Does Education Increase IQ? Evidence From Schooling Studies
Author: Angela Bogdanova · Published: October 12, 2026 · Editorial Policy
Research cluster: Intelligence, Reasoning & Cognitive Abilities · Written by Artificial Sapiens
Education can raise performance on intelligence tests. The strongest evidence comes from studies that exploit differences in school-entry dates, changes to compulsory-schooling laws, and measures of cognitive ability collected before educational pathways diverge. The influential Ritchie and Tucker-Drob (2018) meta-analysis combined 142 estimates from 42 data sets involving more than 600,000 participants and reported gains equivalent to approximately 1–5 IQ points per additional year of education. That range summarizes research estimates, not a promise that every student gains a fixed number of points.
The central scientific question is what schooling changes. A higher score can reflect acquired knowledge, strategies for reasoning and solving unfamiliar problems, test-relevant skills, and changes in several cognitive abilities. The evidence supports an effect of education on measured cognitive performance; it does not establish that every effect operates through a single underlying general-intelligence factor. This distinction matters when interpreting test results, educational policy, individual development, and popular claims about getting smarter.
The short answer: does education increase IQ?
Yes. Across multiple quasi-experimental research designs, additional formal schooling is associated with causally interpretable improvements in scores on a range of intelligence and cognitive-ability tests. The most cited quantitative synthesis found positive effects for its three major research designs, with effects differing substantially by study method and test type. The authors described education as a particularly consistent and durable influence on measured intelligence Ritchie and Tucker-Drob, 2018.
The answer needs four qualifications. First, these are average differences estimated across groups and populations, so one person's response can differ. Second, IQ points are standardized test-score units, with the conventional IQ standard deviation equal to 15; they are not units of brain growth. Third, the experimental leverage often comes from reforms or birth-date cutoffs rather than individually randomized years of school. Fourth, gains in observed scores can arise through broad and specific pathways, and deciding whether latent general intelligence itself changed requires additional measurement models.
In practice, the strongest conclusion is that schooling contributes to cognitive development and to the capacities assessed by intelligence tests. It is an important environmental influence that belongs alongside development, health, social conditions, and prior cognitive differences in any serious account of intelligence.
What does an increase in IQ actually mean?
An IQ score summarizes performance on a specified set of tasks relative to a standardization sample. Contemporary batteries often assess combinations of verbal reasoning, knowledge, fluid reasoning, visuospatial ability, working memory, and processing speed. The exact combination depends on the instrument, its age norms, its construction, and its scoring rules. A full-scale IQ score is a composite, and two batteries may sample cognitive abilities differently.
A gain of three IQ points represents approximately 0.20 of a standard deviation on the familiar mean-100, standard-deviation-15 scale. It is a meaningful research metric for comparing groups, but interpretation requires attention to test reliability, confidence intervals, the test version, age norms, language, and opportunities to learn the content. The Standards for Educational and Psychological Testing provide the professional framework for evaluating validity, score uses, and the conditions under which interpretations are justified.
The distinction between an observed score and a latent psychological construct is crucial. The general factor of intelligence, usually called g, describes common variance across cognitive tasks. Broad abilities, including fluid and crystallized abilities, characterize additional patterns of variation. A composite IQ can change because some constituent tasks improve more than others, even when a particular model of latent g remains unchanged. Conversely, a valid effect on a range of different tests suggests schooling reaches beyond simple memorization of classroom facts.
An individual's apparent gain across repeated testing may also include practice, familiarity, fatigue differences, examiner effects, or regression to the mean. Strong studies try to separate those explanations from added exposure to education.
Why a correlation between education and IQ cannot settle causation
People who complete more schooling tend, on average, to perform better on cognitive tests. Two processes can generate that relationship. Schooling may develop abilities that help on the tests. Earlier cognitive performance may also help a student remain in education, obtain qualifications, or enter academically demanding programs. Both processes can operate simultaneously.
Family resources, school opportunities, neighborhood context, educational policy, health, language, and prior academic performance can influence both schooling and subsequent scores. A conventional regression of IQ on years of education can therefore mix a causal schooling effect with selection and confounding. Adjusting for measured earlier ability helps, although unmeasured influences can remain.
Researchers address this problem by finding sources of educational variation that have less to do with a student's prior ability. Reforms affecting school-leaving ages, eligibility rules based on dates of birth, and long-term studies with childhood IQ measures provide useful leverage. Their strength comes from comparing people exposed to different amounts of schooling for reasons partly outside individual choice. Their limitations come from the assumptions needed to treat that variation as a credible counterfactual.
How scientists estimate the causal effect of schooling
Longitudinal studies that control for earlier intelligence
In a longitudinal design, researchers measure cognitive ability before people complete different amounts of education, then test them again in later life. The earlier measure helps separate starting differences from later divergence. If education still predicts subsequent test performance after adjustment for childhood scores, the finding is consistent with a schooling effect.
The major advantage is temporal ordering: earlier test performance is observed, rather than reconstructed from memory. A limitation remains because people can differ in unmeasured motivation, opportunity, health, family circumstances, and school quality. A regression coefficient from such a study therefore deserves more causal caution than a credible policy-based natural experiment. In the 2018 meta-analysis, the prior-intelligence-control design yielded an average estimate of 1.197 IQ points per educational year in its initial model Ritchie and Tucker-Drob, 2018.
Compulsory-schooling reforms as natural experiments
Governments sometimes increase the minimum number of years that students must attend school. When a reform is introduced in different regions or for different birth cohorts, researchers can compare otherwise similar groups who encountered different rules. Instrumental-variable and difference-in-differences methods can help estimate the effect of the additional schooling induced by the reform.
These methods require specific assumptions. Reform timing should not coincide with other changes that explain the outcome; comparison groups need credible counterfactual trends; and, for instrumental-variable interpretations, the relevant policy variation must affect the cognitive outcome through the schooling pathway specified by the model. Education reforms can alter curriculum, peers, school resources, expectations, and credentials alongside years enrolled. A policy estimate is therefore often best understood as the effect of the schooling package induced by that reform.
The 2018 meta-analysis estimated about 2.056 IQ points per additional year for its policy-change design before expanded covariate adjustment. That is an estimate for studies in that design family, not a universal constant for school systems everywhere Ritchie and Tucker-Drob, 2018.
School-entry cutoffs and regression discontinuity
Most school systems specify a birthday cutoff for entry into a grade. Children born just on opposite sides of that date can be similar in age yet receive noticeably different amounts of formal schooling by the time a test is given. Comparing their cognitive outcomes, with careful attention to the cutoff, helps isolate school exposure from biological maturation.
The approach has a powerful intuitive appeal: if two children are nearly the same age, differences associated with their grade placement may illuminate the contribution of schooling. But cutoff compliance, delayed entry, parents' decisions, and the mixture of school experience with other grade-related factors can complicate interpretation. The results may also identify a local effect for children around the eligibility threshold rather than everyone at every age.
A pioneering study by Cahan and Cohen (1989) compared the contribution of school experience with chronological age in children's intelligence development. In the 2018 synthesis, school-entry-cutoff designs produced the largest initial average estimate, 5.229 IQ points per school year. That estimate differed markedly from the other designs and included tests closely related to achievement Ritchie and Tucker-Drob, 2018.
What the landmark 2018 meta-analysis actually found
The paper How Much Does Education Improve Intelligence?, by Stuart J. Ritchie and Elliot M. Tucker-Drob, was published in Psychological Science. It synthesized 142 effect sizes from 42 data sets covering 615,812 participants across three quasi-experimental design categories. Its goal was to estimate how much additional education changes cognitive test performance once research designs have addressed some of the usual self-selection problem.
Its initial design-specific models found roughly 1.20 IQ points per year after accounting for earlier intelligence, 2.06 for educational-policy changes, and 5.23 for school-entry cutoffs. A joint model yielded an overall estimate of 3.394 points. These figures come from different design families and should be read together with the differences in outcome tests, age groups, and assumptions. The widely reported 1–5-point range conveys the spread across the main approaches better than treating 3.4 as a universal prediction for one extra grade.
Some more extensively adjusted analyses reduced the estimated effects. Where larger covariate sets were available, the prior-intelligence design produced an estimate of 0.903 points and the policy-change design 1.852 points. The revised estimates remained positive. This matters: a strong scientific summary reports sensitivity to adjustment as well as the headline result full analysis.
The authors examined age at intervention, age at later testing, type of cognitive test, and possible publication bias. Their findings supported schooling effects across broad categories of cognitive ability, including fluid and crystallized measures, while showing meaningful heterogeneity. Estimates based on achievement-linked tests, particularly within school-cutoff designs, can be larger than estimates from less curriculum-dependent measures.
A meta-analysis inherits the assumptions and limits of its constituent studies. Combining more than half a million observations increases the breadth of evidence; it does not randomly assign years of education, make every school system equivalent, or answer every question about mechanisms.
A concrete natural experiment: Norway's school reform
Norway gradually expanded compulsory schooling from seven to nine years in different municipalities during the mid-twentieth century. Researchers Christian N. Brinch and Taryn Ann Galloway studied men born from 1950 to 1958, using later military-conscription cognitive test scores. The staggered implementation allowed comparison of cohorts and municipalities exposed to different educational requirements.
Their sample included 107,223 men in the main analysis. The reform was associated with approximately 0.163 additional years of schooling on average and a 0.603-point rise on the converted IQ scale. Using an instrumental-variable analysis, the authors estimated an effect of approximately 3.69 IQ points per additional year of schooling among those whose education was influenced by the reform Brinch and Galloway, 2012.
The distinction between 0.6 and 3.7 is essential. The first is the average effect of the reform on scores in the study population; the second scales that change by the reform-induced increase in years of education. It is misleading to say that every student received two extra years and therefore gained 7.4 IQ points. The first-stage change in measured educational attainment was much smaller, and the instrumental-variable estimate applies under its identifying assumptions.
This study gives strong evidence that a policy increasing adolescent schooling can affect cognitive test scores near age 19. Its setting also shapes external validity: historical Norway, compulsory reform, a male-only military testing sample, and particular tasks cannot automatically represent modern schooling in every country.
A second line of evidence: school entry and age
The Cahan and Cohen (1989) school-age analysis separated time in formal education from chronological age. It remains influential because schools organize children's experiences in ways that do not line up perfectly with birthdays. A child can acquire more reading, reasoning, numerical, and test-relevant experience than a slightly younger peer while being separated by only weeks in age.
The comparison provides a different kind of evidence from compulsory school-leaving reforms. It centers on the early school years, while reforms such as Norway's involve adolescents. Results from both directions strengthen the broad conclusion that education matters, but they need not have equal magnitudes or affect the same cognitive domains.
A 2025 systematic synthesis of schooling and executive functions used cutoff-related research in children aged about 4.5–9 years. Across twelve studies covering approximately 1,611 children, the main pooled estimate was small but positive, Hedges g = 0.24 (95% CI 0.13–0.36) Donenfeld and colleagues, 2025. This concerns executive-function tasks, such as inhibition or working memory, rather than full-scale IQ. It strengthens the view that school exposure can influence cognitive processes while requiring a distinction between task families.
Does schooling increase general intelligence, or particular skills?
This is one of the most important unresolved interpretive questions. Intelligence batteries produce correlated task scores, but an educational gain in a test battery does not automatically identify which latent factors changed. School may strengthen vocabulary and knowledge, improve numerical techniques, expand familiarity with abstract symbols, support attention during testing, and encourage strategies for unfamiliar problems. The resulting pattern can raise composite scores.
Ritchie, Bates, and Deary (2015) studied a Scottish longitudinal sample of 1,091 people who had intelligence testing around age 11 and a battery of ten cognitive tests around age 70. Their structural equation models indicated that the association between education and later cognitive test scores was expressed primarily through specific cognitive skills rather than being mediated by the general factor g. That finding challenges the simple inference that every IQ-score gain is a gain in the same underlying domain-general ability.
The 2018 meta-analysis, meanwhile, reported educational gains across broad categories of tests, including fluid and crystallized outcomes Ritchie and Tucker-Drob. These findings can coexist because they answer related but different questions. One asks whether test performance improves; another asks which statistical level of a latent-variable model best explains those improvements in a particular longitudinal population.
A more precise research program needs repeated measurement with multiple indicators per ability, defensible measurement invariance over time and across education groups, and models that allow both common-factor and domain-specific paths. It also needs to distinguish actual improvements in reasoning from increasing fluency with the formats and cultural tools used by tests.
Does education raise fluid intelligence?
Fluid intelligence refers to reasoning and problem solving where solutions are relatively independent of previously learned factual content. Typical tasks may involve identifying patterns, applying rules, making comparisons, or solving novel configurations. It is tempting to assume that schooling can change only factual knowledge and therefore should have no influence on these tests.
The evidence is broader. The 2018 synthesis reported positive effects for measures classified as fluid and crystallized. Some fluid tests involve skills in symbol handling, sustained attention, rule inference, task familiarity, and motivation that can be influenced by education. An effect on a fluid reasoning test therefore deserves to be taken seriously, although the extent of its transfer to very different problems remains a separate question.
Changes in scores should be attributed at the level actually measured. A fluid-reasoning test gain supports a schooling effect on that outcome; determining whether it changes a general latent fluid factor across many unrelated tasks requires further modeling. The Ritchie and Tucker-Drob meta-analysis supports effects across broad domains while leaving some mechanism questions open.
Does education raise crystallized intelligence?
Crystallized intelligence concerns the organized knowledge and learned competencies developed through experience, including vocabulary and general information. Formal education provides structured exposure to language, mathematical concepts, scientific explanations, and cultural knowledge, so a relationship with this domain has an especially direct pathway.
A student who encounters technical vocabulary, works through increasingly complex texts, or studies historical and scientific frameworks acquires representations that may later support reasoning as well as test performance. Schooling can also improve strategies for retrieving and using knowledge. The schooling evidence across broad cognitive categories is consistent with these routes Ritchie and Tucker-Drob, 2018.
Even here, test construction matters. A measure of vocabulary or knowledge can partly depend on educational access and cultural familiarity, and fairness depends on how scores are used. A high score is evidence of strong performance on the test under its conditions, rather than proof of a single innate quality.
What about executive functions and working memory?
Executive functions involve coordinating behavior toward goals, including inhibitory control, switching, and working-memory-related operations. Classroom routines ask students to follow instructions, maintain relevant information, filter competing responses, and adapt to changing demands. Those activities make an educational effect plausible.
The Donenfeld et al. meta-analysis identified a small pooled benefit from schooling on executive-function outcomes in early childhood, using comparisons informed by grade-entry cutoffs. The finding should be interpreted as evidence about executive-function task performance in the studied age range. It cannot be treated as a numerical estimate of IQ gain, nor can it establish that every executive-function component changes equally.
Working memory, executive functions, fluid reasoning, and IQ overlap empirically but remain different constructs. A schooling study that measures one of them should report that outcome and avoid relabeling it as another.
Is the effect sustained after students leave school?
An IQ benefit observed during the school years is one result; a difference detectable decades later is a stronger claim about persistence. The 2018 synthesis examined age at later assessment and found evidence of benefits beyond immediate schooling. Its policy-change subset did not show a significant age-related weakening of the effect, while the prior-intelligence-control subset showed a declining estimated association with longer follow-up Ritchie and Tucker-Drob, 2018.
This pattern supports the possibility of enduring gains while also demonstrating why persistence must be discussed by research design. People with additional schooling may encounter different work, continuing learning, income, health environments, and cognitively demanding activities over their lives. These downstream pathways may help maintain skills, yet their mediating roles require dedicated causal evidence.
In the Scottish cohort examined by Ritchie, Bates, and Deary (2015), education was associated with particular cognitive outcomes many decades after childhood ability had been measured. That is informative about long-term associations, but its structural model also cautions against equating all later gains with a single increase in latent g.
Does the number of years matter more than educational quality?
Years of education is an accessible measure, but it is a coarse indicator of what people actually learn. An additional year may contain more or less instruction depending on attendance, curriculum, teachers, class environment, language accessibility, and the student's opportunities to engage. Reforms often change compulsory attendance together with curriculum and institutional features, so the treatment is rarely an isolated quantity of time.
The question of marginal returns also depends on the age at which schooling is extended and on whether the next year introduces new material, repeats earlier content, or leads to a qualification. The empirical literature supports the causal importance of schooling, but it does not establish a single effect size for every grade, program, country, or educational system.
A study of schooling reforms across 21 countries by Cappellari, Checchi, and Ovidi (2023) reports positive effects on internationally comparable adult numeracy and literacy outcomes, with variation linked to completing qualifications and the timing of schooling. Its outcome measures are cognitive skills, not full-scale IQ, and the 2023 working-paper evidence should be distinguished from the specific IQ meta-analysis.
For educators and policymakers, the implication is to measure substantive learning as well as time enrolled. The relevant goal is durable growth in abilities, participation, and knowledge rather than maximizing a single standardized number.
Does an extra year of college have the same effect as an extra year of school?
The evidence does not justify treating all educational years as interchangeable. A final year of high school and a year of advanced university work differ in curriculum, selection into the program, age, intensity, and the cognitive demands placed on students. The estimated average effect of an additional year in studies of compulsory schooling cannot simply be applied to voluntary higher education.
College students are selected on earlier achievement, resources, interests, and access. Longitudinal designs can adjust for some selection, but isolating the causal effect of attending university on intelligence requires a credible comparison group. Courses can develop specialized knowledge and intellectual practices while the size of any resulting composite IQ change remains an empirical question.
Research should specify what kind of education was added, whom it affected, which cognitive outcomes were measured, and how much time elapsed before testing. Those details are more informative than an undifferentiated statement that a degree increases IQ.
Do all students gain the same number of IQ points?
No. A study-level average hides considerable variation in baseline ability, schooling exposure, institutional context, instructional content, language, and later life experience. Individual outcomes also contain measurement error. It is therefore scientifically incorrect to predict one student's score by multiplying years of school by a published mean effect.
The evidence on differences by sex, socioeconomic background, or earlier ability depends on the design and outcome. A particular reform may produce similar average benefits across subgroups while another may not. Observed group differences require careful analysis of sampling, access, confounding, and measurement comparability; they cannot determine the prospects of any specific learner.
Some studies explore whether schooling reduces, preserves, or increases variation associated with prior family or genetic differences. Those analyses concern statistical associations within populations and particular contexts. They offer no basis for interpreting heredity as individual destiny or for assigning fixed cognitive potential to social groups.
Why school may improve cognitive test performance: plausible pathways
Accumulated knowledge and conceptual tools
Schooling teaches language, mathematical notation, scientific categories, written reasoning, and culturally shared systems for organizing information. These tools improve how learners encode and interpret tasks and allow knowledge to support new inferences. This is an especially clear pathway for knowledge-rich or verbally loaded measures.
Repeated practice in reasoning
Students compare arguments, solve multi-step problems, classify examples, explain causes, and test rules. Such activities can develop strategies used on cognitive assessments. Whether a strategy transfers to a novel domain depends on overlap between learning and assessment as well as the learner's understanding of when the strategy applies.
Attention, self-regulation, and task familiarity
Formal instruction involves attending to instructions, working under constraints, sustaining effort, and adapting to evaluation. These experiences may improve performance partly through executive processes and familiarity with structured testing. The small schooling-associated executive-function effect reported by Donenfeld et al. is relevant, although mechanisms cannot be uniquely identified from that pooled estimate.
Opportunities and downstream experiences
More education may change occupations, continuing learning, literacy practices, and social environments that sustain specific cognitive skills. This creates a possible pathway from initial schooling into long-term performance. Demonstrating the size of the pathway requires more than showing that more-educated adults score higher.
These mechanisms are compatible and can reinforce one another. Current evidence identifies education as a cause of improved test performance more confidently than it partitions every point of that improvement among mechanisms.
Is the effect just teaching to the test?
Learning test-relevant content is one legitimate source of improved test scores. If a vocabulary test measures words encountered in school, that acquisition is a real gain in knowledge. Yet researchers also care whether the improvement generalizes to tasks that were not specifically taught.
The landmark synthesis found schooling effects in both achievement-classified and other cognitive tests, although achievement-related tasks could have larger gains, especially in school-entry-cutoff studies Ritchie and Tucker-Drob, 2018. This reduces the plausibility of a narrow explanation based only on direct classroom rehearsal of test items. It does not prove unlimited far transfer to all activities requiring intelligence.
The decisive question is construct validity: what interpretation does the task support, and how similar is it to the instruction received? Researchers improve inference by using multiple outcome types, checking measurement invariance, and distinguishing immediate skill acquisition from transfer to unrelated problems.
Schooling versus brain-training apps
A structured school curriculum and a short computerized cognitive-training program expose learners to very different experiences. Formal education includes language, knowledge, problem solving, social interaction, and repeated application over long periods. A brain-training app often optimizes performance on a restricted set of tasks.
A second-order meta-analysis by Sala and colleagues (2019) found that gains from cognitive-training programs commonly transfer to closely related tasks, whereas credible far-transfer effects were small or absent after accounting for relevant biases. That conclusion pertains to the cognitive-training studies it synthesized.
The schooling evidence cannot be reduced to claims about an app, and the weak far transfer in a narrow training literature cannot be used to dismiss formal education. The intervention, duration, population, and measured outcomes must align before evidence can be compared.
School closures, disrupted learning, and cognitive development
Interrupted schooling creates important questions about lost instructional exposure and opportunities to learn. Research on learning loss can document differences in reading, mathematics, or academic attainment, but an achievement decline is not automatically an equivalent decline in full-scale IQ. A causal statement about intelligence requires appropriate cognitive outcomes and a design that distinguishes disruption from other concurrent events.
School closures may also coincide with changes in family stress, health, economic conditions, sleep, digital access, and social support. These pathways complicate attributing a later cognitive difference to the absence of classroom time alone. The best practical response is to measure the abilities of concern directly, support opportunities for effective learning, and resist translating one outcome into another.
What schooling studies can and cannot prove
They can support average causal effects on specified cognitive tests when assignment rules, preexisting trends, and study assumptions are credible. Different quasi-experiments together strengthen confidence that education matters. They can also reveal variation in effect estimates by timing, outcome type, and study design.
They cannot prove that one extra year produces the same IQ increase for every person or that a gain on a composite score consists entirely of change in g. They cannot establish that more time in education is always superior to better instruction. They do not automatically translate adult literacy or numerical-skill gains into standardized IQ gains.
They also cannot determine whether an individual has an intellectual disability, giftedness, or a clinical condition. Those questions require appropriate comprehensive assessment, context, and professional judgment when relevant. An observed IQ score is neither a diagnosis nor a complete account of a person's abilities.
How to read a new study claiming that education increases intelligence
Start with the outcome. Was the measure a recognized standardized intelligence battery, a single reasoning task, a vocabulary test, executive function, achievement, or an adult numeracy assessment? These outcomes answer different questions.
Then inspect the design. Did the study use a credible policy discontinuity, a school-entry cutoff, or only ordinary statistical adjustment? Was cognitive performance measured before the educational difference arose? Was the analysis designed in advance, and were alternative explanations tested?
Next examine the scale. Did the authors report IQ points, a standardized mean difference, percentile ranks, raw test points, or correlations? A Hedges g of 0.24 on executive functions is not 0.24 IQ points, just as an effect on a literacy scale is not directly a full-scale IQ effect.
Finally, ask about durability, representativeness, uncertainty, and mechanisms. Whose education changed, at what age, under which system, and how much later was the test administered? Did the model estimate a total score effect, a domain-specific effect, or a latent-factor change? Those distinctions separate a useful scientific finding from an overstated headline.
What the evidence means for students and families
An individual can learn more, strengthen reasoning skills, and gain proficiency through education. Intelligence test scores are responsive to experience at a population level, and educational opportunities deserve serious consideration in how cognitive abilities develop. A single test result should be interpreted within its norms, reliability, opportunities to learn, and the person's broader profile.
For families, the practical priorities are sustained access to meaningful instruction, appropriate support for learning differences, rich language and numerical experiences, and educational environments that allow participation. An exclusive focus on increasing a full-scale IQ number may obscure more directly valuable outcomes such as literacy, comprehension, independent problem solving, and academic confidence.
A student who receives extra instruction in an area of difficulty can improve that area, even if no one measures a corresponding change in IQ. Educational gains are valuable on their own terms. Research about intelligence helps clarify one mechanism of development, not replace the goals of education.
What the evidence means for teachers and policymakers
The most defensible interpretation of the research is that formal education is part of the causal environment shaping cognitive performance. Compulsory-schooling policies can affect test outcomes; high-quality opportunities to acquire and apply knowledge are therefore meaningful candidates for public investment.
Policy evaluation should assess who benefits, which skills improve, whether gains persist, and how benefits compare with costs and alternative uses of resources. It should account for instruction quality and equity of access as well as attendance years. Since natural experiments often capture a bundle of changes, designing educational improvements requires evidence about specific program components.
School systems should also resist stigmatizing interpretations of differences in scores. Differences may reflect multiple interacting pathways, including measurement, opportunity, language, and environment. The central task is to provide learners with access to instruction and support that helps them develop.
Common myths about education and intelligence
Myth: IQ is fixed at birth
IQ test performance shows meaningful stability in many settings, but stability is compatible with environmental effects. Longitudinal correlations between people do not mean the distribution or an individual's performance cannot respond to schooling. Evidence from educational reforms establishes that at least some test outcomes change causally with educational exposure.
Myth: every extra school year adds five IQ points
The 1–5-point figures summarize design-specific average estimates across studies. The Norway estimate of about 3.7 points is an instrumental-variable estimate for people affected by a particular reform, while the overall reform effect on scores was about 0.6 points. Neither is a universal growth rate for an individual's IQ.
Myth: schooling only increases factual knowledge
Knowledge acquisition is important, but quasi-experimental effects also appear in tests classified beyond narrowly curriculum-based achievement. The pattern does not eliminate the need to distinguish general factors from domain-specific skills.
Myth: a higher IQ score proves that latent g increased
IQ composites are observed variables built from test performance. A change can occur through several components. The Ritchie, Bates, and Deary (2015) structural modeling study gives a direct reason to investigate domain-specific paths separately.
Myth: brain games and school have identical effects
Interventions need to be compared on their actual content and outcomes. Evidence that a brief program boosts performance on trained tasks does not establish durable broad transfer, while formal schooling has a distinct quasi-experimental evidence base.
Frequently asked questions
How many IQ points does one extra year of school add?
The best-known 2018 meta-analysis reported approximately 1–5 IQ points per extra year across its quasi-experimental designs. Its initial design-specific means were about 1.2, 2.1, and 5.2 points. The size depends on method, age, outcome, and population; it is an average research estimate rather than an individual prediction Ritchie and Tucker-Drob.
Is the schooling effect on IQ causal?
There is credible causal evidence because multiple natural-experiment designs exploit policy changes or school-entry rules that alter educational exposure for reasons partly independent of prior ability. Causal interpretation remains conditional on design assumptions and on the precise outcome measured.
Does finishing high school raise IQ?
Longer formal education has a positive average effect on cognitive test scores in the studied settings, but finishing high school involves both time spent learning and the selection and credential effects associated with completion. A universal numerical effect for earning a diploma cannot be inferred from the meta-analytic per-year range.
Does university education increase intelligence?
University study can develop specialized knowledge and intellectual skills, but the causal IQ effect of university attendance is harder to isolate because participation is strongly selected. Evidence from compulsory-schooling reforms cannot be transferred mechanically to college.
Can adults improve their IQ through education?
Some education-associated differences remain detectable when cognitive outcomes are measured later in adulthood. Adult learning can improve specific knowledge and skills; the extent to which a particular adult program increases a standardized full-scale IQ score depends on direct evidence for that program.
Does homeschooling have the same IQ effect as conventional school?
The quoted 1–5-point estimate does not compare homeschooling with conventional schooling. Those formats differ in instruction, context, and selection. Comparable causal research with validated cognitive outcomes would be needed to estimate any difference.
Do school grades and IQ measure the same thing?
No. Academic grades and achievement tests assess performance in particular educational contexts, while IQ tests attempt to summarize specified cognitive abilities relative to norms. They are related but distinct measurements, and changes in one cannot automatically be expressed as changes in the other.
Does education affect fluid intelligence or just crystallized intelligence?
The 2018 synthesis reported positive effects on both categories of cognitive tests. Whether all improvements reflect latent general ability or more specific processes remains an open modeling question.
If IQ is partly heritable, how can education influence it?
Heritability describes variation in a particular population and context; it does not mean a characteristic is immutable in individuals. Genetic influences and environmental effects can coexist. Evidence that schooling changes average scores is compatible with meaningful genetic contributions to individual differences.
Are there harms in using IQ gains as the main goal of education?
A narrow score goal can miss literacy, social participation, specialist knowledge, creativity, health, and other outcomes that educational systems serve. Measurement must be used validly, and an IQ change should be interpreted as one possible research outcome rather than a complete judgment of students or schools.
Conclusion
Formal education is one of the clearest studied environmental influences on cognitive test performance. The convergence of schooling-policy reforms, school-entry cutoffs, and longitudinal adjustment makes the effect more credible than a simple education-IQ correlation. The landmark meta-analysis estimates average benefits in the approximate range of one to five standardized IQ points per additional year, with substantial variation across methods and outcomes.
The deepest interpretive issue is what those points represent. Schooling can develop knowledge, reasoning strategies, test-relevant competence, and other specific cognitive processes. Higher observed IQ scores are real findings, while any claim about change in latent g must be tested with an appropriate model and multiple measures. Good research keeps these levels of analysis distinct and studies how durable skills emerge from educational experience.
Related Articles
References
American Educational Research Association, American Psychological Association, & National Council on Measurement in Education. (2014). Standards for Educational and Psychological Testing. https://ncme.org/resources/books/testing-standards/
Brinch, C. N., & Galloway, T. A. (2012). Schooling in adolescence raises IQ scores. Proceedings of the National Academy of Sciences, 109(2), 425–430. https://doi.org/10.1073/pnas.1106077109
Cahan, S., & Cohen, N. (1989). Age versus schooling effects on intelligence development. Child Development, 60(5), 1239–1249. https://pubmed.ncbi.nlm.nih.gov/2805901/
Cappellari, L., Checchi, D., & Ovidi, M. (2023). The effects of schooling on cognitive skills: Evidence from education expansions. IZA Discussion Paper No. 15876. Working paper. https://www.iza.org/publications/dp/15876/the-effects-of-schooling-on-cognitive-skills-evidence-from-education-expansions
Donenfeld, J., Mudundi, M., Blaser, E., & Kaldy, Z. (2025). School changes minds: A meta-analysis shows that schooling modestly improves children's executive functions. Journal of Experimental Child Psychology, 262, 106371. https://doi.org/10.1016/j.jecp.2025.106371
Ritchie, S. J., Bates, T. C., & Deary, I. J. (2015). Is education associated with improvements in general cognitive ability, or in specific skills? Developmental Psychology, 51(5), 573–582. https://doi.org/10.1037/a0038981
Ritchie, S. J., & Tucker-Drob, E. M. (2018). How much does education improve intelligence? A meta-analysis. Psychological Science, 29(8), 1358–1369. https://doi.org/10.1177/0956797618774253
Sala, G., Aksayli, N. D., Tatlidil, K. S., Tatsumi, T., Gondo, Y., & Gobet, F. (2019). Near and far transfer in cognitive training: A second-order meta-analysis. Collabra: Psychology, 5(1), 18. https://doi.org/10.1525/collabra.203