Gender and IQ: Are Men or Women Smarter?

There is no meaningful difference in average general intelligence between men and women. Decades of large-sample research place the gap at essentially zero. Real, replicated sex differences exist — but they sit on specific abilities and on the spread of scores, not on overall IQ.

The evidence for that flat average is strong. Reviewing 46 meta-analyses spanning abilities, communication and personality, Janet Hyde (2005) found that 78 percent of measured gender differences were small or close to zero — the basis of what she named the gender-similarities hypothesis. A 2022 meta-analysis went further, concluding that even the small mean differences once seen in children were largely artifacts of older test versions, with modern batteries showing no statistically significant gap in general ability. As an interpretation rather than a finding of my own: this is one of the more settled questions in intelligence research, and it is also one of the most persistently misreported.

Where it gets genuinely interesting is below the surface. Average IQ is a blend, and balanced tests are built so that male- and female-favouring abilities cancel out. Pull the blend apart and patterns appear.

The average: a genuine tie

Start with the number people actually want: overall IQ. Across well-sampled modern studies, the average general intelligence of men and women is effectively identical. When a difference does surface, it is usually a point or two and it does not survive proper analysis of the underlying general factor, g.

A representative example: a standardisation study of the WAIS-III in over 1,300 adults observed a male edge of about 3.6 IQ points in raw "ability in general" — but the authors showed this came from specific group factors and test composition, not from g itself, where no sex difference held (Colom et al., 2002). That distinction matters enormously. The IQ score you read off a test is a sum of many abilities; the deeper general factor that drives the famous correlations between IQ and life outcomes shows no sex gap. Confuse the two and you can manufacture a difference that isn't really there.

So when someone confidently tells you one sex is smarter, the first question worth asking is: smarter on what, exactly? Overall intelligence is not where the action is. To understand why a single IQ number can hide so much, it helps to know how IQ tests are actually scored and combined from their component parts.

Where specific abilities diverge

Here the patterns are real, if modest. On average, women outperform men on verbal fluency — the rapid generation of words — with an effect size around d = 0.33 (Hyde, 2005). Women also tend to lead on processing speed and certain memory tasks. Men, on average, lead on some visual-spatial tasks.

One spatial task stands out as the largest reliable cognitive sex difference anywhere in the literature: three-dimensional mental rotation. The male advantage there runs to roughly d = 0.8 to 1.0, which is genuinely large by the standards of this field (Hyde, 2005; Maeda & Yoon, 2013). If you have ever seen those puzzles asking which rotated block matches the target, that is the task. The gap is robust, it has held steady across decades while other spatial gaps narrowed, and it is the one finding that resists the "it's all near-zero" summary. The broader picture of how these abilities are categorised sits in our overview of spatial intelligence and what it measures.

Notice what these differences do at the level of overall IQ: they pull in opposite directions and roughly cancel. A verbal-fluency edge for one sex and a mental-rotation edge for the other net out close to zero when you sum across a full battery. That cancellation is not an accident, as the next section explains. The same logic underlies the long-studied split between verbal and nonverbal IQ, where group profiles can differ even when totals match.

To see where your own profile sits across verbal, numerical and spatial domains rather than as one flattened number, the Free IQ Test gives a quick baseline across core cognitive areas in about fifteen minutes.

The variability debate

This is the part that gets misused most, so I'll be careful with it. The greater-male-variability hypothesis proposes that even with equal averages, male scores spread out more — producing slightly more men at both the highest and lowest extremes. A frequently cited meta-analysis of 242 studies covering over 1.2 million people found a variance ratio of about 1.08, meaning roughly 8 percent more variance among males (Lindberg et al., 2010) — which those authors explicitly described as not meaningfully different from equal.

And the evidence cuts both ways. Some large datasets find the greater male spread; others find equal variability, and at least one meta-analysis found greater female variability on the standard Raven's Progressive Matrices. A 2022 analysis reported no sex difference in the variability of g at all. So the honest status is: a small male-variance tendency appears in some samples, it is far from universal, and its size is modest where it shows up.

Why does this matter? Because at the far tail of a distribution, even a tiny variance gap multiplies. If men are slightly more spread out, you would expect somewhat more men several standard deviations above the mean — the territory discussed in our look at what counts as a high IQ and how rare it is. But a tail effect is exactly the kind of thing that gets stretched into claims it cannot support. A small, contested difference in spread says nothing about any individual, and nothing about averages.

"The variability question is where careful science most often gets hijacked. A variance ratio of 1.08 is a real, interesting statistic about distributions — and it is routinely inflated into sweeping claims about who belongs at the top of a field. The data do not carry that weight."

— Adam Imran, MS Clinical Psychology · DesperateMinds

Why tests show no gap

There is a design reason the headline number comes out flat, and it is worth understanding because it sounds like circular reasoning until you look closely. IQ test makers, going back decades, deliberately balanced their batteries so that items favouring one sex were offset by items favouring the other. Where a subtest produced a large sex gap, it was often revised or weighted to keep the overall scale fair.

Does that mean the "no difference" result is just baked in by the test designers — a fix rather than a finding? Partly, on overall IQ. But it does not erase the real subtest differences, which are measured openly and not hidden. And the deeper result stands independently: studies using the general factor g, which is extracted from the correlations among subtests rather than their sum, keep finding no sex difference regardless of how the battery is balanced. The flat average is partly a design choice; the flat g is not.

Biology, culture, or both?

For the differences that are real — chiefly mental rotation — the cause is contested, and anyone claiming certainty in either direction is overselling. Two facts pull against a purely biological story. First, spatial gaps shrink substantially with training; Hyde noted programs that improved students' spatial performance and, in one case, lifted women's retention in an engineering major from 47 to 77 percent. Second, the size of cognitive sex differences varies by country and shifts over time, which a hardwired trait should not do so readily.

Against a purely cultural story, some differences appear early and a few spatial gaps have proven stubborn. A twin study testing the idea that prenatal testosterone transfer explains female spatial ability found no such effect across 14 spatial measures, which complicates the simplest hormonal accounts (Vuoksimaa et al., 2010). The reasonable position — and the one most researchers hold — is that these patterns reflect a tangle of biology and environment whose proportions we cannot yet cleanly separate. The category most shaped by experience is the knowledge-and-skill side of cognition, the crystallized intelligence that training and exposure build directly.

I'll add one honest qualification to the optimistic training story. That mental rotation responds to practice does not prove the original gap was caused by unequal practice — it only shows the gap is malleable. Malleable and environmental are not the same thing, and tidy summaries often blur them.

The overlap that gets ignored

Every number in this article describes a group average, and group averages are nearly useless for predicting an individual. Even the large mental-rotation difference means the male and female distributions overlap enormously — plenty of women outperform most men on it, and plenty of men sit below the female average.

That is the single most important thing to carry away, and it is the thing internet arguments forget first.

A difference of d = 0.8 still leaves the two distributions sharing the large majority of their range. For the near-zero differences — overall IQ, reading comprehension, most everything — the overlap is almost total. Knowing someone's sex tells you essentially nothing about their intelligence. The averages are a fact about populations; they are not a fact about the person in front of you.

What this research can't tell you

A few honest boundaries. Much of the foundational variability data came from samples that skewed toward particular countries and eras, and some classic studies drew heavily on male-only military datasets, which limits how far they generalise. Cross-cultural coverage has improved but remains uneven.

This research also says nothing about cause at the individual level, nothing about potential, and nothing about worth. It measures average performance on particular tasks at particular moments, under particular conditions. People sometimes reach for these findings to justify conclusions about who should do which jobs or enter which fields — a leap the data flatly cannot support, given the overlap and the malleability we have already seen. If you take one thing from the science, let it be how small and how conditional these differences are, not how large.

Conclusion

On the question that draws people to this topic — is one sex smarter? — the answer is a clean no. Average general intelligence is, as best we can measure it, equal. Beneath that average lie real but modest differences in specific abilities that pull in opposite directions and a small, genuinely disputed difference in score spread, all of it swamped by the overlap between the two groups. The science here is more careful and more boring than the culture war built on top of it, and that gap between what the data say and what people claim they say is the real story worth telling.

Frequently Asked Questions

References