EV-0001
Integrity tests predict job performance and counterproductive behavior (the landmark, now contested)
Ones, D. S., Viswesvaran, C., & Schmidt, F. L. (1993). Journal of Applied Psychology, 78(4), 679–703 · Source
What this entitles us to say
Large-scale meta-analytic evidence that measures of integrity predict both supervisor-rated job performance and counterproductive behavior at work — though the size of that prediction is disputed (see EV-0002).
What it found
A meta-analysis of 665 validity coefficients drawn from around 576,460 data points, reporting that integrity tests showed generalizable predictive validity for both overall job performance and counterproductive workplace behavior — disciplinary problems, absenteeism, rule-breaking, theft.
Published estimates in the literature citing this work vary with the corrections applied: roughly .34–.41 for job performance and around .47 for counterproductive behavior.
Why it matters here
It is the most-cited empirical support for the book's central premise — that character is not a soft adjunct to performance but a predictor of it. If integrity can be measured well enough to forecast whether someone will steal or quit, then "character drives outcomes" is an empirical claim, not a sentiment.
The limits
Do not cite this alone. Its headline figures were directly challenged in 2012 by a meta-analysis using stricter inclusion criteria, which found validity roughly a third to a half of these estimates (EV-0002). The central criticism is that this analysis drew heavily on unpublished studies authored by the companies selling the tests. Reported figures also vary by correction method (~.34 to ~.41 for job performance, ~.47 for counterproductive behavior), so quote a range, never a single decimal. Finally: this literature is about pre-employment selection tests, which is not what CDL does — see "How to say it honestly."
What we do not say
"Studies prove integrity predicts performance." It is contested (EV-0002), the figures vary by method, and — most importantly — this research is about hiring tests used to screen applicants. CDL is not a selection instrument and must never borrow the credibility of one. We are a development program, and our own assessment is explicitly not a validated selection instrument. Citing this to imply otherwise would be the exact dishonesty the product is built against.