IQ by Race: What Causes Score Gaps Between Groups?

Average IQ test scores differ across racial and ethnic groups in the United States and in cross-national comparisons, but the scientific evidence overwhelmingly points to environmental causes rather than genetic ones. The gaps are real in the narrow sense that they show up on test scorecards, yet decades of research into socioeconomic status, education quality, environmental toxins, test design, and the psychology of test-taking itself have revealed a web of factors that accounts for these differences without invoking innate biology. The story is more complicated, and more interesting, than a simple ranking of group averages.

What the Score Gaps Actually Look Like

In the United States, the most frequently cited gap has been between Black and white Americans, historically averaging around 15 points on standardized IQ tests. Hispanic Americans have tended to score somewhat below white Americans as well, while Asian Americans have scored at or slightly above white averages. These figures are population averages, meaning the distributions overlap enormously. Most individuals in any group score within the same broad range. Still, the group-level differences have persisted for decades in American testing data, which is why they attract so much attention and controversy.

What has also persisted, though it gets less attention, is narrowing. The Black-white gap in the U.S. shrank substantially over the second half of the twentieth century, a timeline that tracks with desegregation, expanded educational access, reductions in childhood lead exposure, and other environmental improvements. That narrowing is itself strong evidence against a fixed genetic explanation, because the gene pools of these populations did not change meaningfully over that period.

Whether Tests Measure the Same Thing Across Groups

Before interpreting any score gap, you need to know whether the test is measuring the same underlying abilities in each group. This is the question of measurement invariance, and the answer is often no. A study examining a neuropsychological battery across racial and ethnic subgroups found that while the tests showed full invariance across sex and over repeated measurements, they showed only partial invariance across racial and ethnic groups, with nonequivalent score intercepts in the language domain. In plain terms, the test items did not function identically across groups, meaning raw score comparisons could be misleading.

1PubMed Central. Measurement invariance of neuropsychological measures of cognitive aging across race/ethnicity by sex/gender groups

A separate analysis of the RAKIT IQ test used with ethnic minorities in the Netherlands illustrated the problem even more sharply. When researchers accounted for intercept differences between groups, what had appeared to be negligible test bias turned out to be quite severe. Ignoring those intercept differences made the test look fair; including them revealed that the scores systematically understated minority performance.

2Educational Measurement: Issues and Practice. Measurement Invariance in Confirmatory Factor Analysis: An Illustration Using IQ Test Performance of Minorities

This does not mean every IQ test is useless across groups. A study of South African engineering students found that the Raven’s Advanced Progressive Matrices behaved the same way for African and non-African students at the same university, with items functioning equivalently and concurrent validity holding up.

3International Journal of Selection and Assessment. Construct Validity of Raven’s Advanced Progressive Matrices for African and Non‐African Engineering Students in South Africa

But that was among highly selected university students sharing the same educational environment. When researchers administered Raven’s tests to samples in South Sudan, one of the least developed countries in the world, scores were so low they appeared to reflect a fundamental lack of familiarity with the test format rather than actual cognitive ability. The researchers concluded that a test built around abstract shapes and analytic reasoning was simply unsuitable there and likely to heavily underestimate average intelligence.

4Personality and Individual Differences. The Raven’s test performance of South Sudanese samples: A validation of criticisms of the utility of Raven’s among Sub-Saharan Africans

The takeaway is that comparing IQ scores across groups requires first establishing that the instrument is measuring the same construct in each group, and that step frequently fails.

The Weight of Socioeconomic Status

Socioeconomic status is one of the strongest predictors of IQ scores, and its effects compound over time. A longitudinal study tracking children from infancy through adolescence found that children from low-SES backgrounds scored about 6 IQ points lower at age 2 than children from high-SES backgrounds. By age 16, the gap had ballooned: boys from low-SES homes scored roughly 15 points lower, and girls about 17 points lower, than their high-SES counterparts.

5PubMed Central. Socioeconomic status and the growth of intelligence from infancy through adolescence

That widening gap is not trivial. It is roughly the same magnitude as the Black-white IQ gap in the United States, and it arises entirely within groups that share a racial category. The implication is straightforward: when racial groups differ systematically in wealth, neighborhood quality, nutrition, and access to stimulating environments, those differences alone can produce the observed score gaps.

Neuroscience research has traced the pathways through which this happens. Childhood socioeconomic status shapes neural development, particularly in brain systems supporting language and executive function.

6PubMed Central. Socioeconomic status and the brain: mechanistic insights from human and animal research

A study of Brazilian children quantified the relationship: SES accounted for roughly a quarter to a third of the variance in IQ, verbal memory, working memory, language, and executive functions, with the effects strongest in children under nine.

7Psicologia: Reflexão e Crítica. Influence of family socioeconomic status on IQ, language, memory and executive functions of Brazilian children

In the United States, racial wealth gaps are extreme and well documented. The median Black household holds a fraction of the wealth of the median white household, and this disparity has roots in centuries of slavery, legal segregation, redlining, and discriminatory policy. When you control for SES, racial IQ gaps shrink, sometimes dramatically. They do not always disappear entirely, but the residual gap is much smaller and is subject to additional environmental explanations.

Education Quality, Not Just Years of Schooling

Education has a direct, causal effect on IQ scores. A meta-analysis covering more than 600,000 participants across 42 datasets found that each additional year of education raised cognitive-ability scores by about 1 to 5 IQ points, with effects that persisted across the lifespan and appeared across all broad categories of cognitive ability studied.

8PubMed Central. How Much Does Education Improve Intelligence? A Meta-Analysis

A Norwegian natural experiment reinforced this. When Norway extended compulsory schooling from seven to nine years in the 1960s, the additional two years of education produced a substantial IQ gain measured at age 19.

9PubMed Central. Schooling in adolescence raises IQ scores

And research has shown that schooling’s cognitive benefits go beyond crystallized knowledge. One study found that a year of schooling contributed meaningfully to working memory development, with the schooling effect double that of simply aging one more year for working memory in the age range studied.

10npj Science of Learning. Schooling substantially improves intelligence, but neither lessens nor widens the impacts of socioeconomics and genetics

But the number of years spent in school is only part of the picture. A review of cross-cultural influences on Wechsler IQ tests, using data from a southern African sample, found that quality of education mattered as much as quantity. Black African first-language participants who had received advantaged-quality education scored comparably to white English first-language participants and to the U.S. standardization sample. Those who had received disadvantaged education scored significantly lower. The researchers warned that failing to account for education quality could lead to faulty conclusions about the effects of ethnicity and even neuropsychological misdiagnosis.

11PubMed. Cross-cultural effects on IQ test performance: a review and preliminary normative indications on WAIS-III test performance

This finding alone undercuts a genetic interpretation. If people from different racial groups score similarly when given comparable educational environments, the group-level gap is tracking education quality, not biology.

Lead and Other Environmental Toxins

Environmental lead exposure provides one of the most concrete demonstrations of how a non-genetic factor can produce large, racially stratified IQ differences. A systematic review and meta-analysis of studies in children under 12 found that those exposed to lead concentrations above 10 micrograms per deciliter for more than 4.5 years scored roughly 23 IQ points lower than controls. Even shorter exposure periods produced significant decrements.

12PubMed Central. The effect of lead exposure on IQ test scores in children under 12 years: a systematic review and meta-analysis of case-control studies

There is no safe threshold. A landmark study found that IQ declined by about 7.4 points as lifetime average blood lead concentrations increased from 1 to just 10 micrograms per deciliter, a range previously considered safe. The damage was steepest at the lowest concentrations, meaning the first increments of lead exposure were the most harmful per unit.

13PubMed Central. Intellectual impairment in children with blood lead concentrations below 10 microg per deciliter

And the effects are not just childhood phenomena. A study using a nationally representative sample of U.S. older adults, linked to historical data from 1940, found that people who grew up in cities with lead pipes and water chemistry that promoted lead leaching had worse cognitive functioning decades later. About a quarter of the association between childhood lead exposure and late-life cognition was explained by differences in educational attainment, suggesting lead harmed cognition both directly and by undermining schooling.

14PubMed Central. Childhood lead exposure is associated with lower cognitive functioning at older ages

The racial dimension is critical. In the United States, Black and Hispanic children have historically been exposed to far higher levels of lead than white children, a disparity rooted in housing segregation, proximity to industrial pollution, and older housing stock with lead paint. When an environmental toxin that can erase 7 or more IQ points hits one racial group harder than another, it creates a group-level IQ gap that has nothing to do with genetics.

The Flynn Effect and the Malleability of IQ

IQ scores rose steadily throughout the twentieth century across many countries, a phenomenon known as the Flynn effect. A meta-analysis confirmed this pattern as robust and widespread.

15PubMed Central. The Flynn effect: a meta-analysis

The gains were enormous, on the order of 3 points per decade in many populations, meaning an average person today would outscore their grandparent by 15 or more points on the same test. A study using Norwegian family data showed that both the rise and the more recent slowdown in IQ scores are environmentally caused, because the trends appear within families across generations sharing the same gene pool.

16PubMed Central. Flynn effect and its reversal are both environmentally caused

The Flynn effect matters for racial comparisons because it demonstrates that IQ is far more responsive to environmental conditions than a purely genetic model would allow. If the entire population of a country can gain 15 IQ points in a few decades through better nutrition, more schooling, greater exposure to abstract thinking, and reduced disease burden, then a 15-point gap between racial groups sharing the same era but not the same environments is entirely plausible as an environmental artifact.

Stereotype Threat and the Psychology of Testing

The conditions under which a test is taken also matter. The landmark experiments by Steele and Aronson showed that when Black participants took a difficult verbal test described as diagnostic of their intellectual ability, they underperformed relative to white participants. When the same test was described as non-diagnostic, the gap disappeared, even with SAT scores statistically controlled.

17PubMed. Stereotype threat and the intellectual test performance of African Americans

Simply making participants aware that their race was associated with stereotypes about intellectual ability was enough to depress performance.

Subsequent research has extended the finding. A study of Black adolescents on standardized math tests found that stereotype threat affected not just overall scores but test-taking strategies themselves, altering how students approached problems.

18Harvard Educational Review. The Effects of Stereotype Threat on Standardized Mathematics Test Performance and Cognitive Processing

The research landscape on stereotype threat has grown more complicated over the years. More recent work has found that the classic interaction pattern (race by threat condition) does not always replicate cleanly. One recent study failed to find strong evidence for the classic effect but did find that among Black students, higher working memory capacity moderated the threat, helping higher-capacity students maintain performance under stereotype-activating conditions.

19PubMed Central. Higher trait working memory capacity may benefit standardized test performance under race-related stereotype threat

The mechanism is probably real but less uniform and more context-dependent than the original experiments suggested. What remains clear is that psychological context can influence scores, and standard testing environments are not psychologically neutral for all test-takers.

Why Heritability Within Groups Says Nothing About Gaps Between Groups

One of the most persistent misconceptions in this area involves heritability. IQ is substantially heritable within populations, meaning that within a given group sharing broadly similar environments, genetic differences account for a significant portion of individual variation. Some commentators have leaped from this fact to the claim that the gap between racial groups must also be partly genetic. That leap is logically invalid.

A 2024 analysis in the Proceedings of the National Academy of Sciences laid this out formally: perfect knowledge of within-group heritability provides no information about between-group heritability. Even if you could estimate the genetic contribution to a between-group difference, that estimate would not tell you the direction of the genetic and environmental components.

20PubMed Central. Heritability within groups is uninformative about differences among groups: Cases from behavioral, evolutionary, and statistical genetics

The standard illustration makes the point intuitive. Take two handfuls of genetically identical seeds. Plant one handful in rich soil with ample water and sunlight, and the other in poor soil with inadequate water. Within each plot, the variation in plant height will be substantially genetic, because the environment is held roughly constant within each plot. But the difference in average height between the two plots is entirely environmental. Knowing that height is “heritable” within each plot tells you nothing about why one plot is taller than the other.

The same logic applies to IQ and racial groups. Within any group, genetic variation contributes to individual differences in test scores. But the gap between groups that have experienced dramatically different environments, stretching from slavery and colonialism to modern segregation and wealth inequality, cannot be attributed to genetics on the basis of within-group heritability. The argument simply does not follow.

Gene-Environment Interactions and the Limits of Genetic Prediction

The relationship between genes and environment is not additive in a simple way. Research on the Scarr-Rowe hypothesis has explored whether the heritability of cognitive ability changes depending on socioeconomic environment. The hypothesis holds that in poorer environments, environmental deprivation constrains cognitive development so strongly that genetic potential cannot express itself fully, reducing measured heritability.

21PubMed Central. The Scarr-Rowe Interaction in Complete Seven-Year WISC Data From Louisville Twin Study: Preliminary Report

If this pattern holds broadly, it means that the very populations facing the most environmental disadvantage are also the ones in which genetic potential is most suppressed, making score gaps even less informative about underlying genetic capacity.

Modern genomics has added another wrinkle. Polygenic scores, which combine thousands of tiny genetic effects to predict traits like educational attainment or cognitive ability, lose their predictive accuracy when applied to populations different from the one used to develop the score. As one analysis noted, nearly fifty years after the original debates about locality in genetics, a new problem of portability has emerged: polygenic predictions break down across populations.

22PubMed Central. Half a century later and we’re back where we started: How the problem of locality turned in to the problem of portability

Since the vast majority of genome-wide association studies have been conducted on people of European descent, polygenic scores for cognitive traits cannot be meaningfully compared across racial groups. The tool simply does not transfer, which closes off what might seem like the most direct genetic test of group differences.

Why Early Intervention Studies Have Not Settled the Debate

Some researchers have looked to early childhood intervention programs for evidence about whether environmental improvements can close racial IQ gaps. The Infant Health and Development Program (IHDP), a large-scale randomized trial providing intensive early support to low-birth-weight children, did find meaningful cognitive gains for low-income children. But when the researchers examined whether the program differentially benefited minority children in a way that would close racial gaps specifically, they did not find consistent impact patterns favoring minority children as they did for low-income children.

23PubMed Central. Can Intensive Early Childhood Intervention Programs Eliminate Income-Based Cognitive and Achievement Gaps?

This result has sometimes been misinterpreted as evidence for a genetic component to racial gaps. That interpretation misreads the finding. An early childhood program lasting a few years cannot undo the full suite of environmental differences between racial groups, many of which are structural and begin before birth and continue throughout a lifetime. Neighborhood quality, school funding, exposure to chronic stress, environmental toxins, healthcare access, and accumulated family wealth all differ systematically by race in the U.S. and are not addressed by a short-term educational intervention. The IHDP result tells you about the reach of one specific program, not about the limits of environmental influence in general.

Testing in Employment and Institutional Settings

The practical stakes of group-level score differences are high. Cognitive ability tests are widely used in hiring, school placement, and military selection. When those tests produce different average scores across racial groups, they can have a disparate impact on access to jobs and education. A review in the employment context noted the ethical and legal concerns surrounding differential validity and differential prediction of cognitive tests across racial groups.

24Annual Review of Organizational Psychology and Organizational Behavior. Differential Validity and Differential Prediction of Cognitive Ability Tests: Understanding Test Bias in the Employment Context

Differential prediction means that the same test score may not predict the same level of job or academic performance for members of different groups. If a test overpredicts performance for one group and underpredicts for another at the same score, using a single cutoff creates systematic unfairness even if the test is technically “valid” in aggregate. This is not just a statistical footnote. It affects who gets hired, who gets promoted, and who gets into selective schools.

Courts and regulatory bodies in the United States have long grappled with this. The Uniform Guidelines on Employee Selection Procedures, established in 1978, require that when a selection procedure produces adverse impact against a protected group, the employer must demonstrate that the procedure is job-related and consistent with business necessity. The existence of group-level IQ differences does not, legally or scientifically, justify treating those differences as reflecting ability differences without examining the fairness of the instrument itself.

Brain Imaging and the Search for Biological Substrates

Some researchers have turned to neuroimaging to look for brain-based explanations of cognitive differences. A cross-ethnicity study using diffusion tensor imaging and functional MRI in two cohorts examined how brain structure-function coupling related to age and cognition. The study found that weaker coupling between brain structure and function was associated with increased age and worse cognitive performance, and that coupling in visual networks related to visuospatial ability.

25PubMed. Relationships between brain structure-function coupling in normal aging and cognition: A cross-ethnicity population-based study

What such studies consistently reveal is that brain structure and function are shaped by experience. The same SES and environmental factors that drive IQ differences also shape brain development, as the neuroscience research on socioeconomic status and neural development has documented. Finding a brain correlate of a cognitive difference does not mean the difference is innate. It means the brain reflects the environment it developed in, which is exactly what you would expect of an organ shaped by decades of experience, nutrition, stress, and stimulation.