Aggregating data across several test administrations is a useful strategy for increasing the statistical power of adverse impact analysis.... Show moreAggregating data across several test administrations is a useful strategy for increasing the statistical power of adverse impact analysis. Typically, before such an analysis is conducted, a test of homogeneity is conducted to ensure that the degree of adverse impact is consistent across samples. An alternative approach would be to use hierarchical linear modeling to estimate the average and variability in adverse impact. The current study explored the patterns of variability of adverse impact in a police officer selection test across test administrations, departments, and geographic regions. Significant mean test score differences were found between African-American and White test takers. Further, the size of mean group differences varied significantly across test administration and departments, but not between geographic regions. The implications of these findings for scientists and practitioners alike are discussed. M.S. in Psychology, May 2014 Show less
