| Abstract: | This paper reviews and critiques research pertaining to the three most commonly used honesty tests. Honesty tests, sometimes called integrity tests, purportedly predict theft by employees. Secondary analyses were performed to remove the effect of faking good from validity coefficients with various self-report criteria, and to supplement other analyses. The honesty tests were found to have: (a) virtually no foundation in personality or attitude theory, (b) a corrected average correlation of 0.08 with objective indices of theft (95′/0 confidence interval: 0.03 < r < 0.14), (c) a Taylor-Russell utility of approximately 1% above base rate of success, and (d) a false positive rate of 0.44 if only nontrivial thefts are considered. The honesty tests reviewed are of such marginal validity (less than 1% of the criterion variance accounted for) that their continued use in preemployment settings is seriously questioned. The comparability of these review results with those of other reviewers is discussed along with the social and legal implications of honesty testing. |
