Does IQ Actually Matter? What It Predicts and What It Doesn’t

Most writing on this question picks a side. Either IQ is the single best predictor of everything that matters, or it is a discredited relic that measures only how good you are at IQ tests. The evidence supports neither, and the accurate picture is more interesting than either camp allows.

Two findings frame it. The correlation between cognitive ability and job performance was revised downward in 2022 after a statistical flaw was found in decades of meta-analyses. And the association between childhood IQ and how long you live is large, robust, and almost entirely absent from popular discussion.

What IQ predicts, at a glance

The third column is the one to keep your eye on. Even where the relationship is real, most of what happens to people is happening for other reasons.

Work: smaller than the textbooks said

For roughly twenty-five years the standard citation was Schmidt and Hunter (1998), which reported a correlation of about 0.51 between general cognitive ability and job performance and established cognitive tests as the best single predictor available. That figure went into textbooks, training courses and hiring practice everywhere.

In 2022, Sackett, Zhang, Berry and Lievens re-examined how those meta-analyses had corrected for range restriction, the statistical adjustment used because studies typically sample people already selected into a job. They found the correction had been applied in a way that systematically overstated validity across a wide range of selection tools. Their revised estimates:

• Structured interviews: about 0.42
• Cognitive ability tests: about 0.31

Cognitive ability is no longer top of that table. A structured interview, which assesses how someone actually reasons about the work in front of them, now predicts performance better than an IQ-type test does.

A correlation of 0.31 is real and worth taking seriously. It is also small enough that over ninety percent of the variation in how people perform at work comes from somewhere else. Anyone telling you that cognitive ability is the dominant factor in career outcomes is working from a figure that was corrected three years ago.

Health: larger than almost anyone knows

Here the surprise runs the other way, and it comes from an entire field most people have never heard of: cognitive epidemiology, which studies the relationship between intelligence test scores and later illness and death.

The headline result is striking. A meta-analysis by Calvin and colleagues (2011) found that a one standard deviation advantage in early-life cognitive ability, which is fifteen IQ points, was associated with roughly a 24% reduction in the risk of death during follow-ups running up to 69 years.

The association appears for all-cause mortality and for specific causes including cardiovascular disease, accidents and suicide. Large Scottish studies by Deary, Batty and colleagues, which followed almost an entire population tested as children in 1932 and 1947, found the relationship held after controlling for childhood socioeconomic circumstances. It is not simply a proxy for having grown up comfortable.

Why it exists is less settled than that it exists. The leading explanation is fairly mundane: managing your health is itself a cognitively demanding job. Understanding risk, following complex treatment regimens, navigating a healthcare system, noticing when something is wrong and avoiding accidents all draw on the same reasoning and problem-solving the test measures. Competing explanations include cognitive ability serving as a marker of general bodily system integrity, and the ordinary pathway through education and occupation into safer work and better conditions.

Whatever the mechanism, this is the outcome where the case for IQ mattering is strongest, and it is almost never the one people are thinking about when they ask the question.

How to read these numbers

Correlations get quoted constantly and understood rarely, and the arithmetic matters for this question specifically.

A rough rule: square the correlation to get the share of variation explained. So a correlation of 0.31 accounts for about 9.6% of the variation. A correlation of 0.5 accounts for 25%. Even a strikingly strong correlation of 0.7 would leave half the variation unaccounted for.

This cuts against both of the popular positions at once. It means IQ is not destiny, because the unexplained share is always the larger one. It also means IQ is not nothing, because a factor accounting for ten percent of the variation in something as consequential as job performance is genuinely substantial as behavioural science goes. Very few things about a person predict that well.

The other thing to hold onto is that these are population statistics. They describe how outcomes are distributed across large groups. They cannot tell you what will happen to any individual, and the spread around every one of these relationships is wide enough to contain almost any personal story you can imagine.

What IQ does not predict

Worth being specific rather than gesturing at “other kinds of intelligence”.

• Emotional and social skill. IQ and measures of emotional intelligence correlate only weakly, which means knowing one tells you very little about the other. Covered in our comparison of IQ and EQ.
• Happiness or life satisfaction. There is no substantial relationship. Whatever produces a contented life, this is not the measure of it.
• Conscientiousness and follow-through. Effectively independent of cognitive ability, and at least as important for most outcomes that require sustained work.
• Creativity. The relationship is weaker than assumed and the popular claim that it disappears above IQ 120 is not well-supported either, as set out in our guide to an IQ of 120.
• Wisdom, judgement and character. Not measured, not claimed to be measured, and routinely assumed to come along with a high score. They do not.

Does it stop mattering above a certain level?

This is the most repeated claim in the subject and the evidence is mixed in an interesting way.

For creativity, the classic threshold hypothesis put the cut-off at IQ 120. The most recent and most thorough re-analysis, by Weiss and colleagues in 2020, found no evidence for a threshold under any of four analytical approaches and recommended abandoning the idea.

For achievement, the long-running work following people identified as top-1% reasoners in adolescence finds that differences within that already-selected group still predict doctorates, patents, publications and income decades later. Returns diminish as you go up. They do not switch off.

So the defensible version is that ability keeps mattering at the top, while mattering less per point than it did lower down. Both threshold arguments are examined in the 120 and 130 guides.

What matters more

If ninety percent of the variation in job performance is not cognitive ability, the obvious question is what it is. The honest answer is that no single factor dominates, but several are well evidenced:

• How someone reasons about the actual work. Structured interviews now out-predict cognitive tests, which suggests that domain-situated judgement carries information a general test misses.
• Conscientiousness. Independent of ability, and the personality trait most consistently linked to performance across roles.
• Accumulated domain expertise. Compounds over years in a way that general ability does not, and is the main thing separating people who cleared the same cognitive bar a decade ago.
• Health, timing, network and circumstance. Unglamorous, largely uncontrollable, and collectively enormous.

Frequently asked questions

The bottom line

IQ matters, and not in the places or to the degree most people assume. Its link with job performance is about 0.31 on current best estimates, below structured interviews, leaving over ninety percent of performance variation to other things. Its link with health and longevity is substantial, durable across decades, and survives controlling for social class.

The sensible way to hold a score is as one input among many, informative about a narrow band of reasoning ability and silent on most of what determines how a life goes. It is not nothing, which is what the sceptics get wrong. It is not destiny, which is what the enthusiasts get wrong. It is a moderately useful number with a large amount of unexplained variance sitting on top of it, and that remainder is mostly where you live.