concerning any disagreement over statistics. To the contrary, we are in perfect agreement as to the underlying assumptions that must be met with situations using statistical inference. Rather, our disagreements revolve around questions of law and logic.
55 Indiana L.J. 515 (1980).
Shoben, Elaine W., "In Defense of Disparate Impact Analysis Under Title VII: A Reply to Dr. Cohn" (1980). Scholarly Works. Paper 581.