Adverse Impact Implications of Selection Instrument Group Score Differences.
Abstract
Human resources decision-makers are concerned when mean inter-group score differences on selection measures are observed. Moreover, they are not concerned with the magnitude of the differences per Se, but rather with whether those score differences will manifest themselves as adverse impact. An analytical approach was used to estimate for various combinations of selection ratio and minority applicant group representation. the maximum group score difference that would not violate the "four-fifths" rule. In addition. applicant pools of specific sizes with no mean inter-group score difference on the selection measure were considered to compute the conservative likelihood of encountering an adverse impact situation in a specific applicant sample. The results clearly suggest that the identification of adverse impact and its statistical substantiation will often occur in small applicant pools (i.e. -. 100), even when there is a small inter-group difference on the selection measure. For larger samples (i.e., 500), the results suggest that adverse impact will often he indicated when small mean inter-group selection measure differences are present. It is not clear to what degree the adverse impact found would be statistically substantiated. Research focusing on adverse impact and its statistical substantiation is needed for specific inter-group difference/applicant pool size combinations to create a clearer equivalence between intergroup differences and adverse impact.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1997
- Accession Number
- ADA338809
Entities
People
- Jay M. Silva
Organizations
- U.S. Army Research Institute for the Behavioral and Social Sciences