Statistical Issues in the University of Michigan Reverse Discrimination Cases

• Activity Number: 58
• Type: Contributed
• Date/Time: Sunday, August 3, 2003 : 4:00 PM to 5:50 PM
• Sponsor: Social Statistics Section
• Author(s): Malcolm J. Sherman*+
• Companies: The University at Albany, State University of New York
• Keywords: racial preference ; diversity ; multiple regression ; affirmative action ; higher education

Abstract

The Supreme Court is expected to decide in July 2003 two important higher education affirmative action cases that originated at the University of Michigan. Universities generally justify racial preferences in admissions as a means of promoting diversity, but only the U of M has tried (with 850 pages of regressions) to demonstrate empirically that students learn to "think in deeper, more complex ways in a more diverse environment." The statistical evidence proved decisive in one trial court. Critics, including the presenter, subsequently challenged these claims. A distinguished educational researcher who favors racial preferences had previously analyzed the same data and found no connections between educational outcomes and the percent of minorities. The purported benefits relate only to diversity activities (e.g., taking an ethnic studies course), not to the proportion of minorities. According to Michigan's expert witness, diversity activities should be more effective on campuses with more minorities, but the predicted interactions are absent. The talk will describe the statistical disagreements and the role that statistical arguments played in the Supreme Court's decision.