creator May, Jill L. advisor Morris, Scott B. With increased diversity in the U.S. workforce, there has become a greater need for organizations to defend against accusations of discrimination and thus a need to produce more accurate adverse impact statistics in discrimination cases. The purpose of the present study is to investigate two adjustments to a confidence interval procedure for calculating adverse impact to see if they give more adequate performance over the unadjusted Morris and Lobsenz (2000) confidence interval procedure. A Monte Carlo simulation revealed that using an adjustment for small sample sizes alleviates some of the problems found for the confidence interval procedure found in previous research. Specifically, adding a 0.5 adjustment to the confidence interval procedure increases the accuracy over the unadjusted procedure and the Agresti and Caffo (2000) procedure. Implications for practioners and researchers are included in the discussion. Submitted by Dana Lamparello (dlampare@iit.edu) on 2012-08-21T19:09:34Z No. of bitstreams: 2 Thesis_JM_Final.pdf: 730152 bytes, checksum: f994f6b2dbe60532955e97b577492805 (MD5) My thesis signed title page0001.pdf: 282774 bytes, checksum: 30eb4a59da227d24c4f3490c842404de (MD5) Made available in DSpace on 2012-08-21T19:09:34Z (GMT). No. of bitstreams: 2 Thesis_JM_Final.pdf: 730152 bytes, checksum: f994f6b2dbe60532955e97b577492805 (MD5) My thesis signed title page0001.pdf: 282774 bytes, checksum: 30eb4a59da227d24c4f3490c842404de (MD5) Previous issue date: 2012-05 M.S. in Psychology, May 2012 2012-03-27 2012-05 http://hdl.handle.net/10560/2807 en Thesis born digital application/pdf In Copyright http://rightsstatements.org/page/InC/1.0/ Restricted Access PSYCH / Institute of Psychology Illinois Institute of Technology Affiliated department