What is Predictive validity?
Predictive validity is the correlation between a score measured before a hiring decision and an outcome measured afterwards, such as supervisor-rated performance. It is the form of validity evidence that matters most in selection, because it is the only one whose study design matches the way the test is actually used.
The numbers were revised downwards in 2022, and the revision is the news
For twenty-five years the reference point in personnel selection was Schmidt and Hunter's 1998 meta-analysis, which reported general mental ability as the strongest single predictor of job performance at around 0.51. Those figures are still quoted in vendor material today.
Sackett, Zhang, Berry and Lievens re-analysed the evidence in 2022 and found that corrections for range restriction had been applied systematically too aggressively — estimates derived from predictive designs had been used to correct concurrent studies, where applicants had not been selected on the measure in the first place. Correcting that produced a materially different ordering: structured interviews at 0.42, job knowledge tests at 0.40, empirically keyed biodata at 0.38, work samples at 0.33, and cognitive ability at 0.31.
Two things follow. Cognitive ability is a good predictor rather than a uniquely dominant one, and structured interviewing is not a soft complement to testing but a comparable instrument in its own right. Any vendor still quoting 0.51 in 2026 is quoting a superseded figure.
What a correlation of 0.35 means in practice
It means the instrument shifts the odds, not that it identifies the right person. Squared, it accounts for roughly an eighth of the variance in the outcome, which sounds dismal and is nevertheless commercially significant at volume, because a small shift applied across thousands of decisions moves a great many people.
It also means no single instrument should carry a decision alone. Combinations of predictors that measure different things — a structured interview plus a work sample, rather than two cognitive tests — add more than either does by itself.
There has been a second revision since, and almost nobody has it
The 2022 figures above are already not the newest ones. Sackett and colleagues published a follow-up in 2023 collecting estimates that had appeared since, and it revises cognitive ability downward again — to 0.23, on the basis of a meta-analysis by Griebe and colleagues of 113 twenty-first-century studies reporting a mean observed validity of 0.16. The authors state the consequence plainly: using that value drops cognitive ability from fifth to twelfth in the ranking of the predictors they examined. Assessment centres move the other way in the same table, from 0.29 to 0.33.
The practical difference is smaller than the reordering suggests and it is not nothing. At a base rate of 50% and a selection ratio of one hire in ten, a validity of 0.31 produces a success rate of 71.5% among those hired and a validity of 0.23 produces 66.0% — five and a half fewer good hires per hundred.
Read the second column, not just the first
Every figure in a meta-analysis is the mean of a distribution, and the spread is published beside it. Structured interviews carry a standard deviation of 0.19 around their 0.42, integrity tests 0.20 around their 0.31, cognitive ability 0.14 around its 0.31. Two instruments with the same headline number can differ substantially in how dependably that number transfers to your setting.
The discipline that follows is cheap: model any selection decision twice, once at the published mean and once a standard deviation below it, and take the decision that survives both. A business case that only works at the mean is a business case that only works half the time.
Not the same as concurrent validity
Concurrent studies test people who already hold the job and correlate scores with current performance. Predictive studies test applicants and wait. Concurrent designs are cheaper and their samples have already been through selection, which narrows the range and distorts the coefficient.
Criterion validitySources
- Sackett, P. R., Zhang, C., Berry, C. M. & Lievens, F. (2022). Revisiting meta-analytic estimates of validity in personnel selection: Addressing systematic overcorrection for restriction of range. Journal of Applied Psychology, 107(11), 2040–2068
- Sackett, P. R., Zhang, C., Berry, C. M. & Lievens, F. (2023). Revisiting the design of selection systems in light of new findings regarding the validity of widely used predictors. Industrial and Organizational Psychology, 16, 283–300 — the source of the 0.23 re-revision (open access)
- SIOP — Is Cognitive Ability the Best Predictor of Job Performance? (summary of the revised estimates)
Read next
Related terms
More terms beginning with P
Check a selection process against the four-fifths rule
Free, no signup, computed in your browser — with the remedy, not just the verdict.
Last reviewed