Law and Statistics of Combining Categories: Walmart and Employment Discrimination Cases

• Activity Number: 488
• Type: Invited
• Date/Time: Wednesday, August 4, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Committee on Law and Justice Statistics
• Author(s): Albert J. Lee*+ and Alan J. Salzberg
• Companies: Summit Consulting, LLC and Analysis & Inference, Inc.
• Keywords: reference class ; Chow Test ; grouped data ; discrmination ; Simpson's Paradox

Abstract

The New York Times reported (August 31, 2009): "the Obama administration is planning a major revival of high-impact civil rights enforcement against policies, in areas ranging from housing to hiring, where statistics show that minorities fare disproportionately poorly." What statistics should the DOJ bring to bear in prosecuting such cases and how can companies ensure their practices are appropriate? The typical case will compare the minority group at the company to some reference group, but how does the selection of that group affect the outcome? We look at two discrimination cases, against Johnson & Johnson and Wal-Mart, and find that careful, or careless, choice of a comparison group can result in entirely different statistical conclusions.