Meta-Analysis of Integrity Tests: A Critical Examination of Validity Generalization and Moderator Variables.

摘要

A comprehensive meta-analysis was conducted to investigate whether integrity test validities are generalizable and to estimate differences in validity due to potential moderating influences. The database included 665 validity coefficients across 576,464 data points. Results indicate that integrity test validities are positive and in many cases substantial for predicting both job performance and counterproductive behaviors on the job such as theft, disciplinary problems, and absenteeism. Validities were found to be generalizable. The estimated mean operational predictive validity of integrity tests for supervisory ratings of job performance is .41. For the criterion of counterproductive behaviors, results indicate that use of concurrent validation study designs may overestimate the predictive criterion-related validity applicable in selection situations. Our results based on external criterion measures (i.e., excluding self reports) and predictive validity studies using applicants indicate that integrity tests predict the broad criterion of organizationally disruptive behaviors better than they predict the narrower criterion of employee theft alone. Our results also indicated substantial evidence for the construct validity of integrity tests. Perhaps the most important conclusion of this research is that despite the influence of moderators, integrity test validities are positive across situations and settings. Personnel selection; Personnel integrity and Reliability; Personnel security; Meta-analysis; Selection validity.