Dyslexia and IQ

Posted on September 29, 2026 by Mahesh

Understanding IQ

Dyslexia and IQ: Is There a Link? What the Research Shows

Dyslexia is a reading and spelling difficulty that occurs at every level of IQ, and the IQ-gap rule was dropped as a US requirement in 2004 and from the IDA definition in 2025. See what IQ tests show for readers with dyslexia, and what the genetics and myths add.

By Alden Cross, Intelligence Research WriterPublished September 29, 2026 · 12 min readReviewed by Aarav Vardhan, Editor-in-Chief

Dyslexia and IQ are separate things. Dyslexia is a difficulty with accurate, fluent word reading and spelling, not a measure of intelligence, and the UK Rose Review states that it “occurs across the range of intellectual abilities.” A person does not need a high, average or any particular IQ to be dyslexic. The old requirement for a gap between IQ and reading was dropped from US special-education rules in 2004, and in October 2025 the International Dyslexia Association replaced its 2002 definition, which had called the difficulty “unexpected” relative to other abilities. This guide covers what changed, what IQ tests show for readers with dyslexia, whether dyslexia lowers IQ scores, what genetics adds, and two popular myths.

Definition: difficulties with accurate and/or fluent word reading and spelling, on a continuum of severity, not a yes-or-no condition.
IQ needed? No. The UK Rose Review says dyslexia occurs across the range of intellectual abilities, and under US IDEA 2004 states “must not require” a severe IQ-achievement discrepancy.
Lower IQ? Not as such, but group profiles are uneven: in Pearson data for 24 children with a reading disorder, working memory averaged 89.8 and verbal comprehension 91.4, against 106.6 and 104.1 in matched controls.
How common? It depends on the cut-off: about 7% at 1.5 standard deviations below the mean, and roughly one in ten children in a 2022 genetics paper.
Genes: a 2022 study found 42 genome-wide significant loci; polygenic scores explained up to 6% of the variance in reading traits.

Three cards summarise dyslexia and IQ. No IQ threshold: the UK Rose Review says dyslexia occurs across the range of intellectual abilities, and US states may not require an IQ-reading gap. An uneven profile: in Pearson WISC-V data for 24 children with a reading disorder, working memory averaged 89.8 against 106.6 in matched controls. Genes: a 2022 study of 51,800 self-reported cases found 42 loci.

What is dyslexia, and what changed in 2025?

For two decades most articles quoted the International Dyslexia Association’s 2002 definition, which said the difficulty was “often unexpected in relation to other cognitive abilities and the provision of effective classroom instruction.” That is no longer the current wording. The IDA board adopted a revised definition on 22 October 2025, and the text no longer contains the word “unexpected.” The IDA page keeps the 2002 version only as a historical reference. In the UK, the British Dyslexia Association now presents a 2025 definition agreed by a panel of experts through the Delphi method, which accepted 42 statements. Neither new definition requires an IQ gap, though the UK wording still relates literacy to a person’s “other attainments.”

Dyslexia definitions compared
Source and year	What it says
International Dyslexia Association, 2002	A neurobiological learning disability with accurate and/or fluent word recognition problems that are “often unexpected in relation to other cognitive abilities”
International Dyslexia Association, 22 October 2025	Difficulties in word reading and/or spelling involving accuracy, speed or both; a continuum of severity; difficulties persist even with instruction that is effective for peers. Phonological and morphological difficulties are “common but not universal.” No mention of unexpectedness.
Rose Review, UK, 2009	“A learning difficulty that primarily affects the skills involved in accurate and fluent word reading and spelling”; occurs across the range of intellectual abilities; a continuum with no clear cut-off points
British Dyslexia Association, 2025 Delphi definition	“A set of processing difficulties that affect the acquisition of reading and spelling,” with literacy weak in relation to age, teaching and other attainments
DSM-5, 2013	Specific learning disorder with impairment in reading; academic skills well below the average range on appropriate tests; not better explained by other disorders. Dyslexia stays in the descriptive text.

The debate is still live. Snowling, Hulme and Nation (2020) note that the failure to find qualitative differences between poor readers with and without an IQ gap led the discrepancy definition to “fall from favour,” but that loosening the criteria also caused “diagnostic confusion.” Elliott and Grigorenko (2024) argue that diagnosing by the IDA definition then in force “often results in scientifically questionable diagnoses.” Treat any 2002-era definition, and any IQ-gap rule of thumb, as out of date.

Do you need a certain IQ to be dyslexic?

No, and the history explains why people think you do. A 1968 World Federation of Neurology definition described children who fail to attain reading skills “commensurate with their intellectual abilities.” Rutter and Yule’s 1975 paper on Isle of Wight data seemed to support the discrepancy idea, on the strength of a “hump” in the distribution of reading achievement. When US regulations for Public Law 94-142 arrived in 1977, specific learning disability was identified by a significant discrepancy between IQ and achievement. Later studies questioned the original finding because it applied no exclusion criteria, so its no-discrepancy group included many children with brain injury or intellectual disability.

The evidence then accumulated against the gap rule:

Stuebing and colleagues (2002) meta-analyzed 46 studies of the IQ-discrepancy classification and found “little evidence in support of the validity” of it.
Fletcher and colleagues (1994) compared discrepancy and low-achievement definitions in 199 elementary school students and found no support for the discrepancy version; Stanovich and Siegel (1994) reached a similar conclusion with 907 children.
Stuebing and colleagues (2009) pooled 22 studies and found IQ explained very little of how well children responded to reading instruction (R² = 0.03).
IDEA 2004 (34 CFR 300.307) says states “must not require” a severe discrepancy between intellectual ability and achievement, and must permit a process based on response to research-based intervention.

Two nuances keep this from being an absolute. US rules still exclude reading problems that are primarily the result of intellectual disability, and evaluators may still look for a pattern of strengths and weaknesses relative to age, grade-level standards or intellectual development. And a 2012 Lancet review by Peterson and Pennington found that research does not support the external validity of the distinction between age-based and IQ-based definitions, yet recommends that children meeting either one be identified and treated. Dyslexia is also not all-or-nothing. In Shaywitz’s 1992 study of 414 Connecticut children, which defined dyslexia by an IQ-reading discrepancy score, it “occurs in degrees,” and the model predicted, and the data indicated, that only 7 of 25 children classed as dyslexic in grade 1 would be classed as dyslexic in grade 3. For the wider picture of how learning disabilities relate to scores, see our guides to low IQ, intellectual disability and learning disability and ADHD, autism and dyslexia test scores.

Does dyslexia lower IQ scores?

Not as a single number, but the profile is often uneven. The clearest public data come from Pearson’s technical report on the WISC-V in 24 children aged 6 to 16 with a specific learning disorder in reading. Their averages, next to matched controls, show where the gaps sit:

WISC-V index scores: 24 children with a reading disorder versus matched controls
Index	Reading-disorder group	Matched controls
Verbal Comprehension	91.4	104.1
Visual Spatial	98.4	106.6
Fluid Reasoning	97.6	105.7
Working Memory	89.8	106.6
General Ability Index	95.1	107.2
Naming Speed Literacy	80.0	103.8

Read this with care. It is a small group, the controls averaged 104 to 107, so the gaps overstate the distance from the norm of 100, and the report excluded the processing-speed subtests, so there is no Full Scale IQ. A much larger clinical sample points the same way: Toffalini and colleagues (2017) studied 1,049 Italian children diagnosed with a specific learning disorder on the WISC-IV and found that all subgroups shared weaknesses in working memory and processing speed, with the working-memory and processing-speed composite about one standard deviation below the general ability composite. Among adults, Callens and colleagues (2012) compared 100 Dutch-speaking students with dyslexia and 100 without learning disabilities and found no significant difference in fluid intelligence, and slightly lower crystallized scores, which they attributed to retrieving verbal information from long-term memory. The authors also warn that university samples may be selected by admission criteria. Our guides to WISC-V index scores, full-scale IQ versus index scores and reading an IQ test report explain why one number can hide an uneven profile, and the Woodcock-Johnson guide covers the achievement tests that measure reading directly.

Can reading itself change IQ scores?

It may, which is one reason group averages need care. Ferrer and colleagues (2010) reported that in typical readers reading and IQ develop in a dynamically linked way over time, a pattern “not perceptible in dyslexic readers,” suggesting the two develop more independently. In a general-population twin sample of 1,890 identical pairs tested from age 7 to 16, Ritchie, Bates and Plomin (2015) found that earlier within-pair differences in reading were associated with later differences in intelligence, not explained by reading exposure and not limited to verbal tasks. That study did not examine dyslexia, and Stanovich’s 1986 “Matthew effect” idea that reading experience feeds cognitive growth has been contested, but together they raise the possibility that less reading practice could hold back vocabulary-heavy scores over time. Our guides to reading and IQ and verbal versus non-verbal scores cover the same ground from other angles.

What do genes have to do with it?

A great deal, and it is polygenic. Family studies put heritability at up to 70%. A genome-wide study by Doust and colleagues (2022) in Nature Genetics, compared 51,800 adults who reported a dyslexia diagnosis with 1,087,070 controls and found 42 independent genome-wide significant loci, 15 of them in genes linked to cognitive ability or educational attainment and 27 new. Polygenic scores explained up to 6% of the variance in reading traits, so they are nowhere near a diagnostic test. Genetic correlations with cognitive measures were negative but partial, ranging from −0.19 for childhood and adolescent performance IQ to −0.50 for adult verbal-numerical reasoning, none beyond 0.5. Of 75 previously reported dyslexia associations, none reached genome-wide significance in the new analysis. The authors note the limits: self-reported diagnosis in a consumer-genetics cohort, with potential collider bias. Our guide to whether IQ is genetic covers the parallel debate for intelligence.

Your own number

Where would your own score land?

Take the IIF-certified assessment and get your score with the scale it was measured on, the percentile it corresponds to and the confidence range around it — the three figures most online tests leave out.

Find your IQ score now! →

Secure & encryptedInstant results10–20 minutes

Are dyslexic people more intelligent, or more likely to be entrepreneurs?

No source we reviewed reports higher IQ in dyslexia. The British Dyslexia Association says many dyslexic people show strengths in reasoning and in visual and creative fields, but its page gives no citation for that sentence, and Callens found no difference in fluid intelligence. The most repeated statistic is that 35% of US entrepreneurs are dyslexic. It traces to Julie Logan’s 2009 survey, in which 102 entrepreneurs and 37 corporate managers responded, a 7.0% response rate. Of the entrepreneurs, 36 (35%) scored four or more traits on a dyslexia checklist, against 3 of 37 managers (8%). A checklist is a trait screen, not a diagnosis, and Logan cautioned that the corporate sample was too small to test for an effect. The result is an association in a small, low-response, self-selected sample, and it says nothing about IQ or about whether dyslexia causes business success.

Claims that Einstein, Edison or da Vinci were dyslexic are posthumous labels. A diagnosis rests on standardized reading and spelling tests that cannot be given to historical figures, and a 1987 paper in the Journal of Learning Disabilities asked whether such figures were really learning disabled at all. For a related myth about a different condition, see ADHD and IQ.

How is dyslexia identified, and where does IQ fit?

Assessment focuses on word reading, spelling and decoding, plus phonological awareness and rapid naming. In the Pearson data above, the largest gap of all was on naming speed for literacy (80.0 against 103.8). Peterson and Pennington note that diagnosis “requires setting a somewhat arbitrary cutoff on a continuous variable,” which is why prevalence depends on the definition: about 7% at a cut-off of 1.5 standard deviations below the mean, and roughly one in ten children in Doust’s genetics paper. US law lists dyslexia among the conditions that can be a specific learning disability; in the UK, the British Dyslexia Association reads the Equality Act 2010 as covering dyslexia as a disability, although whether a particular person is covered depends on the Act’s “substantial” and “long-term” test. Only a qualified assessor can diagnose, and an online IQ quiz cannot. Timed, text-heavy online tests were not evaluated in the research we reviewed, so a low score on one is not a finding. IQ is a weak predictor of how children respond to reading instruction (the 2009 meta-analysis above), which is why identification leads to teaching rather than to an IQ verdict; for what persistence traits add on top, see our guide to grit vs IQ. The idea of being gifted and having a learning difficulty at once (twice-exceptional) is covered in our guide to high IQ and ADHD, and IQ testing for children covers the assessments themselves.

Quick answers

Does dyslexia lower IQ? Not as such. Group averages on some index scores, especially working memory and verbal comprehension, run lower, but individuals vary widely.
Can you be dyslexic with a high IQ? Yes. The Rose Review says dyslexia occurs across the range of intellectual abilities.
Do you need an IQ score to be diagnosed? No. US rules say states must not require a severe IQ-achievement discrepancy, and UK definitions have no IQ criterion.
Did the definition of dyslexia change? Yes. The IDA adopted a new definition on 22 October 2025 that drops “unexpected,” and the British Dyslexia Association now presents a 2025 Delphi definition.
How common is dyslexia? About 7% at a cut-off of 1.5 standard deviations below the mean, roughly one in ten children in a 2022 study; the figure depends on the definition.
Is dyslexia genetic? Partly. Family studies suggest heritability up to 70%, and a 2022 study found 42 associated loci, but polygenic scores explain up to 6% of variance in reading traits.
Are 35% of entrepreneurs dyslexic? No. 36 of 102 survey respondents scored four or more traits on a checklist, at a 7.0% response rate.

The bottom line

Dyslexia is a reading-and-spelling difficulty that occurs at every level of IQ. The idea that you need an IQ gap to qualify has been dropped from US special-education rules and, as of October 2025, from the International Dyslexia Association’s own definition. What IQ tests can add is a profile: in the data above, lower working-memory and verbal scores than reasoning scores, with wide variation between individuals. That is a description of strengths and weaknesses, not a verdict on intelligence, and it cannot come from an online quiz.

Sources

International Dyslexia Association: Definition of dyslexia (adopted 22 October 2025), with the 2002 definition as a historical reference.
British Dyslexia Association: What is dyslexia? (Delphi definition, 2025).
Carroll, J.M. et al. (2025): Toward a consensus on dyslexia: findings from a Delphi study, Journal of Child Psychology and Psychiatry, 66(7), 1065-1076.
Snowling, M.J., Hulme, C. and Nation, K. (2020): Defining and understanding dyslexia: past, present and future, Oxford Review of Education, 46(4), 501-513.
34 CFR 300.307: Specific learning disabilities (IDEA 2004 regulations).
Stuebing, K.K. et al. (2002): Validity of IQ-discrepancy classifications of reading disabilities: a meta-analysis, American Educational Research Journal, 39(2), 469-518.
Stuebing, K.K. et al. (2009): IQ is not strongly related to response to reading instruction: a meta-analytic interpretation, Exceptional Children, 76(1), 31-51.
Peterson, R.L. and Pennington, B.F. (2012): Developmental dyslexia, The Lancet, 379(9830), 1997-2007.
Shaywitz, S.E. et al. (1992): Evidence that dyslexia may represent the lower tail of a normal distribution of reading ability, New England Journal of Medicine, 326(3), 145-150.
Pearson (2016): Q-interactive Technical Report 13, WISC-V in children with specific learning disorder in reading and in mathematics.
Toffalini, E., Giofrè, D. and Cornoldi, C. (2017): Strengths and weaknesses in the intellectual profile of different subtypes of specific learning disorder, Clinical Psychological Science, 5(2), 402-409.
Callens, M., Tops, W. and Brysbaert, M. (2012): Cognitive profile of students who enter higher education with an indication of dyslexia, PLoS ONE, 7(6), e38081.
Ferrer, E. et al. (2010): Uncoupling of reading and IQ over time: empirical evidence for a definition of dyslexia, Psychological Science, 21(1), 93-101.
Ritchie, S.J., Bates, T.C. and Plomin, R. (2015): Does learning to read improve intelligence? Child Development, 86(1), 23-36.
Doust, C. et al. (2022): Discovery of 42 genome-wide significant loci associated with dyslexia, Nature Genetics, 54(11), 1621-1629.
Logan, J. (2009): Dyslexic entrepreneurs: the incidence; their coping strategies and their business skills, Dyslexia, 15(4), 328-346.
Adelman, K.A. and Adelman, H.S. (1987): Rodin, Patton, Edison, Wilson, Einstein: were they really learning disabled? Journal of Learning Disabilities, 20(5), 270-279.

Share this article

Know someone who keeps seeing these numbers quoted without the scale they were measured on? Send it to them.

Keep reading

All articles →

Grid rating how far dyslexia, ADHD and autism affect four IQ test domains: verbal comprehension, processing speed, working memory and matrix reasoning, which is largely spared

Taking a Test

ADHD, Autism and Dyslexia in IQ Testing

A test score is a measurement, and a measurement can be biased by the conditions under which it is taken. ADHD, autism and dyslexia each depress performance on particular kinds of item without lowering the reasoning underneath. Here is what actually interferes, condition by condition, and what to do next.

August 29, 20267 min read

Understanding IQ

Low IQ vs Learning Disability

A low IQ, an intellectual disability and a specific learning disability are three different things, and the UK uses "learning disability" for the second. What each requires, how each is assessed, the US and UK terms, and what to do if you are worried.

September 21, 20268 min read

Understanding IQ

Grit vs IQ

Grit is weakly to moderately linked to school achievement and overlaps almost completely with conscientiousness. See what the original grit studies, two large meta-analyses, the marshmallow test and a retracted test-motivation paper really show about grit vs IQ.

September 29, 202610 min read

Adoption and IQ

Posted on September 29, 2026 by Mahesh

Research & Evidence

Adoption and IQ: What Adoption Studies Show About Nature and Nurture

Adoption can raise average IQ by about 3 to 4 points in a large Swedish sibling study and by 8 to 20 in a French study of neglected children, yet adoptees come to resemble their birth parents. See what adoption and twin studies show, and what they cannot.

By Arjun Vardhan, News & Research WriterPublished September 29, 2026 · 12 min readReviewed by Aarav Vardhan, Editor-in-Chief

A bar chart shows four measured IQ gains from adoption, each against a different comparison group: 3.18 points in a Swedish half-sibling comparison, 4.41 points in a Swedish full-sibling comparison, 7.7 points for neglected French children adopted into lower-SES homes and 19.5 points for those adopted into higher-SES homes.

Adoption and IQ: does adoption raise a child’s IQ? Adoption studies show that the home a child grows up in can move average IQ, and how far depends on where the child starts: about 3 to 4 points in a large Swedish sibling study, and roughly 8 to 20 points in a French study of neglected children moved into better-off homes. But adopted children’s IQ also comes to resemble their birth parents’ more than their adoptive parents’ as they grow up, and identical twins reared apart correlate about .70 in adult IQ. Both findings hold at once: environment can shift the average level, especially for children in poor starting conditions, while genes account for more of the differences among people raised in ordinary, similar-quality homes. This guide sets out what each major study measured, what the numbers are, and what adoption designs cannot show.

Swedish siblings (2015): adopted-away men scored 4.41 IQ points higher than their home-reared full brothers (436 sibships); a half-sibling replication gave 3.18 points (2,341 sibships).
French adoptees (1999): neglected children with a pre-adoption IQ near 77 gained 7.7 points in lower-SES adoptive homes and 19.5 points in higher-SES homes by adolescence.
Colorado Adoption Project: adoptees resembled their adoptive parents slightly in early childhood but not at all by adolescence.
Twins reared apart (1990): an identical-twin IQ correlation of .69 on the Wechsler adult scale, with a warning not to extend it to extreme disadvantage.
Romanian institutions: at age 8, children placed in foster care averaged IQ 81 against 107 for never-institutionalized children.

Why do researchers use adoption studies?

An adopted child inherits genes from one family and grows up in another, which pulls apart two things that are tangled together in ordinary families. If adoptees end up resembling their birth parents, genes are doing the work; if they resemble their adoptive parents, or outscore siblings who stayed behind, the home is. Adoption designs have limits, and the biggest is that adoptive parents are screened and are on average better off than birth parents, a point we return to below. For the wider debate see our guides to whether IQ is genetic, twin studies and IQ and socioeconomic status and IQ.

Does adoption raise IQ? What the meta-analysis found

The broadest evidence is a 2005 meta-analysis in Psychological Bulletin by van IJzendoorn, Juffer and Klein Poelhuis, covering 62 studies and 17,767 adopted children. Its answer depends on the comparison group. Adopted children scored higher on IQ tests than the siblings or peers who stayed behind. They did not differ in IQ from their nonadopted peers or siblings in their new communities, though their school performance and language abilities lagged behind and more of them developed learning problems. In other words, adoption looks like a large improvement relative to what the child left behind, and like a normal starting point relative to peers in the new community.

A widely quoted range comes from Nisbett and colleagues (2012), who describe a 12-point to 18-point rise in IQ when children are adopted from working-class to middle-class homes. They add that the meta-analysis figure of 18 points included comparisons with extremely deprived institutional settings. Those are large numbers, and the designs below suggest they sit at the top of the range rather than being the typical case.

How big is the gain? Four measured numbers

The Swedish study uses national registers and gives the most cautious estimate. Kendler and colleagues (2015) used national registers to find male siblings where one was raised by biological parents and the other was adopted away, with IQ taken from the military conscription exam at age 18 to 20. Across 436 full-sibling sets the adopted brother scored 4.41 points higher (standard error 0.75); a replication with 2,341 half-sibling sets gave 3.18 (standard error 0.34). The authors call this smaller than differences in earlier, smaller studies, and note that the environmental difference between the adoptive and biological families was not especially large and that the adoptive families were screened by the adoption agency. The effect also ran both ways: children placed in the best-educated homes scored highest, while those placed in homes less educated than the family of origin “actually performed worse” than their siblings.

The French study shows a bigger effect at a lower starting point. Duyme, Dumaret and Tomkiewicz (1999) picked 65 children, from 5,003 adoption files, who had been abused or neglected in infancy and were adopted between ages 4 and 6, with a pre-adoption IQ below 86 (mean 77). By adolescence the mean IQ was 85 in adoptive families of lower socioeconomic status (SES) and 98 in higher-SES ones, gains of 7.7 and 19.5 points. One caution is ours, not the paper’s: children chosen for a low first score tend to score somewhat higher on retest by chance alone, the effect described in our guide to regression to the mean, so some of the 7.7 may reflect that statistical rebound rather than adoption itself. A classic 1989 French cross-fostering study of 38 children, as summarized in a 2018 review, found children born to higher-SES families and adopted into lower-SES homes averaged 107.5, against 103.6 for children born to lower-SES families and adopted into higher-SES homes, which suggests, on our reading, that birth background and adoptive home both mattered.

Four adoption gains, and what each compares
Study	Comparison	Gain
Kendler et al. 2015, Sweden (half-siblings)	Adopted-away men against home-reared half-brothers, 2,341 sibships, IQ at conscription	3.18 points
Kendler et al. 2015, Sweden (full siblings)	Adopted-away men against home-reared full brothers, 436 sibships	4.41 points
Duyme et al. 1999, France (lower-SES homes)	65 neglected children adopted at 4 to 6, pre-adoption IQ near 77; measured in adolescence	7.7 points
Duyme et al. 1999, France (higher-SES homes)	Same children, adopted into higher-SES families	19.5 points

Do adopted children come to resemble their birth parents or their adoptive parents?

The Colorado Adoption Project followed adopted children and their biological and adoptive parents for 20 years, alongside matched non-adoptive families. Plomin and colleagues reported in 1997 that adopted children resemble their adoptive parents “slightly in early childhood but not at all” in middle childhood or adolescence, while becoming more like their biological parents, to the same degree as children in non-adoptive families. A later analysis of the project’s sibling data by Petrill and colleagues (2004), covering 245 adoptive and 245 biological families with IQ measured from age 1 to 16, found considerable genetic influence at each age and modest shared-environmental influence, and reported that selective placement was negligible for the cognitive measures.

This fits a pattern documented in twin and adoption work: the share of IQ differences linked to genes rises with age. Plomin and Deary (2015) put heritability at about 20% in infancy, rising to perhaps 80% in later adulthood, with cross-sectional twin data from 11,000 pairs showing roughly 40% in childhood, 50% in adolescence and 60% in young adulthood. That is consistent with parental influence looking strongest in early childhood and weakest in adulthood. See also IQ scale by age and whether intelligence comes from mother or father.

The two findings do not contradict each other. A 2018 review by Sauce and Matzel puts it in one line: “the magnitude of a correlation is independent of changes in means.” Adoption can raise the average score of a group while genes still explain most of the differences among individuals within it.

What do twins reared apart show?

The Minnesota Study of Twins Reared Apart, reported by Bouchard and colleagues in Science in 1990, covered more than 100 sets of reared-apart twins or triplets, with the main analysis on 56 sets of identical twins. The IQ correlation was .69 on the Wechsler adult scale (48 pairs) and near .78 on other composites; three earlier studies gave .64 to .74. The authors concluded that about 70% of the variance in IQ was associated with genetic variation. Identical twins raised together, mostly children, correlated .86 across 34 studies, against .72 for the mostly adult reared-apart twins; the samples differ in age as well as in rearing, so the comparison is not clean.

The paper is also explicit about its limits. Few of the twins were reared in real poverty and none was intellectually disabled, so the estimate “should not be extrapolated to the extremes of environmental disadvantage,” and the findings “do not define or limit what might be conceivably achieved in an optimal environment.” The authors add that the evidence does not detract from the value of parenting, education and other interventions.

Your own number

Where would your own score land?

Find your IQ score now! →

Secure & encryptedInstant results10–20 minutes

What happens when the early environment is very poor?

The strongest evidence on severe deprivation comes from Romania. In the Bucharest Early Intervention Project, abandoned children were randomly assigned to foster care or to remain in institutions. By 42 and 54 months, children who stayed in institutions scored markedly below both never-institutionalized children and those moved to foster care, and the gains were largest for the youngest children placed. At age 8, Fox and colleagues (2011) tested 103 of the original 136 children on the WISC-IV: the foster-care group averaged 81.46, the care-as-usual group 76.16 and never-institutionalized children 107.00. The gap between foster care and care as usual was only marginal (p = .07), and the care-as-usual average rose between 54 months and age 8, partly because Romania began moving institutionalized children into foster homes. Even the foster-care group was about 25 points below never-institutionalized peers. Our guide to toxic stress and IQ covers the trial in more depth.

The English and Romanian Adoptees study followed children adopted into UK families after up to 43 months of severe deprivation. Sonuga-Barke and colleagues (2017) found that among those who had spent more than six months in an institution, rates of IQ below 80 were markedly higher than in UK-born adoptees at ages 6 (p = .0001) and 11 (p = .0016), but had fallen to normal rates by age 22 to 25 (p = .76). Other difficulties did not fade: autism-spectrum-like symptoms, disinhibited social engagement and inattention stayed higher, and about a fifth of that group (15 people) had no problems at any assessment. IQ can recover when the environment improves, but a score is not the whole picture. For more on adversity and thinking skills, see our news piece on childhood adversity and cognition.

What adoption studies cannot show

Adoptive homes are screened. Adoptive parents are on average better off than birth parents. Nisbett and colleagues argue this restricted range may make adoption studies underestimate environmental effects; other researchers dispute that, so it stays open.
The comparison group decides the answer. Left in an institution, left with birth parents, or living next door in the adoptive community give very different gains.
Birth-family effects are not only genes. Prenatal conditions travel with birth mothers, and parental education is only an imperfect indicator of environmental quality, as Kendler and colleagues note.
Extreme deprivation is not ordinary adoption. The Romanian studies show what happens at the far end and should not be stretched to typical adoptions.
Numbers differ by test and age. The studies used different tests at different ages, from military conscription exams in Sweden to the WISC-IV in Bucharest and the Wechsler adult scale in Minnesota, so gains in points are not perfectly comparable.
Nothing here describes an individual. These are averages across groups. They say nothing about any one child’s potential or any one family’s parenting.

Adoption research also reaches beyond IQ. In a natural experiment in which Korean-born adoptees were quasi-randomly assigned to American families, Sacerdote (2007) found that parents’ education and family size predicted adoptees’ education, income and health more than parents’ income did, and that shared family environment explained 14% of the variation in educational attainment. That study did not measure IQ. Traits such as persistence are shaped by genes as well: in a UK twin study, perseverance of effort was 37% heritable, a finding covered in our guide to grit vs IQ.

Quick answers

Does adoption raise IQ? On average, yes, relative to siblings or peers left behind: about 3 to 4 points in a Swedish sibling comparison and roughly 8 to 20 points in a French study of neglected children. Adoptees do not differ in IQ from peers in their new communities.
Do adopted children take on their adoptive parents’ IQ? Not by adolescence. In the Colorado Adoption Project they resembled adoptive parents slightly early on and not at all by middle childhood.
How much of IQ is genetic? It rises with age, from about 20% in infancy to perhaps 80% in later adulthood, and about 70% in the Minnesota twin study of mostly middle-class adults.
Does early neglect permanently lower IQ? Not always. Romanian adoptees with more than six months of deprivation had normal rates of IQ below 80 by age 22 to 25, though other difficulties persisted.
Does high heritability mean parenting does not matter? No. The twin authors say their findings do not detract from the value of parenting and education, and the Swedish authors say the malleability they found is fully consistent with genetic influence.

The bottom line

Adoption studies say two things that sound opposed but are not. Environment can change the average IQ of a group, by a few points in ordinary comparisons and by much more when children leave severe deprivation. And as children grow, differences between people raised in ordinary homes track their birth parents more than their adoptive ones. Both come with limits: adoptive homes are screened, deprivation studies describe extremes, and none of the numbers applies to a single child.

Sources

van IJzendoorn, M.H., Juffer, F. and Klein Poelhuis, C.W. (2005): Adoption and cognitive development: a meta-analytic comparison of adopted and nonadopted children’s IQ and school performance, Psychological Bulletin, 131(2), 301-316.
Kendler, K.S., Turkheimer, E., Ohlsson, H., Sundquist, J. and Sundquist, K. (2015): Family environment and the malleability of cognitive ability: a Swedish national home-reared and adopted-away cosibling control study, PNAS, 112(15), 4612-4617.
Duyme, M., Dumaret, A.-C. and Tomkiewicz, S. (1999): How can we boost IQs of “dull children”? A late adoption study, PNAS, 96(15), 8790-8794.
Bouchard, T.J. et al. (1990): Sources of human psychological differences: the Minnesota Study of Twins Reared Apart, Science, 250, 223-228.
Plomin, R., Fulker, D.W., Corley, R. and DeFries, J.C. (1997): Nature, nurture, and cognitive development from 1 to 16 years: a parent-offspring adoption study, Psychological Science, 8(6), 442-447.
Petrill, S.A. et al. (2004): Genetic and environmental contributions to general cognitive ability through the first 16 years of life, Developmental Psychology, 40(5), 805-812.
Plomin, R. and Deary, I.J. (2015): Genetics and intelligence differences: five special findings, Molecular Psychiatry, 20(1), 98-108.
Nisbett, R.E. et al. (2012): Intelligence: new findings and theoretical developments, American Psychologist, 67(2), 130-159.
Nelson, C.A. et al. (2007): Cognitive recovery in socially deprived young children: the Bucharest Early Intervention Project, Science, 318, 1937-1940.
Fox, N.A., Almas, A.N., Degnan, K.A., Nelson, C.A. and Zeanah, C.H. (2011): The effects of severe psychosocial deprivation and foster care intervention on cognitive development at 8 years of age: findings from the Bucharest Early Intervention Project, Journal of Child Psychology and Psychiatry, 52(9), 919-928.
Sonuga-Barke, E.J.S. et al. (2017): Child-to-adult neurodevelopmental and mental health trajectories after early life deprivation: the young adult follow-up of the longitudinal English and Romanian Adoptees study, The Lancet, 389(10078), 1539-1548.
Sacerdote, B. (2007): How large are the effects from changes in family environment? A study of Korean American adoptees, Quarterly Journal of Economics, 122(1), 119-157.
Sauce, B. and Matzel, L.D. (2018): The paradox of intelligence: heritability and malleability coexist in hidden gene-environment interplay, Psychological Bulletin, 144(1), 26-47.

Share this article

Know someone who keeps seeing these numbers quoted without the scale they were measured on? Send it to them.

Keep reading

All articles →

Chart showing how the heritability of IQ rises from childhood to adulthood

Research & Evidence

Is IQ Genetic?

Heritability estimates for IQ rise from roughly 0.4 in childhood to 0.7 or higher in adulthood, which sounds decisive until you know what heritability measures. Here is what twin and adoption studies actually show, and the four conclusions people draw from them that the statistic does not support.

August 27, 20267 min read

Research & Evidence

Toxic Stress and IQ

Researchers in Romania did something that can never be done again for ethical reasons: they randomly assigned some institutionalized children to foster care and left others in the institution, then followed both groups for twenty years. The IQ gap that opened up has held at every check-in since. What has not held up as cleanly is the famous claim about exactly when the window closes.

September 10, 20266 min read

Understanding IQ

Grit vs IQ

September 29, 202610 min read

Grit vs IQ

Posted on September 29, 2026 by Mahesh

Understanding IQ

Grit vs IQ: Which Predicts Success More? What the Studies Found

By Alden Cross, Intelligence Research WriterPublished September 29, 2026 · 10 min readReviewed by Aarav Vardhan, Editor-in-Chief

Three cards compare correlations. Grit and school achievement: a weighted correlation of .19 across 137 studies and 285,331 people. Grit and conscientiousness: a corrected correlation of .84 across 88 samples and 66,807 people. Intelligence and school grades: a corrected correlation of .54 across 240 samples and 105,185 people, corrected for range restriction, so not like for like with the grit figures.

Grit vs IQ: does perseverance beat intelligence? In the research, no. Two large reviews find that grit correlates only weakly to moderately with school achievement (a weighted correlation of .19 across 137 studies), while a review of 240 samples puts intelligence tests and school grades at .54 after corrections. Grit also overlaps almost completely with the Big Five trait of conscientiousness (.84) and is nearly unrelated to cognitive ability (.05). It behaves like a separate effort trait that adds a little to IQ, not a replacement for it. This guide walks through the original grit studies, the meta-analyses, the marshmallow test, and a widely repeated claim about paying people to raise their IQ scores that was retracted in 2025.

Grit and school achievement: a weighted correlation of .19 across 137 studies and 285,331 people, roughly 4% of the differences in achievement.
Grit and conscientiousness: a corrected correlation of .84, so close to the same trait; overall grit adds almost nothing once conscientiousness is counted.
Grit and cognitive ability: about .05 corrected, meaning grittier people are not systematically smarter or less smart.
Marshmallow test: the link to teenage achievement shrank by about two thirds once family background and early cognitive ability were controlled.
Retraction: the 2011 paper behind “paying people adds 10 IQ points” was retracted by its journal on 1 May 2025.

What is grit, and how is it measured?

Grit is defined as perseverance and passion for long-term goals. The psychologist Angela Duckworth and colleagues introduced it in a 2007 paper in the Journal of Personality and Social Psychology, along with a questionnaire. The original Grit Scale has 12 items split into two factors of six: consistency of interests and perseverance of effort. People rate each item from 1 to 5, and a short 8-item version, the Grit-S, was published in 2009. Both are scored by averaging, so the maximum is 5. Everything rests on self-report, which matters later. If you are weighing effort against ability in a wider sense, our guides to growth mindset and intelligence and talent versus practice cover the neighboring evidence.

What did the original grit studies find?

The 2007 paper reported six studies. Its abstract says grit “accounted for an average of 4% of the variance in success outcomes” and showed incremental predictive validity over IQ and conscientiousness. Incremental means it added something on top, not that it outranked anything. The table shows what each sample actually found, including the parts that rarely make it into popular summaries.

The 2007 grit paper, sample by sample
Sample	What grit predicted	How IQ-type scores compared
Adults aged 25 and older (N = 1,545 and 690)	More schooling than same-age peers; one standard deviation of extra grit meant 35% lower odds of being a frequent career changer	No IQ measure; analyses controlled for age and the Big Five
Psychology majors at an Ivy League university (N = 138)	GPA, r = .25 (and .34 with SAT held constant)	SAT and GPA r = .30; grit and SAT r = −.20, so the grittier students had slightly lower SAT scores
West Point cadets (N = 1,218 and 1,308)	Finishing the first summer of training; about 1 in 20 cadets dropped out, and one standard deviation of grit meant more than 60% higher odds of finishing (Study 4)	West Point’s own Whole Candidate Score, built from SAT, class rank, leadership and fitness, did not predict who finished
Spelling Bee finalists aged 7 to 15 (N = 175)	Reaching a later round, with age controlled	With verbal IQ and age in the model, grit was no longer statistically significant; verbal IQ was

Two details deserve emphasis. Grit did not relate positively to IQ in these samples, so it is not a proxy for intelligence; at the Ivy League school it was slightly negatively related to SAT scores, and the authors speculated that smarter students “may be slightly less gritty.” But the Spelling Bee result cuts the other way from how grit is often described: once verbal IQ entered the model, grit stopped being a significant predictor, though the authors noted that missing IQ data cut the degrees of freedom by more than half. The samples are also highly selected, which shrinks IQ correlations as well as grit ones.

What do the big reviews say about grit?

Two large meta-analyses give the clearest picture. Credé, Tynan and Harms (2017) pooled 88 independent samples and 66,807 people. Grit correlated .18 with academic performance after correction for measurement error (.15 uncorrected), and .84 with conscientiousness, but only .05 with cognitive ability. After conscientiousness was accounted for, overall grit explained no additional variance in overall academic performance or high-school GPA and almost none for college GPA. Only the perseverance-of-effort half of grit added a small amount, and the meta-analysis did not confirm the two-part structure the scale assumes. Lam and Zhou (2022) covered 137 studies and 285,331 people and found a weighted correlation of .19 between grit and academic achievement, .21 for perseverance of effort and .08 for consistency of interest.

A UK twin study points the same way. Rimfeld and colleagues studied 4,642 sixteen-year-olds from the Twins Early Development Study. The Big Five personality traits explained 5.6% of the variance in GCSE grades, and adding grit raised that by only 0.5 percentage points. Perseverance of effort was 37% heritable, and its genetic correlation with conscientiousness was .86. That study used no intelligence measure, so it tests grit against personality rather than against IQ.

How the correlations line up (corrected figures are not like for like)
Relationship	Correlation	Study and sample
Grit and academic performance	.18 corrected (.15 observed)	Credé et al. 2017: 88 samples, 66,807 people
Grit and academic achievement	.19 weighted, observed	Lam and Zhou 2022: 137 studies, 285,331 people
Grit and cognitive ability	.05 corrected	Credé et al. 2017: 21 samples, 11,513 people
Intelligence tests and school grades	.54 corrected, incl. range restriction	Roth et al. 2015: 240 samples, 105,185 people
Cognitive tests and job performance	.31 (2022 re-estimate; .51 in 1998)	Sackett et al. 2022, employees
Conscientiousness and job performance	.19 (2022; .31 in 1998)	Sackett et al. 2022, employees

These are not like-for-like comparisons, so read the table as a map, not a scoreboard. The .54 is corrected for measurement error and range restriction and comes from school grades; grit’s .18 is corrected for measurement error only; the .19 is an uncorrected average. Job-performance validity for cognitive tests is itself contested: the older .51 figure was re-estimated at .31 in 2022 after Sackett and colleagues argued that earlier range-restriction corrections were too generous, and structured interviews now top that table. Even so, the direction is consistent: in these reviews cognitive ability shows a larger correlation with outcomes than grit or conscientiousness does, which fits our guides to whether IQ predicts school success and what a cognitive test at work predicts. For personality more broadly, see IQ versus personality.

What about self-control and the marshmallow test?

Self-control is a nearby idea with better evidence behind it. The Dunedin study followed a complete birth cohort of 1,037 children in New Zealand to age 32, with 96% still taking part. Childhood self-control, rated from nine measures between ages 3 and 11, predicted adult health, substance dependence, finances and criminal convictions “even after accounting for social class origins and IQ,” and the pattern held in a sibling comparison of about 500 pairs. It is not a contest with IQ, though: in that cohort self-control and IQ correlated .44, and the design is observational, so it shows self-control adds prediction beyond IQ, not that it exceeds it.

The marshmallow test is weaker than its reputation. Watts, Duncan and Quan (2018) ran a conceptual replication on a larger, more varied sample: 918 US children with valid delay data, with the main analyses on 552 whose mothers had not completed college. An extra minute of waiting at age 4 went with about a tenth of a standard deviation in achievement at 15, an association about half the size of the original. It fell by two thirds with controls for family background, early cognitive ability and the home environment, and became non-significant once early cognitive and behavioral measures were added (standardized coefficient .24, then .08, then .05). Most of the variation came from whether a child could wait at least 20 seconds. How much a home can shape such traits at all is a different question; see our guide to adoption and IQ.

Your own number

Where would your own score land?

Find your IQ score now! →

Secure & encryptedInstant results10–20 minutes

Does motivation change IQ test scores? A retracted claim

A popular argument runs that IQ tests mostly measure motivation, because paying people raises their scores by about 10 points. It traces to a 2011 paper by Duckworth and colleagues in PNAS, which reported a meta-analytic effect of 0.64 standard deviations (about 9.6 points on a scale with a standard deviation of 15, our conversion; the paper reports standard-deviation units only). That paper was retracted on 1 May 2025. The journal issued an expression of concern in February 2025, and in the retraction notice the authors explained that a retracted 1978 article had contributed three large effect sizes and 24% of the meta-analytic sample. Without it the effect fell to 0.47; after adjusting for publication bias it was 0.16, with a confidence interval (−0.02 to 0.34) that includes zero. All four publication-bias tests now indicated bias, and the authors decided to retract the article.

What survives is narrower. Effort and engagement plausibly affect any test performance, which is why conditions matter, as covered in our guide to the factors that affect IQ test results and the news piece on what practice actually changes. But a ten-point payment bonus is not an established fact, and it should not be repeated as one.

What should you take from grit vs IQ?

Treat it as both, not either. IQ predicts school and job outcomes moderately; grit and conscientiousness add a small, real increment on top.
Ask which half of grit you mean. Perseverance of effort carries the prediction; consistency of interest adds little.
Do not use grit scores to judge people or schools. Duckworth wrote in 2016 that she worried she had “contributed, inadvertently,” to high-stakes character testing, and a 2015 review by her and Yeager concluded that no available measure suits between-school accountability.
Expect self-report bias. High-performing charter-school students rated themselves lower on grit than district-school peers in Duckworth’s own data.
Be cautious about grit interventions. The correlations are modest and observational, and we did not review any trial showing that raising grit raises achievement.

For a wider view of what predicts success beyond a single score, see our guides to IQ versus EQ and whether IQ predicts income, and for a condition where persistence and instruction interact, dyslexia and IQ.

Quick answers

Is grit more important than IQ? No source we reviewed supports that. Grit adds a small amount to prediction; cognitive ability predicts school grades more strongly in the largest reviews, though the figures are not like for like.
Is grit the same as conscientiousness? Almost. The corrected correlation is .84, and the twin study found a genetic correlation of .86 for perseverance of effort.
Are gritty people smarter? No. Grit and cognitive ability correlate about .05.
Does the marshmallow test predict success? Weakly. The link to age-15 achievement shrank by two thirds with controls and lost significance with early cognitive measures.
Does paying people raise IQ scores by 10 points? Not established: the paper that claimed it was retracted in 2025.
Can you measure grit reliably? Mostly by self-report, and its author says character questionnaires are not ready for high-stakes use.

The bottom line

Grit is real enough as a label for sustained effort, but the evidence does not support the claim that it beats intelligence. Across the largest reviews it correlates weakly to moderately with achievement, overlaps almost completely with conscientiousness, and adds very little once that trait is counted. The best-supported version of the grit story is modest: talent and sustained effort both matter, effort is hard to measure, and neither a marshmallow nor a payment can be read as a shortcut to a score.

Sources

Duckworth, A.L., Peterson, C., Matthews, M.D. and Kelly, D.R. (2007): Grit: perseverance and passion for long-term goals, Journal of Personality and Social Psychology, 92(6), 1087-1101.
Duckworth, A.L. and Quinn, P.D. (2009): Development and validation of the Short Grit Scale (Grit-S), Journal of Personality Assessment, 91(2), 166-174.
Credé, M., Tynan, M.C. and Harms, P.D. (2017): Much ado about grit: a meta-analytic synthesis of the grit literature, Journal of Personality and Social Psychology, 113(3), 492-511.
Lam, K.K.L. and Zhou, M. (2022): Grit and academic achievement: a comparative cross-cultural meta-analysis, Journal of Educational Psychology, 114(3), 597-621.
Rimfeld, K., Kovas, Y., Dale, P.S. and Plomin, R. (2016): True grit and genetics: predicting academic achievement from personality, Journal of Personality and Social Psychology, 111(5), 780-789.
Roth, B. et al. (2015): Intelligence and school grades: a meta-analysis, Intelligence, 53, 118-137.
Sackett, P.R., Zhang, C., Berry, C.M. and Lievens, F. (2022): Revisiting meta-analytic estimates of validity in personnel selection: addressing systematic overcorrection for restriction of range, Journal of Applied Psychology, 107(11), 2040-2068.
Moffitt, T.E. et al. (2011): A gradient of childhood self-control predicts health, wealth, and public safety, PNAS, 108(7), 2693-2698.
Watts, T.W., Duncan, G.J. and Quan, H. (2018): Revisiting the marshmallow test: a conceptual replication investigating links between early delay of gratification and later outcomes, Psychological Science, 29(7), 1159-1177.
PNAS (2025): Retraction for Duckworth et al., “Role of test motivation in intelligence testing,” PNAS, 122(18), e2508040122.
Duckworth, A.L. and Yeager, D.S. (2015): Measurement matters: assessing personal qualities other than cognitive ability for educational purposes, Educational Researcher, 44(4), 237-251.
Duckworth, A.L. (2016): Don’t grade schools on grit, The New York Times, 26 March 2016.

Share this article

Know someone who keeps seeing these numbers quoted without the scale they were measured on? Send it to them.

Keep reading

All articles →

Chart comparing three waves of growth mindset evidence, showing a weak pooled intervention effect, a targeted national trial that worked for lower-achieving students, and a later review finding widespread bias

Research & Evidence

Growth Mindset and Intelligence

The claim that believing intelligence is malleable makes you achieve more is one of the most widely taught ideas in education, and one of the most heavily contested in the research literature. Three waves of evidence have now landed. Here is what each found, and the question none of them actually answered.

September 7, 20266 min read

Research & Evidence

Talent vs Practice

The 10,000-hour rule made a strong claim: expert performance is built by practice, not given by talent. When researchers pooled the studies that had actually measured it, practice explained about a quarter of the difference between people in games and almost none in professional work. Here is what fills the rest of the gap.

September 7, 20266 min read

Research & Evidence

Adoption and IQ

September 29, 202612 min read

What IQ Is a Moron?

Posted on September 29, 2026 by Mahesh

Scores & Scales

What IQ Is a Moron? Idiot, Imbecile and Moron as Old IQ Labels

In the old clinical scheme a moron had an IQ of roughly 50 to 70, an imbecile about 20 or 25 to 50 and an idiot below that, according to Lewis Terman in 1916. See who coined the words, how the cut-offs moved, what the 1913 UK Act said and why the labels were dropped.

By Leon Falkner, Data & Research AnalystPublished September 29, 2026 · 11 min readReviewed by Aarav Vardhan, Editor-in-Chief

A bar chart draws the 1916 classification of Lewis Terman on an IQ scale from 0 to 150. The three retired labels are highlighted: idiot below 20 or 25, imbecile 20 or 25 to 50 and moron 50 to 70, with adult mental levels of up to about 3, 3 to 7 and about 7 to 11 years. The other bands run from border-line deficiency at 70 to 80 up to near genius above 140.

What IQ is a moron? In the old clinical scheme, a “moron” was someone with the mildest form of what is now called intellectual disability, and in Lewis Terman’s 1916 Stanford-Binet tables the group sat at an IQ of roughly 50 to 70. An “imbecile” sat at about 20 or 25 up to 50, and an “idiot” below that. These were never official cut-offs. The original 1910 definitions counted mental age, not IQ; different authorities drew the lines in different places; and Terman himself called the sub-grade bands “not very secure.” This guide covers who defined each word, how the lines moved, what a 1913 UK Act said, why the words were dropped and what replaced them. All three words are offensive today and are quoted here only as history.

Moron: an IQ of roughly 50 to 70 in Terman’s 1916 tables; a mental age of about 7 or 8 up to 12 in the 1910 scheme.
Imbecile: an IQ of about 20 or 25 to 50; a mental age of 3 to 7.
Idiot: an IQ below about 20 or 25; a mental age of up to about 2 or 3.
Origin: “moron” was coined as a new technical term in 1910, from a Greek word for foolish; “idiot” and “imbecile” are far older.
Retired: Rosa’s Law (2010) replaced “mental retardation” with “intellectual disability” in US federal law, and DSM-5 (2013) bases severity on everyday functioning, not IQ bands.

What IQ was a moron, an imbecile and an idiot?

The nearest thing to an original IQ table comes from Lewis Terman’s 1916 The Measurement of Intelligence, which introduced the Stanford-Binet in the United States (see our guide to the Stanford-Binet). Terman labeled every IQ below 70 “definite feeble-mindedness” and then subdivided the group: those between 50 and 70 “include most of the morons,” those between 20 or 25 and 50 are “ordinarily to be classed as imbeciles,” and those below 20 or 25 are idiots. He also translated the three grades into the mental level of an adult, shown next to the 1910 mental-age bands in the table.

The three labels in two old systems
Old label	IQ band (Terman, 1916)	Adult mental level (Terman)	Mental age (1910 scheme, as Goddard reported it)
Idiot	below 20 or 25	up to about 3 years	up to and including 2 years
Imbecile	20 or 25 to 50	3 to 7 years	3 to 7 years, inclusive
Moron	50 to 70 (“most of the morons”)	about 7 to 11 years	7 to 12 years; some sources say 8 to 12

The tidy version that circulates online — idiot 0 to 25, imbecile 26 to 50, moron 51 to 70 — is a later simplification, and we found no primary source that prints it. Terman’s own version has “or” boundaries at 20 or 25, says 50 to 70 covers most morons, and adds that the lines between such groups are “absolutely arbitrary, a matter of definition only.” Sources also disagree by a year at each end of the mental-age bands: Goddard’s 1914 book gives the moron range as 7 to 12, the Online Etymology Dictionary gives 8 to 12, and Terman’s adult figure is about 7 to 11.

Where did the words come from?

Two of the three are far older than IQ testing. Idiot comes from the Greek idiotes, a “private person” or layman, and by the early fourteenth century English used it for a person unable to reason. Imbecile comes from the Latin imbecillus, “weak, feeble,” and first described bodily weakness in the 1540s; the mental sense arrived in the mid-eighteenth century. Moron is different: it was invented for the job.

In 1910 the American Association for the Study of the Feeble-Minded adopted a three-grade scheme, and the psychologist Henry H. Goddard supplied a new word for the top grade, taken from the Greek moros, “foolish, dull.” Goddard’s 1914 book Feeble-Mindedness: Its Causes and Consequences reports the scheme in mental-age years: idiots up to and including two, imbeciles three to seven and, in his words, “for those from seven to twelve a new term has been invented; they are now called morons.” He added that the new word “designates almost exactly what is meant by ‘feeble-minded’ in England.” Coined as a clinical term in 1910, “moron” was in print as an insult by 1918 and was later dropped from technical use, according to the Online Etymology Dictionary.

How did mental ages become IQ bands?

Mental age is the age at which an average child performs like the person tested. A ratio IQ divides it by chronological age and multiplies by 100, so a child of 10 with a mental age of 5 scores 50. Adults cause a problem, because mental age stops rising in the mid-teens. Terman handled it by ignoring age above 16; his case notes give scores such as “approximately 65 (not counting age beyond 16).” Using 16 as the divisor (our arithmetic), a mental age of 8 is an IQ of 50 and a mental age of 11 is about 69. That is why “IQ 50 to 70” and “an adult mental age of about 7 to 11” describe nearly the same people, and why Goddard’s upper limit of 12 works out nearer to IQ 75. Modern tests no longer report mental age at all. Our guides to IQ and mental age and deviation IQ versus ratio IQ explain why, and the piece on standard deviation 15 versus 16 explains why older scales do not line up with today’s.

Who drew the line where?

The words disappeared, but the underlying question — where does the line sit? — kept moving. The table shows how far the cut-off wandered over a century.

How authorities drew the line, 1910 to today
Authority and year	Term used	Where the line sat
Goddard and the American Association, 1910 (as reported in 1914)	idiot, imbecile, moron	mental age up to 2; 3 to 7; 7 to 12. No IQ figures.
Terman, Stanford-Binet, 1916	“definite feeble-mindedness”	IQ below 70; morons mostly 50 to 70
Kuhlmann (as quoted by Wechsler)	mental defectives	IQ below 75
Wechsler-Bellevue, 1944	defective; borderline	65 and below; 66 to 79
WAIS, as tabulated in Wechsler’s 1958 book	defective; borderline	69 and below; 70 to 79
AAMD manual, 1959	mental retardation, borderline included	one standard deviation below the mean (about 85 on today’s scale of 15); “almost 15 percent” of the US population
AAMD manual, 1973	mental retardation	back to about 70; the ceiling may extend to 75
DSM-IV-TR, 2000 (as quoted in Atkins v. Virginia)	“mild” mental retardation	an IQ of 50–55 to approximately 70
DSM-5, 2013	intellectual disability	about two standard deviations below the mean (an IQ of about 70 or below), plus limits in everyday functioning
AAIDD, current definition	intellectual disability	“around 70 or as high as 75,” plus adaptive limits and onset before age 22

The 1959 row is the striking one. The American Association on Mental Deficiency moved the line to one standard deviation below the mean, a step its own 1973 manual says could take in almost 15 percent of the US population, and then reversed it. On a normal curve the share of people below 85 is 15.9 percent (our calculation). A cut-off is a decision, not a discovery, and the same person could sit inside or outside a diagnosis depending on which manual was in force. Note that Wechsler’s tables never used the three words at all; his labels were “defective” and “borderline.”

What did UK law say? The Mental Deficiency Act 1913

British law used the same vocabulary and defined it differently. Section 1 of the Mental Deficiency Act 1913 named four classes, each by what a person could do rather than by any test score: idiots, unable “to guard themselves against common physical dangers”; imbeciles, “incapable of managing themselves or their affairs”; feeble-minded persons, who needed “care, supervision, and control”; and “moral imbeciles,” defined by a permanent mental defect coupled with “strong vicious or criminal propensities.” The statute contains no IQ or mental-age figure. Goddard himself observed that his new American word “moron” covered almost exactly what the English called “feeble-minded.” The Mental Health Act 1959 later replaced the four classes with “severe subnormality” and “subnormality,” and British services now say “learning disability,” which means what the United States calls intellectual disability — a terminology trap covered in our guide to low IQ, intellectual disability and learning disability.

Your own number

Where would your own score land?

Find your IQ score now! →

Secure & encryptedInstant results10–20 minutes

Why were the words dropped?

A clinical word for a stigmatized condition tends to turn into an insult, and its replacement then absorbs the stigma. “Moron” made the jump within about eight years of its coinage. The field’s response was to keep changing the label: by 1973 the American Association on Mental Deficiency’s manual was titled Manual on Terminology and Classification in Mental Retardation, and in 2010 Congress passed Rosa’s Law (Public Law 111-256, signed 5 October 2010), which replaced “mental retardation” with “intellectual disability” in a list of federal education, health and rehabilitation statutes. The law renamed rather than redefined: its own text says the new term means a condition previously referred to as mental retardation. The Social Security Administration made the same substitution in its disability listings from 3 September 2013, and DSM-5 renamed the diagnosis the same year.

What were the labels used to justify? Buck v. Bell

The categories were not harmless bookkeeping. In Buck v. Bell (1927) the US Supreme Court upheld Virginia’s compulsory sterilization law, and Justice Holmes closed the opinion with the line “Three generations of imbeciles are enough.” The opinion itself describes Carrie Buck, her mother and her daughter as “feeble-minded” and states no IQ or mental age; “imbeciles” appears only in that sentence. Lutz Kaelber of the University of Vermont counts compulsory sterilization laws in over 30 US states that led to more than 60,000 sterilizations of disabled people. The wider story of how IQ testing was bent to that agenda, and where the popular version overreaches, is in our guide to eugenics and the history of IQ testing.

How is intellectual disability defined today?

Modern diagnosis is not “an IQ below 70.” DSM-5 requires limits in intellectual functioning and in adaptive functioning across conceptual, social and practical skills, and describes the intellectual limit as roughly two standard deviations below the population mean, an IQ of about 70 or below. The American Association on Intellectual and Developmental Disabilities says a score “around 70 or as high as 75” indicates a significant limitation and places onset before age 22. The World Health Organization’s ICD-11 describes the mild range as roughly two to three standard deviations below the mean. Severity is now set by everyday functioning rather than by an IQ band. Our guide to what an IQ of 70 means explains the measurement-error zone around the line, and our news pieces on the Hamm v. Smith decision and the Social Security score rules show how courts and benefit rules handle the same numbers.

The bell curve helps here, with one warning. On a normal model with a mean of 100 and a standard deviation of 15, about 2.3 percent of people would score below 70 (roughly 1 in 44), 0.04 percent below 50 (roughly 1 in 2,300) and almost nobody below 25. Those are model figures, not prevalence counts: the American Psychiatric Association puts intellectual disability at about 1 percent of the population, well under the 2.3 percent the curve implies, because a diagnosis also needs limits in daily functioning. Our IQ percentile calculator and IQ bell curve pages let you check any score. One more confusion worth clearing: dyslexia and other specific learning disorders are not intellectual disability and occur at any IQ, as our guide to dyslexia and IQ explains.

Quick answers

What IQ is a moron? Roughly 50 to 70 in Terman’s 1916 tables; a mental age of about 7 or 8 up to 12 in the 1910 scheme.
What IQ is an imbecile? About 20 or 25 to 50 in Terman’s tables; a mental age of 3 to 7.
What IQ is an idiot? Below about 20 or 25; a mental age of up to about 2 or 3.
Who invented the word “moron”? Henry H. Goddard proposed it in 1910, from the Greek moros, and the American Association for the Study of the Feeble-Minded adopted it.
Was there ever an official IQ table for the three words? No. The 1910 scheme used mental age, and Terman’s IQ bands were approximate suggestions he called “not very secure.”
Are the words still used in diagnosis? No. Current manuals say intellectual disability (UK services say learning disability), and the old words survive only as insults.
Is IQ 70 the modern cut-off? It is a guide, not a line: about 70 to 75, plus limits in everyday functioning that begin in the developmental period.

The bottom line

“Moron,” “imbecile” and “idiot” began as the three grades of one clinical scale, defined first by mental age in 1910 and then given approximate IQ bands by Terman in 1916: about 50 to 70, 20 or 25 to 50, and below that. The bands were soft from the start, the cut-off for the whole group moved by a full standard deviation within living memory, and the words turned into insults within a decade. What replaced them is not a new set of IQ labels but a two-part test — a low score plus limits in everyday life — that no online quiz can apply.

Sources

Terman, L.M. (1916): The Measurement of Intelligence. Houghton Mifflin, pp. 78-80.
Goddard, H.H. (1914): Feeble-Mindedness: Its Causes and Consequences. Macmillan, pp. 3-4.
Wechsler, D. (1958): The Measurement and Appraisal of Adult Intelligence, 4th ed. Williams & Wilkins.
Grossman, H.J., ed. (1973): Manual on Terminology and Classification in Mental Retardation. American Association on Mental Deficiency.
Online Etymology Dictionary: entries for moron, idiot and imbecile.
Mental Deficiency Act 1913 (3 & 4 Geo. 5 c. 28), section 1, transcription.
Buck v. Bell, 274 U.S. 200 (1927).
Atkins v. Virginia, 536 U.S. 304 (2002), footnote 3 (DSM-IV-TR).
Rosa’s Law, Public Law 111-256 (2010).
Social Security Administration (2013): Change in terminology, mental retardation to intellectual disability, 78 FR 46499.
American Psychiatric Association: DSM-5 fact sheet, Intellectual Disability.
AAIDD: Definition of intellectual disability.
Kaelber, L.: Eugenics, compulsory sterilization in 50 American states, University of Vermont.
World Health Organization: ICD-11, code 6A00.0, mild disorder of intellectual development.

Share this article

Know someone who keeps seeing these numbers quoted without the scale they were measured on? Send it to them.

Keep reading

All articles →

Bell curve with the lower tail shaded below IQ 70 and a highlighted band from 65 to 75 showing the measurement-error zone around the score.

Scores & Scales

What an IQ of 70 Means

An IQ of 70 sits two standard deviations below the mean, at about the 2nd percentile. It is the score most often linked with intellectual disability, but a diagnosis never rests on the number alone: measurement error turns the line into a zone, and adaptive functioning decides who is inside it.

September 16, 20268 min read

Understanding IQ

IQ Testing and Eugenics

Between 1917 and 1919, the United States Army tested roughly 1.75 million recruits in the largest intelligence-testing program the world had yet seen. Within a decade, one psychologist had turned that data into a case for restricting immigration by race. Thirteen years after that, the same psychologist published a retraction few people have ever heard of.

September 10, 20266 min read

Understanding IQ

Dyslexia and IQ

September 29, 202612 min read

Creativity and IQ

Posted on September 28, 2026September 28, 2026 by Mahesh

Understanding IQ

Creativity and IQ: What the Threshold Theory Gets Wrong

Creativity and IQ correlate only weakly. The largest meta-analysis ever run on the question, covering 45,880 people, found no significant threshold at IQ 120, just a modest overall link of about .17 to .25. See what the strongest evidence shows, and why the classic claim does not hold up.

By Arjun Vardhan, News & Research WriterPublished September 28, 2026 · 8 min readReviewed by Aarav Vardhan, Editor-in-Chief

A line chart plots creative achievement against IQ score. The classic threshold claim draws a flat line above IQ 120. The largest meta-analysis instead finds a similar, modest correlation below and above 120, with no significant break at that point.

Creativity and IQ correlate only weakly, and not in the way the famous threshold theory claims. Are creative people more intelligent? Only slightly. That theory says intelligence and creativity correlate up to about IQ 120, then the link flattens out completely. The largest meta-analysis ever run on the question, pooling 447 correlations from 45,880 people, found a correlation of about .17 overall and no significant difference between the below-120 and above-120 groups — its author concluded plainly that “this study does not support threshold theory.” This guide covers where the threshold idea came from, what the two largest meta-analyses actually found, a more recent study that found the breakpoint moves depending on how creativity is measured, and what the field’s most famous creativity-test follow-up study found instead.

The classic claim: intelligence and creativity correlate below about IQ 120; above it, no further relationship (Guilford, 1950; Getzels and Jackson, 1962).
The largest test of it: a 2005 meta-analysis of 45,880 people found a weighted mean correlation of .174, with no significant difference above versus below IQ 120.
A 2021 update, correcting for measurement error: 34,610 people, overall correlation .25 — still modest, not the strong link pop psychology sometimes implies.
Creative-test scores predicted personal achievement, not public achievement, in a 50-year follow-up of Torrance Test takers; public creative achievement needed high scores on both creativity and intelligence.

Where the threshold theory came from

J.P. Guilford’s 1950 address to the American Psychological Association, titled simply “Creativity,” argued that standard intelligence tests miss an entire class of abilities: divergent thinking, the capacity to generate many different valid responses to an open-ended problem, rather than converge on the one correct answer an IQ test rewards. Getzels and Jackson followed in 1962 with Creativity and Intelligence: Explorations with Gifted Students, comparing high-IQ and high-creativity groups of students and finding they overlapped only partly. Out of that era came the threshold claim, credited across the literature to Barron (1961), Getzels and Jackson (1962), Guilford himself (1967) and MacKinnon (1962): below roughly IQ 120, a certain amount of intelligence is needed to be creative, so the two correlate; above it, intelligence stops being the limiting factor and the correlation should flatten toward zero.

It is one of the most widely repeated claims in popular writing about intelligence and creativity. It is also, on the best current evidence, not well supported as a hard cutoff.

The largest test of the threshold claim found no threshold

Kyung Hee Kim’s 2005 meta-analysis, “Can Only Intelligent People Be Creative?”, pooled 447 correlation coefficients from 21 studies published between 1961 and 2004, covering 45,880 people. The overall weighted mean correlation between intelligence and creativity was .174 (95% confidence interval .165 to .183) — positive, but modest.

Kim (2005): IQ-creativity correlation by IQ band
IQ band	Correlation (r)	Number of coefficients
Below 100	.260	part of 65 below-120
100 to 120	.140	part of 65 below-120
120 to 135	.259	part of 14 above-120
Above 135	-.215	only 2 coefficients, unreliable
All below IQ 120	.201	65 coefficients
All above IQ 120	.235	14 coefficients

If the threshold theory were right, the correlation above 120 should collapse toward zero. Instead it was slightly higher than below 120, and the difference was not statistically significant (chi-square 2.004, p = .157). Adding an IQ-120 threshold term to a regression model did not improve how well it fit the data. Kim’s own conclusion, in her words: “This study does not support threshold theory.” A narrower look at just the Torrance Tests of Creative Thinking, covered below, gave a similar mean correlation of .218 across 18 coefficients.

A 2021 update: a somewhat stronger link, still not a threshold

A 2021 update by Gerwig and colleagues in the Journal of Intelligence pooled a larger and more recent set of studies: 849 correlations from 112 studies and 34,610 people, and reported an overall correlation of .25 (95% confidence interval .21 to .30) between intelligence and divergent-thinking scores. That is noticeably higher than Kim’s .174, but the gap is mostly statistical rather than a real change over time: once the newer studies are restricted to the same uncorrected method Kim used, the estimate drops to about .19, close to her figure. The higher headline number in the 2021 study comes mainly from correcting for measurement unreliability in the tests themselves, a standard meta-analytic adjustment, not from creativity and intelligence having grown more closely linked.

A moving breakpoint, not a fixed one

Not every study agrees the relationship is perfectly flat throughout. A widely cited re-analysis by Jauk, Benedek, Dunst and Neubauer, published in Intelligence in 2013, used a statistical technique that finds a breakpoint in the data rather than assuming one at 120. It reported that where the breakpoint falls depends on how strictly creativity is defined: for a simple count of how many ideas someone can generate, the breakpoint sat far lower, around IQ 85; for a demanding measure of genuinely original ideas, it moved up toward roughly 120. In other words, a fixed threshold at 120 may be an artefact of which creativity measure a study happens to use, not a single biological cutoff. The broader pattern across both meta-analyses and this re-analysis is the same: a real but modest link that behaves differently depending on how “creative” is defined and measured, not a clean switch that turns off at a specific score.

Is creativity even part of intelligence?

Part of the disagreement is definitional. Robert Sternberg’s triarchic theory of intelligence names “creative intelligence” — the ability to deal with novelty and generate new ideas — as one of three components, alongside analytical and practical intelligence, arguing standard IQ tests mostly capture only the analytical piece. The Cattell-Horn-Carroll (CHC) model behind most modern IQ tests, by contrast, does not name creativity as one of its broad abilities at all; the closest fit is Glr, long-term storage and retrieval, which covers the kind of fluent, associative idea generation that divergent-thinking tests measure. Our guide to Gardner’s multiple intelligences covers a separate, and more famous, argument that standard IQ tests miss entire categories of ability; see also is intelligence limited to IQ.

Your own number

Where would your own score land?

Find your IQ score now! →

Secure & encryptedInstant results10–20 minutes

What the most famous creativity-test follow-up study actually found

E. Paul Torrance built the Torrance Tests of Creative Thinking (TTCT) in the 1960s to score fluency, flexibility, originality and elaboration on open-ended tasks, and then tracked his test-takers for decades. A 40-year follow-up (Cramond and colleagues, 2005) and a 50-year follow-up (Runco and colleagues, 2010) both looked at how well childhood TTCT scores predicted adult creative achievement. The finding is more precise than the popular version of the story: TTCT scores moderately predicted personal achievement (creative accomplishments in an individual’s own life and work), but predicting public achievement (creative work recognised publicly) needed an interaction between creativity and intelligence scores together. Neither measure alone told the fuller story; the combination did.

What this means if you are trying to identify creative potential

A single IQ score was never built to certify creativity, and a single creativity-test score is not a substitute for one either. Programmes that identify gifted children using only an IQ cutoff can miss students whose strength shows up in open-ended, divergent tasks rather than the convergent, single-right-answer format most IQ subtests use; the Torrance follow-ups suggest the reverse is also true, since public achievement took both kinds of ability together. Neither score alone settles the question of who is creative, and a modest .17 to .25 correlation leaves most of the variation in creativity unexplained by IQ in either direction. For another surprising trait people have tried to link to intelligence, with a very different causal story, see our guide to myopia and IQ.

Quick answers

Are highly intelligent people more creative? Only slightly. The two largest meta-analyses put the correlation at about .17 to .25, positive but modest.
Is there really an IQ 120 cutoff where the link disappears? The best-supported answer is no. The largest meta-analysis (45,880 people) found no significant difference in the correlation above versus below IQ 120.
Can someone with an average IQ be highly creative? Yes. The correlation is far from perfect in either band, and one re-analysis put the breakpoint for basic idea-generation as low as IQ 85.
Does a high IQ guarantee creativity? No. Correlations around .2 to .25 mean intelligence explains only a small share of the differences between people in creative-test scores.
Is creativity part of IQ? Not in the CHC model most modern IQ tests are built on. Sternberg’s triarchic theory treats it as a separate kind of intelligence; most IQ tests measure mainly the analytical kind.

The bottom line

The tidy version of the threshold theory — strong link below IQ 120, none above — does not hold up against the largest evidence. What holds up instead is a real but modest correlation, roughly .17 to .25, that does not flatten sharply at any single score, though where it is strongest can shift with how strictly a study defines “creative.” A high IQ makes highly creative output somewhat more likely, but the Torrance follow-ups suggest that publicly recognised creative achievement is better predicted by intelligence and creativity acting together than by either score alone.

Sources

Guilford, J.P. (1950): Creativity, American Psychologist, 5, 444-454.
Getzels, J.W. and Jackson, P.W. (1962): Creativity and Intelligence: Explorations with Gifted Students. Wiley.
Kim, K.H. (2005): Can only intelligent people be creative? A meta-analysis, Journal of Secondary Gifted Education, 16(2/3), 57-66.
Gerwig, A. et al. (2021): The relationship between intelligence and divergent thinking, a meta-analytic update, Journal of Intelligence, 9(2), 23.
Jauk, E., Benedek, M., Dunst, B. and Neubauer, A.C. (2013): The relationship between intelligence and creativity, new support for the threshold hypothesis by means of empirical breakpoint detection, Intelligence, 41(4), 212-221.
Runco, M.A., Millar, G., Acar, S. and Cramond, B. (2010): Torrance Tests of Creative Thinking as predictors of personal and public achievement, a fifty-year follow-up, Creativity Research Journal, 22(4), 361-368.
Cramond, B., Matthews-Morgan, J., Bandalos, D. and Zuo, L. (2005): A report on the 40-year follow-up of the Torrance Tests of Creative Thinking, Gifted Child Quarterly, 49(4), 283-291.
Sternberg, R.J. (1988): The Triarchic Mind: A New Theory of Human Intelligence. Viking Press.

Share this article

Know someone who keeps seeing these numbers quoted without the scale they were measured on? Send it to them.

Keep reading

All articles →

Diagram of the positive manifold: a correlation grid showing that verbal, spatial, numerical and memory tests all correlate positively with one another, which is the finding multiple intelligences theory predicts should not appear

Research & Evidence

Gardner’s Multiple Intelligences

Gardner’s multiple intelligences theory is one of the most influential ideas in education and almost entirely absent from the tests psychologists use. That gap is not stubbornness. It comes down to one stubborn statistical finding, one missing instrument, and a real insight the theory carries.

August 31, 20267 min read

Understanding IQ

Is Intelligence limited to IQ?

As we explore the fascinating world of intelligence, we will uncover the truth about IQ and the limitations of the human intellect. Beyond traditional measures of intellect, this article explores a wider range of aptitudes that impact our perception and engagement with the natural world.

March 28, 20245 min read

Research & Evidence

Myopia and IQ

Myopia and IQ: are nearsighted people really smarter? A 2018 genetic study of 68,000 people found the causal arrow runs from more education to more myopia, not the other way around. See the historical studies, the genetics behind the link, and why time outdoors helps.

September 28, 20267 min read