Does IQ Predict Success? The Numbers Changed in 2022
- Written by
- IQCognify Editorial Team
- Reviewed for accuracy
- IQCognify Research Review Process
- Last updated
- Reading time
- 6 min
Quick answer
For twenty-five years the standard answer was that general mental ability predicts job performance better than anything else, with a validity of about .5. In 2022 Sackett and colleagues showed that figure had been inflated by a flawed statistical correction applied across decades of meta-analyses. The corrected estimate is .31, and cognitive ability is no longer the strongest predictor of job performance. It now sits behind structured interviews and biodata.
The number everyone learned
In 1998 Frank Schmidt and John Hunter published a review in Psychological Bulletin that became one of the most cited papers in applied psychology. Its headline finding was that general mental ability predicted job performance with a validity of about .51, higher than any other single selection method, and that adding other predictors improved on it only marginally.
That result shaped a generation of hiring practice, textbooks and popular writing about intelligence. If you have ever read that IQ is the best predictor of job performance, you have read Schmidt and Hunter, whether or not you knew it.
What Sackett and colleagues found
Validity coefficients in personnel selection are not raw correlations. They are corrected — for the unreliability of the performance measure, and for range restriction, the fact that a study of people already hired excludes everyone who was rejected and therefore artificially compresses the spread of ability.
Range restriction is real and correcting for it is legitimate. Sackett, Zhang, Berry and Lievens examined how the correction had actually been implemented across the meta-analytic literature and identified a systematic error: the artifact distributions used to make the correction assumed a degree of restriction that the primary studies did not have. Applied at scale, across decades, this inflated the reported validity of a wide range of predictors — and cognitive ability most of all.
| Selection method | Prior meta-analytic value | Sackett et al. (2022) |
|---|---|---|
| Structured interviews | .48 | .42 |
| Biodata | .32 | .38 |
| General mental ability tests | .52 | .31 |
| Integrity tests | .42 | .31 |
| Conscientiousness tests | .22 | .19 |
The conclusion, in the authors' own framing
The validity of cognitive ability tests is reduced from .52 to .31, and cognitive ability tests are no longer the strongest predictor of job performance — instead they lag behind structured interviews and biodata.
A further update restricting the analysis to validity studies conducted in the twenty-first century produced a lower figure still: an operational validity of about .22. The older estimates rested mostly on twentieth-century data, gathered when work looked different.
What a validity of .31 actually means
Correlations are easy to misread in both directions, so it is worth being concrete.
Squaring the coefficient gives the proportion of variance accounted for. A validity of .31 means cognitive ability accounts for roughly 9.6% of the variation in job performance. Around ninety per cent of why some people perform better than others at work is something else — motivation, experience, health, management, opportunity, temperament, luck, and the fit between a person and a role.
But .31 is not nothing
By the standards of psychology, a validity of .31 is a substantial effect, and across a thousand hires it produces real gains. The mistake is to read a population-level predictor as an individual-level forecast. It tells you how to bet across many people. It tells you very little about any one of them.
Both errors are common. One camp treats a .31 correlation as proof that IQ determines your career. The other treats it as proof that IQ is meaningless. Neither follows.
Beyond work: what else does IQ predict?
Job performance is one outcome among many, and IQ relates to them with very different strengths. Strenze's meta-analysis of longitudinal studies gives the broad picture.
| Outcome | Approximate correlation | Variance accounted for |
|---|---|---|
| Educational attainment | ~.56 | ~31% |
| Occupational status | ~.43 | ~18% |
| Income | ~.21 | ~4% |
The pattern is orderly and revealing. Intelligence predicts most strongly the outcome that most resembles an intelligence test — how far you go in school. It predicts occupational status less well, and income least of all. By the time you reach earnings, more than 95% of the variation is attributable to something other than measured intelligence.
Intelligence also correlates with health and longevity, a finding that has been replicated in large cohorts and is not well explained. Whether it reflects better health decisions, better jobs, or a common underlying factor in system integrity remains open.
The ceiling: IQ does not predict exceptional achievement
Whatever IQ does across the ordinary range, it does not identify the people who will do extraordinary work.
Lewis Terman followed more than 1,500 children selected for an IQ of roughly 140 or above across their entire lives. They did well. None produced work of historic significance. Two boys screened during recruitment, William Shockley and Luis Alvarez, failed to reach the threshold and later won Nobel Prizes in physics.
The anecdote is weaker than it looks
It is usually cited as proof that IQ fails to predict achievement. Warne (2020) showed the arithmetic does not support that reading: Nobel Prizes are rare enough, and Terman's cutoff high enough, that a sample of his size would be expected to contain no laureates whether or not IQ predicts anything. The rejections are a fact; the inference drawn from them is not.
The defensible statement is narrower. High ability raises the probability of exceptional work without coming close to guaranteeing it, and the traits that convert ability into achievement — persistence, taste in problems, tolerance for being wrong in public — are not what an intelligence test samples.
Einstein's IQ — what a number nobody measured is doing at the centre of this conversation.Myths and facts
| Myth | Fact |
|---|---|
| IQ is the best predictor of job performance | Not since 2022. Sackett and colleagues put cognitive ability at .31, behind structured interviews (.42) and biodata (.38). |
| The validity of cognitive ability is around .5 | That figure came from meta-analyses that systematically overcorrected for range restriction. Corrected, it is .31 — and about .22 in 21st-century studies. |
| A .31 correlation means IQ explains a third of job performance | Square it. A validity of .31 accounts for under 10% of the variance in performance. |
| IQ strongly predicts how much you will earn | The correlation with income is about .21, accounting for roughly 4% of the variance. |
| High IQ predicts world-changing achievement | Terman's high-IQ cohort produced no laureates; two boys screened out did. Warne (2020) shows base rates alone explain that, so the anecdote proves less than it is asked to. |
| The 2022 correction shows IQ is meaningless | A validity of .31 is a substantial effect in psychology. It is one strong predictor among several, not the predictor. |
Frequently asked questions
Does IQ predict job performance?+
Yes, but less well than was believed until recently. Sackett, Zhang, Berry and Lievens (2022) found that decades of meta-analyses had systematically overcorrected for range restriction, inflating the validity of cognitive ability tests. The corrected operational validity is .31, down from a prior figure of about .52.
Is IQ still the best predictor of job performance?+
No. In the corrected estimates, structured interviews predict job performance at .42 and biodata at .38, both above cognitive ability's .31. The authors state explicitly that cognitive ability tests are no longer the strongest predictor and now lag behind those two methods.
Why was the old .51 figure wrong?+
Validity coefficients are corrected for range restriction, because studies of people already hired exclude everyone who was rejected and therefore compress the observed spread of ability. Sackett and colleagues found the artifact distributions used to make that correction assumed more restriction than the primary studies actually contained, inflating the results across the literature.
What does a validity of .31 mean in practice?+
Squaring it gives the variance explained: cognitive ability accounts for roughly 9.6% of the variation in job performance. Around ninety per cent of why some people perform better than others at work is attributable to something else. That is still a substantial effect by the standards of psychology, but it is a population-level predictor, not an individual forecast.
How strongly does IQ predict income?+
Weakly. Strenze's meta-analysis of longitudinal research puts the correlation between intelligence and income at about .21, accounting for roughly 4% of the variance. The correlation with educational attainment is much stronger at around .56, and occupational status falls between the two at about .43.
Do high-IQ people achieve more?+
On average and across large groups, yes, with the strength of the relationship depending heavily on the outcome. For exceptional achievement the evidence is weak: Terman's cohort of more than 1,500 children with IQs around 140 or above produced no work of historic significance, while two boys who failed to qualify later won Nobel Prizes. Warne (2020) showed base rates alone would predict that result.
Has the estimate fallen further since 2022?+
Yes. A subsequent update restricting the analysis to validity studies conducted in the twenty-first century produced an operational validity of about .22 for general mental ability tests, compared with .31 from the mostly twentieth-century data used in the 2022 reanalysis.
Does this mean IQ tests are worthless for hiring?+
No. A validity of .31 remains a meaningful effect and, aggregated across many hires, produces real gains. What has changed is its rank: cognitive ability is one strong predictor among several rather than the dominant one, and combining methods predicts better than any single method alone.
Sources
This guide draws on standard psychometric references and peer-reviewed research:
- 1.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.
- 2.Sackett, P. R., Demeke, S., et al. (2023). Updated meta-analysis of general mental ability test validity using 21st-century validity studies, as reported in the updated personnel-selection meta-analytic matrix.
- 3.Schmidt, F. L., & Hunter, J. E. (1998). “The validity and utility of selection methods in personnel psychology.” Psychological Bulletin, 124(2), 262–274.
- 4.Strenze, T. (2007). “Intelligence and socioeconomic success: A meta-analytic review of longitudinal research.” Intelligence, 35(5), 401–426.
- 5.Warne, R. T. (2020). “Low base rates and a high IQ selection threshold prevented Terman from identifying future Nobelists.” Intelligence, 82, 101488.
- 6.Ritchie, S. J., & Tucker-Drob, E. M. (2018). “How much does education improve intelligence? A meta-analysis.” Psychological Science, 29(8), 1358–1369.
- 7.Neisser, U., et al. (1996). “Intelligence: Knowns and Unknowns.” American Psychologist, 51(2). APA.
- 8.Deary, I. J. (2020). Intelligence: A Very Short Introduction (2nd ed.). Oxford University Press.
- 9.American Psychological Association (APA)
Sources are provided for further reading. Organization links point to official sites; academic works are cited in full. See our research standards and editorial team.
Find out your IQ
Take the free IQ test and get your score, percentile, and a full cognitive breakdown in about 12 minutes.
Start Free Test