Affirmative action, duality of error, and the consequences of mispredicting the academic performance of african american college applicants

Authors

  • Jeryl L. Mumpower,

    1. Center for Policy Research and Department of Public Administration and Policy, Nelson A. Rockefeller College of Public Affairs and Policy, State University of New York, Albany
    Search for more papers by this author
  • Radhika Nath,

    1. Center for Policy Research and Department of Public Administration and Policy, Nelson A. Rockefeller College of Public Affairs and Policy, State University of New York, Albany
    Search for more papers by this author
  • Thomas R. Stewart

    1. Center for Policy Research and Department of Public Administration and Policy, Nelson A. Rockefeller College of Public Affairs and Policy, State University of New York, Albany
    Search for more papers by this author

Abstract

The implications of different potential affirmative action policies depend on three factors: selection rate from the applicant pool, base rate of qualified applicants, and accuracy of performance predictions. A series of analyses was conducted under various assumptions concerning affirmative action plans, causes of racial differences in average college admissions test scores, and racial differences in accuracy of performance predictions. Evidence suggesting a lower level of predictive accuracy for African Americans implies that, under a program of affirmative action, both proportionately more false positives (matriculated students who do not succeed) and proportionately more false negatives (rejected applicants who could have succeeded) will be found among African American applicants. Unless equivalent levels of predictive accuracy are achieved for both groups, no admission policy can be fair simultaneously to majority group applicants and African American applicants. © 2002 by the Association for Public Policy Analysis and Management.

Ancillary