Does One Year of Education Raise IQ Test Performance?

What counts as higher intelligence-test performance?

A higher result can mean more correct answers on the items taken, a higher score in one reasoning domain, or a different comparison with a reference group. These claims are not interchangeable. Before asking whether education raised performance, identify what changed. A raw score is observed item performance, such as the number answered correctly under the test’s scoring rules. A domain score summarizes performance in a defined area; its meaning depends on the test’s design. A norm-referenced score interprets performance against a specified comparison group. The Standards for Educational and Psychological Testing explains that norms and score interpretations need a relevant reference population and evidence supporting the intended use. A scaled score or composite transforms or combines results. A claim about general intelligence requires further evidence about what the test measures. None of these scores defines a person’s identity, worth, diagnosis, or fixed potential. Test IQ Free’s assessment covers pattern, number, and spatial reasoning and gives a provisional estimated IQ range. The site states that this estimate is not a validated IQ measurement. Treat it as feedback from this assessment, not as an established comparison with a population. On a retest, describe the raw or domain change first. A higher result alone establishes neither a population rank nor a change in general intelligence.

Sources: Standards for Educational and Psychological Testing (2014); Free IQ Assessment: 50 Adaptive Questions, Estimated IQ Range

How do researchers separate schooling's effect from selection?

A simple comparison between people with more and less schooling cannot show that education itself caused a difference in test performance. The groups may already have differed before their schooling diverged. In How Much Does Education Improve Intelligence?, the authors describe this selection problem: earlier test performance predicts later educational attainment, so the association can run in both directions. Ability may affect time in school, while schooling may also affect later scores. These shared influences make a raw comparison difficult to interpret.

The meta-analysis combined three quasi-experimental design families. These use real-world variation to build a more credible comparison than correlation, without randomly assigning people to years of schooling. One family follows people tested before and after differences in education, adjusting the later-score comparison for earlier test performance. It asks whether people with similar measured starting performance but different subsequent schooling differ later. Other differences may remain.

A second family uses policy changes that altered required schooling. The policy creates variation less directly tied to each pupil's prior characteristics. Researchers compare affected cohorts or places with less-affected ones, relying on assumptions about what would have happened without the reform. A third family compares students born close to a school-entry age cutoff. Children on either side may be similar in age and background but receive different amounts of schooling by a test date. This depends on the cutoff not sorting children in another relevant way.

Each design estimates an average effect under its assumptions, not an outcome promised to every learner. In a policy study, the estimate is most informative about people whose schooling changed because of the policy; it need not describe people who would have chosen the same education either way. The designs differ in population, age, tests and how schooling changes.

Schooling in Adolescence Raises IQ Scores, the Norwegian reform study, shows why the source of extra schooling matters. The reform raised compulsory schooling from seven to nine years, introduced a unified middle-school structure and standardized the minimum academic curriculum. The authors note that interpreting their instrumental-variable estimate as a duration-only effect requires school length to be the exclusive route from reform to scores. Because curriculum and organization changed too, the comparison may capture the reform package rather than time alone. This calls for a precise description of the result, not its dismissal.

When reading an education-effect claim, ask what generated the extra schooling, who changed exposure, and what assumptions connect that change to later test results. Then check whether the study's outcome and population match your question. This is more useful than treating a correlation or an average from a different design as an individual prediction.

Sources: How Much Does Education Improve Intelligence? A Meta-Analysis; Schooling in Adolescence Raises IQ Scores

What does the meta-analysis's 1–5 point conclusion mean?

The “1 to 5 IQ points” finding is a range of average estimates from different research designs, not a prediction interval for an individual. In “How Much Does Education Improve Intelligence? A Meta-Analysis,” the researchers combined 142 effect sizes from 42 datasets involving more than 600,000 participants. They examined studies that adjusted for earlier intelligence, used compulsory-school policy changes, or compared students around school-entry age cutoffs. Across those approaches, the evidence pointed in the same direction: additional education was associated with higher average performance on cognitive tests. But the estimates did not converge on one universal number. The authors first estimated each design family separately using random-effects models, which allow the underlying effect to vary across studies. For studies controlling for prior intelligence, the estimate was 1.197 points per additional year. For studies using policy changes, it was 2.056 points. For school-age cutoff studies, it was 5.229 points. A model combining estimates from all three families gave an average of 3.394 points per year. Researchers rescaled outcomes to a standardized IQ scale (mean 100, standard deviation 15) to compare results reported in different units; this does not make the tests or populations identical. The spread should not be read as a ranking in which one method proves that schooling raises scores by five points and another proves it raises scores by only one. Each design creates its comparison differently and relies on different assumptions. The families adjust for earlier performance, use external changes in required schooling, or compare students around school-entry rules. They address selection differently and can involve different ages, tests, and groups whose schooling changed. The reported estimate therefore describes an average for evidence assembled under a particular design, not the expected outcome for every student who completes another year. There was also meaningful variation among studies within each design family. The authors report statistically significant heterogeneity in all three separate models and in the combined analysis. They tested age at testing and outcome type as possible explanations, but heterogeneity remained. The positive pattern appeared across distinct quasi-experimental approaches, but the pooled average does not identify which operations improved or predict a detectable change for any learner. The result is strongest when stated at the level the evidence supports: across the studies included, additional education had a positive average effect on cognitive-test performance, with design-specific estimates ranging from about one to five standardized IQ points per year. It is not a conversion rule for a course, a personal retest, or an online score. It helps explain why education can affect measured performance on average. They should not add 3.394 points to their own score, treat the range as the likely gain for them, or infer from it alone that every aspect of general intelligence changed. Which tasks changed is a separate question that depends on the outcome measures used.

Sources: How Much Does Education Improve Intelligence? A Meta-Analysis; How Much Does Education Improve Intelligence? A Meta-Analysis

What exactly did Norway's reform change?

The Norwegian result is best understood as an effect of a particular school reform, not as a universal conversion from one extra year of education to IQ points. In municipalities adopting the policy between 1955 and 1972, compulsory schooling expanded from seven to nine years. Because the starting age remained seven, the change mainly affected adolescents around ages 14 to 16. The study, “Schooling in Adolescence Raises IQ Scores,” then examined cognitive test scores recorded at about age 19 among Norwegian men born from 1950 through 1958. Its outcome was later performance on a military-administered test, not a direct measure of every kind of thinking or a forecast for an individual student. The reform also changed what schooling looked like. It created a unified middle school for grades seven through nine and standardized the minimum academic curriculum at that level. Before the change, lower-secondary options differed: access to academically oriented schooling could depend on location, while other local programs offered less standardized practical or vocational courses. The policy therefore altered duration, organization, and academic content together. That matters when interpreting the number often quoted from this study, because the intervention was not simply the same course extended by a year. The paper reports two related estimates that answer different questions. Its difference-in-differences analysis compares changes in scores for cohorts in adopting municipalities with changes in municipalities not yet adopting it. The estimated average effect of the reform itself was an increase of 0.6 IQ points. Separately, the instrumental-variable analysis used exposure to the reform to estimate schooling's effect: the reform increased schooling by an average of 0.16 years, and under the model's assumptions the authors translated the score effect into an estimate of 3.7 IQ points per additional year. That 3.7 figure is not the observed gain from every person's extra year; it is a ratio-based estimate tied to the reform's effect on schooling and its affected population. The key qualification is explicit in “Schooling in Adolescence Raises IQ Scores.” The instrumental-variable interpretation assumes the reform affected scores only through the amount of schooling. Yet the authors note that a standardized curriculum and new school structure could also have changed the quality or nature of instruction. They therefore treat the reform-level estimate as meaningful even if the duration-only interpretation is uncertain. The evidence supports the conclusion that this Norwegian adolescent reform raised later test performance among affected male cohorts. It cannot promise the same change from a present-day adult course, a different school system, or an unrelated test; those settings differ in learners, content, and measured outcome.

Sources: Schooling in Adolescence Raises IQ Scores; Schooling in Adolescence Raises IQ Scores

Can the score rise while general ability does not?

Yes. An intelligence-test score can rise because education improves particular skills sampled by the test, without a corresponding rise in the broad common factor estimated across a varied battery. One study found this pattern, but cannot establish that education never affects general ability. The general factor, often written as g, is a statistical summary of variance that diverse cognitive tests share. It is inferred from how scores covary. If education improves a skill measured by some tasks, those scores may rise without a matching change in the pattern shared across the battery. An overall score combines specific and shared performance. “Is Education Associated With Improvements in General Cognitive Ability, or in Specific Skills?” analyzed 1,091 members of the Lothian Birth Cohort 1936. Participants took an intelligence test at about age 11 and completed ten varied cognitive tests at about age 70. Structural-equation models compared whether years of education were associated with later results through g, directly with particular tests, or both. The researchers adjusted for childhood intelligence, measured before educational duration differed in this cohort. The model with links from education to particular test scores, but no path from education to g, fit better than models that included an education-to-g path. In that model, education was associated with seven of the ten test scores. The authors interpreted the pattern as specific skill associations rather than evidence of a general-factor increase. This is a model comparison in one cohort, not a direct observation of an individual’s g. It supports a narrower reading of score gains: performance on tested tasks can improve without those gains transferring broadly. This finding does not cancel “How Much Does Education Improve Intelligence? A Meta-Analysis,” which combined quasi-experimental studies and found average positive effects on cognitive-test performance. The analyses ask different questions. The meta-analysis synthesizes whether additional education changes scores across its included study outcomes. The Lothian analysis asks how education’s association is distributed across ten later-life tests and their shared factor in one cohort. Score changes can be detected without estimating a latent general factor. Together, the findings support average gains on some intelligence-test outcomes while leaving open whether those gains represent a broad change in g. The Lothian result is limited. Schooling ended decades before the later-life battery. The authors note maturation, work opportunities, other experiences, and cognitive aging could accumulate during that interval and affect skills differently. Adjusting for childhood intelligence cannot isolate education from every later influence. They also leave open an education-related general effect measurable earlier in life but harder to detect later. This cohort, its ten-test battery, and its broad measure of schooling bound the conclusion. A higher composite score is evidence that measured performance changed; it does not alone establish far transfer across all abilities in a battery or a change in g. The Lothian study is a reason to keep those claims separate, not proof that general ability cannot change.

Sources: Is Education Associated With Improvements in General Cognitive Ability, or in Specific Skills?; How Much Does Education Improve Intelligence? A Meta-Analysis

Why can years of schooling affect one cognitive domain more than another?

Education can strengthen operations and knowledge used often in instruction, so outcomes closely related to that learning may change more than unfamiliar or less curriculum-linked tasks. A Swedish study found a contrast between crystallized and fluid test outcomes under a specific school-calendar design. In “The Effect of Schooling on Cognitive Skills,” researchers used differences in the dates when Swedish male conscripts took a cognitive battery. Because test dates fell at different points in the school calendar, some examinees had received more school instruction before testing. The authors describe this variation as conditionally random, making it useful for estimating the effect of additional instruction within that setting. It concerns male conscripts in one education system, not random assignment of learners to different school durations. The reported outcomes differed by task. Ten additional days of school instruction were associated with an increase of about one percent of a standard deviation on crystallized tests. In this study, those tests assessed synonym knowledge and technical comprehension. The same abstract reports almost no effect from extra days outside school, and no increase in scores on fluid spatial and logic tests with additional school days. “Crystallized” here refers to learned knowledge and skills used in those tests; “fluid” refers to the spatial and logic tasks used in this battery. The labels describe measured outcomes, not perfectly separate mental systems. One plausible interpretation is that instruction gave students practice with knowledge or operations closer to the crystallized test content. That is a synthesis about task alignment, not a mechanism the study isolated: the accessible report does not show that familiarity with particular test items caused the result. Nor does the fluid-task finding establish that education cannot affect fluid reasoning. It says that, in this design, added school days did not produce a detected increase on those particular spatial and logic measures. Other ages or measures could yield different results. This domain contrast fits the broader point that education effects depend partly on what an assessment asks people to do. A score built from vocabulary or technical comprehension can respond to learning that a spatial transformation task does not directly sample. The Swedish estimate therefore supports a bounded conclusion about average performance on the measured crystallized tests under the study conditions, alongside a null finding for its measured fluid tests. It cannot be converted into an expected gain for an individual or a different assessment. When a reasoning score changes, the relevant next question is which operations the test sampled: improvement on one operation is not evidence of transfer to another without outcome evidence for that other task.

Sources: The Effect of Schooling on Cognitive Skills

What does school-achievement evidence add?

A gain on a curriculum achievement test is not automatically a gain on broader cognitive-skill measures. In one study of eighth-grade students, random offers to attend oversubscribed charter schools improved mathematics achievement but produced no measured impact on the study’s cognitive-skill assessments. “Doing better at school” and “scoring higher on a cognitive test” are separate outcomes. In “Cognitive Skills, Student Achievement Tests, and Schools,” researchers compared standardized achievement scores with cognitive-skill measures for 1,367 eighth-grade students attending traditional, exam, and charter public schools. Achievement scores and changes in those scores were correlated with cognitive measures. Yet after accounting for fourth-grade test scores, differences in cognitive skills across schools were negligible. A lottery-based comparison among applicants to oversubscribed charter schools found positive math-achievement impacts from random enrollment offers, but no impact on cognitive skills. The accessible publisher abstract gives no effect size, so the result should not be assigned a numerical magnitude. The two outcome types ask different questions. A curriculum achievement test asks what students can demonstrate in taught subjects, such as mathematics. A cognitive assessment samples selected skills using its own tasks. The measures may overlap: the study reports correlations between achievement and cognitive measures. But overlap does not make them interchangeable. An intervention that improves taught mathematics provides evidence of a math-achievement gain. It supports a claim about a cognitive measure only if that measure also changes; a broader claim about general intelligence requires more evidence about what the test represents. This is counterevidence to the assumption that every school improvement must raise every kind of test performance. It does not overturn evidence that additional education can raise average intelligence-test scores in other settings. The study concerns school attendance and school quality among eighth-grade lottery applicants in particular US public-school contexts, not the isolated effect of adding one year of compulsory education. Its null cognitive finding is limited to the outcomes and sample measured. It cannot establish that schooling never affects cognitive skills or that another age group, program, or test would show the same pattern. For a learner, keep the outcome label attached to the result. If grades or a subject achievement score improve, describe that gain directly. Do not translate it into an IQ gain unless a suitable cognitive measure supports that interpretation. Even then, a measured change describes performance on that instrument under its conditions; it does not by itself show what caused the change or establish a shift in general intelligence.

Sources: Cognitive Skills, Student Achievement Tests, and Schools; Is Education Associated With Improvements in General Cognitive Ability, or in Specific Skills?

Does a one-year average predict my before-and-after change?

No. A population average cannot tell one person how much their score should change after a year of education. A retest can describe changed performance, but only if results are comparable in what they measure and how they are scored. Even then, the difference alone does not show that schooling caused it. First ask whether the same instrument and domains were used both times, with similar instructions, timing and administration. Different forms need to be formally linked for their scores to be compared. If the test or scoring changed, a higher result may combine changed performance with changed measurement. The Standards explain that score meaning depends on interpretation and supporting evidence. Then ask what kind of score moved. As defined earlier, a raw score records observed item performance; a norm-referenced score compares it with a specified reference group. A change in raw answers does not by itself say whether someone moved relative to others. That comparison depends on whether the reference population and scoring scale are suitable on both occasions. Validity, as the Standards explain, concerns whether evidence supports an interpretation for its intended use. It is not a general badge attached to a test name. Precision matters too. Test scores include measurement error, so a small difference may be difficult to distinguish from uncertainty without precision information. A point estimate without that information is not an exact measure of change. A displayed band is not necessarily an uncertainty interval: Test IQ Free describes its 50-question result as a provisional educational estimate from an unvalidated conversion, without a validated adult reference population, and says its fixed display band is not a confidence interval. Treat it as low-stakes practice, not a way to calculate personal IQ gain. Repeated exposure adds an explanation. Familiarity with similar item formats, instructions or solution approaches could improve performance on those tasks. That is a possible practice effect, not proof of what happened in one retest. Success on familiar items also does not establish transfer to different tasks or a broad change in general ability. Researchers address alternative causes through comparisons; an individual before-and-after difference has no counterfactual. A careful report says what changed and stops there: “the number of correct responses in this domain increased on the second administration.” It can name the test form, conditions, scoring basis and known limits to comparability. It should not say, “school raised my IQ by the meta-analysis average.” The review’s 1-to-5-point summary describes average estimates across research designs, not a personal target or conversion rule. If a result could affect placement, diagnosis or another consequential decision, ask a qualified examiner whether instruments, norms and precision support comparison for that purpose. A browser assessment cannot settle it. For curiosity, the free test offers reasoning practice and worked explanations; treat its estimated range as provisional. Report an observed change only when the comparison supports it, and leave its cause open unless evidence identifies one.

Sources: Standards for Educational and Psychological Testing (2014); Free IQ Assessment: 50 Adaptive Questions, Estimated IQ Range

Can you assume that two years double the change?

No. The meta-analysis estimated average effects associated with an additional year in studies that could support a particular comparison. It did not show that each successive year produces the same increment, or that two years can be converted into twice one year’s result. The reported range of roughly 1 to 5 standardized IQ points is a synthesis across different designs and outcomes, not a schedule for accumulating points in an individual. A per-year estimate and a multi-year dose-response answer different questions. A study may estimate an average slope by relating an education contrast to later test scores, then express that estimate per year. That does not necessarily follow the same people through successive years, establish that the slope stays constant, or show what happens after a particular amount of schooling. In *How Much Does Education Improve Intelligence? A Meta-Analysis*, the authors explicitly say they could not determine whether effects add across years. They raise diminishing marginal benefits and a possible plateau as questions, while noting that they knew of no rigorous quasi-experimental study directly addressing them. This is an unresolved evidence gap, not evidence that benefits must diminish or stop. The design matters to what “another year” means. Policy studies use a rule or reform that changes schooling for some people; their estimate is tied to those whose education changed under that policy and to its assumptions. The meta-analysis cautions that such instrumental-variable estimates may not generalize across the full range of education. In Norway, the reform expanded compulsory schooling from seven years to nine, but also created a unified middle school and standardized its minimum academic curriculum. *Schooling in Adolescence Raises IQ Scores* notes that curriculum or school quality may have contributed to the result. Its reform is therefore not a clean test of duration alone, and its estimate cannot establish that a second year would duplicate a first-year effect in another setting. To answer the cumulative question more directly, evidence would need credible comparisons across clearly specified amounts of schooling, with prior ability considered and outcomes measured on varied cognitive tasks. It would also need to show whether the estimated change continues, weakens, or levels off as exposure accumulates. These features would make evidence more informative; they are not findings already established by the cited studies. A collection of one-year estimates from different cohorts, policies, tests, and ages cannot simply be added as though it followed one person on an unchanged scale. So do not multiply the 1-to-5-point summary by the number of years studied or use it to promise a future gain. The defensible conclusion is narrower: additional education can raise average performance on some intelligence-test outcomes, while effects across multiple years remain unsettled. That boundary matters when interpreting a retest: a research average is not a personal forecast, and a score change does not by itself show a linear change in general intelligence.

Sources: How Much Does Education Improve Intelligence? A Meta-Analysis; Schooling in Adolescence Raises IQ Scores

What is the responsible next action?

If your goal is a low-stakes reasoning challenge, take the free 50-question assessment and review its worked answers. It covers pattern, number, and spatial reasoning; use feedback to choose an operation. Treat the displayed range as provisional: the conversion has no validated adult reference population, so the band is neither a normed comparison nor a confidence interval. The product page, “Free IQ Assessment: 50 Adaptive Questions, Estimated IQ Range,” calls it unvalidated. A repeat attempt may show different performance on similar tasks, but cannot establish that education caused a change in general intelligence. Familiarity with a format does not demonstrate transfer to other tasks.

If a result may affect placement, clinical care, or another consequential decision, use an assessment selected for that purpose and discuss its interpretation with a qualified examiner. The “Standards for Educational and Psychological Testing” treats validity as evidence supporting a proposed interpretation and use; it does not endorse a browser quiz. Ask which instrument and reference group support the comparison, and how score precision limits conclusions. Describe observed performance before assigning a cause. The paid report adds explanation and a practice plan, not validity.

Identify which operation changed, under what conditions, and what evidence supports the interpretation. For practice, try a task using that operation and check whether improvement extends beyond familiar examples. For consequential decisions, bring the assessment and its intended use to a qualified professional instead of converting a research average into a personal IQ gain.

Sources: Free IQ Assessment: 50 Adaptive Questions, Estimated IQ Range; Standards for Educational and Psychological Testing (2014)