Predictive validity of integrity tests for workplace deviance across industries and countries in the past 50 years: A meta-analytic review

Abstract

The current meta-analysis provides a comprehensive and updated review on integrity-testing findings across industries and countries in the past 50 years (k = 150, N = 67,016). Integrity tests were coded into the types of overt tests, covert tests, biodata, organizational measures, value/moral reasoning/situational judgment tests, integrity-related cognitive ability tests, and novel measures. The criterion measures of workplace deviance included CWBs, unethical pro-organizational behaviors, and other workplace deviant behaviors. For the information source, both computer and manual searches were performed to locate relevant published and unpublished papers. A variety of sources were examined to avoid publication bias, and publication bias analyses were conducted to uphold the methodological rigor. Results indicated that all the integrity tests analyzed were significant in predicting workplace deviance, with an overall mean validity estimate corrected for indirect range restriction and measurement error as .43 (95% CI [.32; .52]; p < .001). Among the tests, the value-oriented tests and cognitive ability tests indicated relatively large validity estimates of .60 (95% CI [.41; .75]; p < .001) and .65 (95% CI [.53; .74]; p < .001), respectively. The relationship between integrity tests and workplace deviance was found to be significantly moderated by the type of integrity test, industry, country, and criterion source. The effect size of integrity tests was largest in predicting deviance in the military and law enforcement sector, and relatively large in the work samples of Canada, Germany, Israel, Romania, and the United States. However, the moderating effects of the nature of deviance, validation sample, validation strategy, publication status, medium of test, and gender, were nonsignificant. Compared with previous reviews, our study was unique in its cross-cultural direction, which included primary studies of integrity testing in countries with different languages (e.g., publications in Chinese) and associated cultural variations. New insights and comparisons with previous meta-analytic findings were discussed.