Joseph L. Gastwirth. San Diego, CA: Academic Press, 1988. Vol. 1: xxii + 465 pp.; Vol. 2: viii + 458 pp. $84.50 each.
A few years ago, a journalist observed that "what demonstrative evidence was to the 1960s and the early 1970s, statistics has become to the 1980s--the hottest way to prove a complicated case" (Lauter 1984, p. 10). With the publication of Statistical Reasoning in Law and Public Policy (SRLPP), what is heard in the courtroom may move into the classroom. Gastwirth, a professor of statistics and economics at George Washington University who has served as a consultant and expert witness in many a case, has produced a two-volume work "designed to serve as a textbook for students of law and public policy as well as a resource for individuals" (Vol. 1, p. xi). I discuss the value of the volumes for both teaching and reference. Although I realize that some statisticians may adopt this book for their undergraduate students, my discussion emphasizes its potential for courses designed primarily for law students, since this is the group with which I am familiar.
As a textbook, these volumes are unique. Although rumor has other statisticians preparing texts on forensic statistics, Gastwirth's is the first of the breed. Because it is so new, I have not used it in the classroom; as such, I can offer no firm pronouncements on its pedagogical value. It has formulas and exercises, but it emphasizes the meaning of the formulas rather than their derivation or the details of the computations. Most of Volume 1 presents basic statistical concepts and methods in forensic or policy contexts. The remainder of Volume 1 and most of Volume 2 discuss applications, often with data drawn from litigation. Volume 2 describes a few more-advanced statistical methods.
The discussion of descriptive statistics in Volume 1 includes ogives, histograms, and typical measures of central tendency and spread. Newer exploratory data-analytic techniques are slighted, but measures of relative inequality and concentration are introduced because of their usefulness in legislative apportionment, tax assessment, antitrust, and other cases. A concise presentation of probability theory, estimation, and hypothesis testing follows, with a chapter devoted specifically to the comparison of proportions--a ubiquitous problem in employment-discrimination cases. The final two chapters in this volume concern comparison of two distributions and correlation and regression analysis. Volume 2 explains matched- or paired-data analysis, clinical trials, and Bayesian inference.
The strength of Gastwirth's introduction to statistics is the attention to legal applications. There are already several nice nonmathematical introductions to statistics available. For instance, I have found Kimble (1978) and Moore (1985) to work well with law students; others will have their favorite texts. I doubt that SRLPP represents a significant advance in clarity or entertainment over this existing literature. Nevertheless, the specific field of study--law and public policy--influences Gastwirth's selection of basic topics and the manner of their presentation.
Furthermore, SRLPP is more comprehensive and more detailed in its coverage than general-purpose undergraduate materials. For example, it discusses the hypergeometric distribution, which has confused some courts (Kaye 1986); the procedures for combining results from related 2 x 2 tables, which have been overlooked by some courts (Kaye 1985); logistic regression, which figured prominently in the unsuccessful challenge to the application of the death penalty in McCleskey v. Kemp (1987); and many biostatistical and epidemiologic concepts that have become decisively important in much litigation and environmental regulation (Fienberg 1988).
The applications chapters are varied and topical. They include sampling evidence, jury selection, employment discrimination, epidemiologic proof in tort and environmental law, and the use of medical and other screening tests. …