Monday, September 29, 2014
In employment-discrimination cases, plaintiffs sometimes present regression analysis to support their claims, particularly for disparate impact claims. In a new paper, Joni Hersch and V. Blair Druhan, “The Use and Misuse of Econometric Evidence in Employment Discrimination Cases,” Washington and Lee Law Review 71(4) (forthcoming December 2014), the authors argue that certain objections by defendants to such regression analysis---omitted variable bias, sample size, and statistical significance—are often invalid yet succeed in making it harder for plaintiffs to prevail. Here’s the abstract:
Experts routinely criticize three aspects of regression analyses presented by the opposing party in employment discrimination cases: omitted explanatory variables, sample size, and statistical significance. However, these factors affect the reliability of the regression results only in very limited circumstances. As a result, valid regression analyses do not provide the critical guidance that they should in employment discrimination cases. Our own statistical analyses of seventy-eight Title VII employment discrimination cases find that merely raising these critiques, even if spurious, reduces plaintiffs’ likelihood of prevailing at trial. We propose that courts adopt a peer-review system in which court-appointed economists, compensated by each party as a percentage of the total payment to econometric expert witnesses, review econometric evidence before the reports are submitted to the judge or jury.
Their sample of Title VII cases comes from a Westlaw search of Title VII cases published between January 2000 and October 2013 in which the words “regression analysis” were present. The total sample includes summary judgment motions, evidentiary motions, trial verdicts, district court opinions and appellate court opinions.