Shadowboxing with Mark Twain: Self-Defense of the Statistical Expert
William E. Feinberg
As a quantitative social scientist, I have served as a statistical expert for the plaintiff in about ten civil cases (although I would not refuse to consider working for the defense, and I now act as a statistical consultant for a defendant1). In this role, I have been identified to the court as a statistical expert, with passing mention of my training as a sociologist. All of my cases have involved suits alleging either race or age discrimination. These are areas in which statistical analysis has come to be an (almost) obligatory part of plaintiff's evidence. I have presented data analysis specific to each suit rather than giving sociological testimony about general patterns of such discrimination in U.S. society.
The major task of the social scientist as statistical expert is to ask statistical questions that complement and enhance the legal issue that is being contested: whether discrimination occurred in the specific situation. One must consider especially whether the data provide evidence leading to statistical conclusions that resist explicit attack from the other side (and sometimes, as we shall see, from the judge). In theory, the task comprises extensions of what a quantitative sociologist does in sociological research. In practice, the changes in milieu from the university to the courtroom and the shift to a contest that involves money and similar considerations rather than an exchange of sociological ideas are important in determining how the task is performed.
Surprisingly, my professional identification as a sociologist has been largely ignored by opposing counsel despite the relatively poor reputation of sociology (compared with economics, for example) in the general populace. Perhaps I perceive no malice toward my identification as a sociologist because I have testified in only one jury trial, but I believe it more likely that opposing counsel regards skepticism about statistics as the point of attack for attempting to discredit my testimony. That does not mean that other potential points of attack are ignored, however. During my lengthy deposition as the