Males in many species, and humans in particular, tend to show more individual variation than females across a wide range of measurable traits. That, in a nutshell, is the male variability hypothesis. The idea has been documented in domains as diverse as IQ scores, brain anatomy, energy expenditure, and economic decision-making, though the pattern is not universal and some traits show the reverse. What makes the hypothesis interesting, and frequently misunderstood, is that greater variability says nothing about which sex is “better” at anything. It means the male distribution is wider: more individuals clustered at both extremes, with fewer near the average.
Where the Evidence Is Most Consistent
Cognitive test data have been examined for sex differences in variability for over a century, and the finding that keeps turning up is that males are more spread out in several ability domains. A review of the research literature found males consistently more variable than females in quantitative reasoning, spatial visualization, spelling, and general knowledge, while also noting that variability differences and differences in averages need to be considered together to draw accurate conclusions.1Review of Educational Research. Sex Differences in Variability in Intellectual Abilities: A New Look at an Old Controversy A separate analysis of general intelligence found that even above average ability levels, males remained more variable, a pattern the authors linked to a mixture of normal variation and genetic or environmental conditions affecting cognitive development.2PubMed. Sex Differences in Variability in General Intelligence: A New Look at the Old Question
More recently, a large meta-analysis using data from 75 studies of children tested on the Wechsler Intelligence Scales confirmed that males were more variable in several cognitive domains, including visual processing and crystallized intelligence. But it also found that females were more variable in processing speed, and that no sex difference in variability existed for short-term verbal memory. The researchers reported a striking regularity: whichever sex had a higher average on a given task also tended to be the more variable sex on that task.3Personality and Individual Differences. Sex differences in cognition: A meta-analysis of variance ratios in the Wechsler Intelligence Scales for Children That finding complicates the picture. Greater male variability is not a blanket rule across all cognitive abilities; it shows up most reliably in domains where males also score higher on average.
Brain structure tells a parallel story. The ENIGMA Consortium’s mega-analysis of MRI scans from over 16,000 healthy people aged one to ninety found that males had greater variability than females in all subcortical brain volumes, all cortical surface area measures, and about 60% of cortical thickness measures. For many of these structures, the pattern held steady across the entire lifespan.4PubMed Central. Greater male than female variability in regional brain structure across the lifespan In other words, one-year-old boys and seventy-year-old men both showed more person-to-person spread in brain anatomy than their female counterparts at the same age.
Beyond the Brain
Greater male variability is not limited to cognition and anatomy. An analysis of a large database of adult human energy expenditure, covering roughly 1,500 males and 3,100 females, found substantially more variation in total, activity-related, and basal energy expenditure among males, even after matching people for age, height, and body composition.5Personality and Individual Differences. Variability in energy expenditure is much greater in males than females Two men of the same size and age can burn surprisingly different amounts of energy in a day; two women of the same size and age tend to be closer together. The reasons are not entirely clear, but the finding suggests that metabolic variability is part of a broader biological pattern.
Behavioral and economic choices also fit. A meta-analysis pooling data from over 50,000 participants in experimental economics studies found converging evidence of greater male variability in time preferences, risk preferences, and social preferences. In some cases the sexes also differed in their averages; in others, average preferences were similar but males were more spread out. The authors argued that theories of gender differences in economic behavior are incomplete if they only look at average differences and ignore the wider spread within men.6PubMed Central. Converging evidence for greater male variability in time, risk, and social preferences
Hearing thresholds offer yet another example. A study examining hearing acuity found clear evidence that males varied more than females and that related hearing measures were more correlated within males than within females. The researchers attributed this to the mosaic pattern of X-chromosome inactivation in females, a mechanism discussed below.7PubMed Central. Sex differences in number of X chromosomes and X-chromosome inactivation in females promote greater variability in hearing among males
Why a Wider Spread Matters at the Tails
One reason the male variability hypothesis attracts so much attention is that even modest differences in how spread out two groups are can translate into dramatic differences at the extremes. If you look at the middle of a bell curve, two groups with the same average but slightly different spreads look almost identical. But move out to the top or bottom one percent, and the group with the wider spread will be heavily overrepresented. This is not an empirical controversy; it is a mathematical certainty. A formal proof showed that if one population has greater variance than another, it will dominate both the upper and lower tails, regardless of where the two averages sit.8Stats. Extreme Tail Ratios and Overrepresentation among Subpopulations with Normal Distributions
In practical terms, this means that small differences in variability can produce large imbalances when you apply any kind of threshold, whether that is a cutoff for gifted education, a diagnostic criterion for intellectual disability, or any screening process that selects from the extreme ends. A review of this issue emphasized that even small group-mean differences, or variance differences alone without any mean difference, can generate marked imbalances in group representation in the tails.9PubMed. On the importance of tail ratios for psychological science This is why the male variability hypothesis has been invoked in debates about sex ratios at the top of math and science achievement: if boys are even slightly more spread out, you would expect to see more boys among both the highest and lowest scorers, even if the average boy and average girl score the same.
Italian data on math and reading performance in school-aged children illustrate this concretely. Boys were more variable in math at every grade tested, with their overrepresentation at the top growing as children aged. In reading, the picture was different and somewhat grimmer for boys: girls dominated the top five percent at every grade, while boys became increasingly overrepresented in the bottom five percent as they progressed through school. By eighth grade, boys were roughly eight times more likely than girls to be among the weakest readers.10Personality and Individual Differences. Sex differences in variability: Evidence from math and reading assessment in Italy Reading is a domain where greater male variability primarily means more boys struggling, not more boys excelling.
Biological Mechanisms
Why would males be more variable? Several biological explanations have been proposed, and they are not mutually exclusive.
The most frequently cited mechanism involves the X chromosome. Females carry two copies of the X chromosome, and in each cell one is largely shut off through a process called X-inactivation. Which copy gets silenced is essentially random, creating a patchwork: some cells express the mother’s X, others express the father’s. This mosaicism acts as a buffer. If one copy of a gene on the X chromosome carries an unusual variant, roughly half a female’s cells will express the other, potentially more typical copy. Males, with only one X chromosome, get no such buffering. Every cell expresses the same single copy, so any unusual variant has its full effect. This has been proposed as one reason for greater male variability in X-linked traits, and the hearing study mentioned earlier found evidence consistent with this mechanism.11PubMed Central. Sex differences in number of X chromosomes and X-chromosome inactivation in females promote greater variability in hearing among males
Beyond the sex chromosomes, there appear to be sex differences in how genes on the non-sex chromosomes are regulated. A study of DNA methylation patterns across the genome identified hundreds of sites on ordinary (autosomal) chromosomes where the chemical marks that control gene activity differ between males and females. About three-quarters of these sites showed higher methylation in females.12PubMed Central. Characterising sex differences of autosomal DNA methylation in whole blood using the Illumina EPIC array While this does not directly prove that these methylation differences cause differences in variability, it shows that the molecular machinery governing gene expression is calibrated differently in male and female cells, potentially contributing to different variability profiles even for traits not linked to the X chromosome.
An evolutionary framing has also been proposed. A theoretical model advanced the idea that sexual selection, where one sex competes more intensely for mates, could drive the evolution of greater variability in the more competitive sex. The reasoning is that when reproductive success is more “winner-take-all,” nature benefits from producing more variation in the competing sex: some individuals will be spectacularly fit, others spectacularly unfit, but the average outcome for the lineage can improve because the winners reproduce prolifically.13arXiv. An Evolutionary Theory for the Variability Hypothesis In most mammals, males face stronger sexual selection, which this model predicts should lead to greater male variability.
Where the Hypothesis Breaks Down
The male variability hypothesis is not a law of nature. It describes a tendency that shows up in many human traits but fails to appear in others, and it does not generalize cleanly to all species.
As noted in the children’s IQ meta-analysis, females were more variable than males in processing speed, and the sexes did not differ in variability for short-term verbal memory.14Personality and Individual Differences. Sex differences in cognition: A meta-analysis of variance ratios in the Wechsler Intelligence Scales for Children Academic grades tell a complicated story as well. A large meta-analysis of school grades across many countries found that boys’ grades were more variable than girls’ in every subject type, but the gap in variability was smaller in STEM subjects than in non-STEM subjects. Girls’ grades were about eight percent less variable in STEM and about thirteen percent less variable in non-STEM.15Nature Communications. Gender differences in individual variation in academic grades fail to fit expected patterns for STEM If the male variability hypothesis were the primary explanation for male overrepresentation at the top of STEM fields, you would expect variability differences to be largest in STEM. Instead, the variability gap was larger in the humanities and languages, where male overrepresentation at the top is not a widely discussed issue. The researchers argued this was evidence against a simple variability-based explanation for STEM gender gaps.
The most striking counterevidence comes from nonhuman animals. A large meta-analysis examined sex differences in personality traits across over 220 species and more than 2,100 effect sizes, covering boldness, aggression, activity, sociality, and exploration. The result was unequivocal: no significant sex difference in variability for any personality trait in any broad taxonomic group. The degree of sexual size dimorphism, often used as a proxy for the intensity of sexual selection, did not predict the magnitude of sex differences either.16PubMed. A meta-analysis of sex differences in animal personality: no evidence for the greater male variability hypothesis If the hypothesis were a straightforward consequence of sexual selection, you would expect to see it in species where males fight over mates. The absence of the pattern in animal personality is a genuine challenge to universalist versions of the hypothesis.
Sex-Role Reversal and What It Teaches
One way to test whether sexual selection drives variability is to look at species where the usual sex roles are reversed. Pipefish, relatives of seahorses, are a textbook example. Males carry the eggs and invest heavily in each pregnancy, while females compete for access to males. Genetic analysis of Gulf pipefish revealed the most extreme polyandry documented in any species, with females showing far greater variation in mating success than males.17PubMed Central. Genetic evidence for extreme polyandry and extraordinary sex-role reversal in a pipefish In this species, sexual selection falls most heavily on females. According to the evolutionary logic of the variability hypothesis, you would expect females to be the more variable sex in traits related to reproductive competition, and the intense variance in female mating success is at least consistent with that prediction. The pipefish case is often cited as a natural experiment: flip the direction of sexual selection, and you may flip which sex shows more variation, though extrapolating from mating success to broader morphological or cognitive variability requires caution.
Paternal Age and Mutational Input
A separate line of evidence connects greater male variability to the biology of sperm production. Men produce sperm continuously throughout adulthood, and each round of cell division introduces a small chance of copying errors in DNA. Because the cells that become sperm have gone through far more divisions than the cells that become eggs, fathers contribute more new mutations to their children than mothers do. Furthermore, the number of new mutations increases with a father’s age. One study found a strong correlation between paternal age and the rate of new single-nucleotide mutations in offspring’s genomes.18PLoS ONE. Paternal Age Explains a Major Portion of De Novo Germline Mutation Rate Variability in Healthy Individuals
Most of these new mutations are harmless, but some are not. A study examining paternal-age effects on five neurodevelopmental and psychiatric disorders found that offspring of 45-year-old fathers had roughly 10 to 20 percent higher risk than offspring of 25-year-old fathers for conditions including autism spectrum disorder, schizophrenia, and intellectual disability.19Nature Communications. Paternal-age-related de novo mutations and risk for five disorders The absolute risk remains low: a condition with a one-percent baseline incidence would rise to about 1.2 percent. But the pattern matters for variability. Because fathers supply a disproportionate share of new genetic mutations, and because many of those mutations affect neurodevelopment, paternal mutation input is one more mechanism that could push the male-line contribution to trait variation upward. This is distinct from the X-chromosome mechanism; it operates through accumulated copying errors in sperm and affects autosomal genes as well.
What the Hypothesis Does and Does Not Explain
Greater male variability is often dragged into political arguments about representation, particularly in STEM careers and in diagnoses of learning disabilities or behavioral disorders. It is worth being precise about what the data can and cannot support.
The data consistently show that boys are overrepresented at both tails of many cognitive and behavioral distributions. More boys in gifted programs, more boys with reading difficulties, more boys diagnosed with intellectual disabilities, and more boys with extreme behavioral profiles. Greater variability is a partial explanation for this pattern. But it is only one factor among many. Social expectations, educational practices, diagnostic biases, and cultural norms also shape who ends up at the extremes. The academic-grades meta-analysis is a useful reminder: the variability gap was smaller in STEM than in non-STEM, the opposite of what a simple variability-based explanation for STEM gender gaps would predict.20Nature Communications. Gender differences in individual variation in academic grades fail to fit expected patterns for STEM
It is also important to recognize that greater variability at the group level says nothing meaningful about any individual. A given woman can easily be more extreme than a given man on any trait where men are, on average, more spread out. Population-level statistics describe distributions, not people. The practical implication for policy is that variability differences, even where they are real and stable, do not tell you what ratio of men to women you should expect in any specific profession or diagnostic category. Too many other factors intervene between a population’s variance ratio and the real-world outcome anyone cares about.
DNA Methylation and the Autosomal Puzzle
One of the more intriguing newer findings is that sex differences in gene regulation are not confined to the sex chromosomes. The study of autosomal DNA methylation mentioned earlier found that hundreds of sites scattered across the non-sex chromosomes show different levels of methylation in males and females, with the majority being more heavily methylated in females.21PubMed Central. Characterising sex differences of autosomal DNA methylation in whole blood using the Illumina EPIC array Methylation is one of the primary chemical switches cells use to turn genes up or down, so widespread sex differences in methylation mean that male and female genomes are being read differently even when the underlying DNA sequence is identical.
How this connects to variability is not yet well established, and researchers are careful to note that association is not causation. But the finding raises the possibility that some of the variability differences between males and females originate not in the genes themselves but in how those genes are regulated. If male cells are, on average, less tightly regulated at certain genomic locations, you would expect more person-to-person variation in the output of those genes among males. Whether that is actually happening awaits further study. For now, autosomal methylation differences are a piece of the puzzle that does not fit neatly into the older X-chromosome explanation and hints at a more complex regulatory landscape underlying sex differences in variability.