How Appropriate Analysis Can Improve the Legal System’s Appreciation of the Strength of Statistical Evidence Submitted in EEO Cases

• Activity Number: 143 - Critical Role of Statistics in Evaluating Real World Evidence for Legal and Regulatory Applications
• Type: Invited
• Date/Time: Monday, July 29, 2019 : 10:30 AM to 12:20 PM
• Sponsor: Stats. Partnerships Among Academe Indust. & Govt. Committee
• Author(s): Joseph Lewis Gastwirth*
• Companies: George Washington University
• Keywords: equal employment; statistical evidence; stratified data; common misunderstandings; regression analysis

Abstract

Although the Supreme Court accepted statistical analysis of data arising in jury and employment discrimination cases in 1977, judges may not appreciate the insights careful statistical analyses can provide. Courts often make “ad hoc” judgments concerning the required sample size, which can lead to one court accepting and another rejecting numerically similar data. Several courts have had difficulty understanding the analysis of data, stratified by job type or location and confuse a proper combination test of promotion or layoff data with a simple aggregation of the data into one large 2x2 table. These and other statistical issues will be illustrated on data from a variety of cases; in some, the courts were misled and others where proper analysis would strengthen the logic underlying a decision. A few appellate judges have observed that both parties ignored a statistical problem. If time permits, statistical procedures resolving these issues will be given.