Anxiety and IQ: what anxiety does to a score, and what a score says about anxiety

0 Quick Answer

Anxiety at the time of testing lowers performance on working memory and reasoning tasks by a measurable amount, test anxiety is a documented cause of lower test scores, and the popular claim that intelligent people are more anxious rests on surveys of high IQ society members that representative cohorts do not confirm; those cohorts find the opposite. Moran's meta-analysis in Psychological Bulletin pooled 177 samples and 22,061 people and found that self-reported anxiety was reliably associated with poorer working memory capacity, at g = 0.334 in the harmful direction, with the effect present on complex span tasks, on simple span tasks such as digit span, and most strongly on dynamic tasks such as the n-back. Hembree's integration of 562 studies in 1988 concluded that test anxiety causes poor performance and that reducing it improves test scores and grades, and von der Embse and colleagues' 30 year update across 238 studies found test anxiety negatively related to standardized tests, entrance examinations and grade point average, with effects in the small to moderate range and largest in the middle school years.

The second question is different and so is the answer. Karpinski and colleagues surveyed 3,715 members of American Mensa and reported elevated rates of anxiety and mood disorders, and a preregistered European replication of 615 Mensa members found a risk ratio of 3.82 for generalized anxiety, with both papers noting that membership of a high IQ society, not high IQ, may explain the result. Representative cohorts, which do not have that problem, point the other way. In the Dunedin birth cohort of 1,037 people followed to 32, lower childhood IQ predicted adult anxiety as well as depression and schizophrenia spectrum disorder. In the Providence cohort, a 15 point advantage in cognitive performance at seven was associated with a 50 percent lower risk of lifetime generalized anxiety disorder. In 1,049,663 Swedish conscripts, each standard deviation of lower IQ carried a hazard ratio of 1.51 for hospital admission with a neurotic disorder. And Brown, Wai and Chabris's analysis of 48,558 people in four representative cohorts found no support for any downside to higher ability on any outcome.

The two findings are compatible, and the reconciliation matters for a person reading their own score. Anxiety is a state that competes with the task in front of a person, and a score taken under it is lower than the same person's score taken calm. Trait anxiety and anxiety disorders are, in the population, somewhat more common at lower levels of measured ability, through routes the cohort studies discuss. Neither fact says anything about an individual, and the sections below take each design in turn.

1 Two Questions That Get Confused

The phrase anxiety and IQ covers a question about a score on a day and a question about a person across decades, and almost every confident claim in circulation answers one while sounding like the other. The first question is whether anxiety at the time of testing changes what a person can do on the tasks an IQ test contains: hold a string of digits, rotate a figure, complete a matrix under a clock. That is a question about state anxiety, including the specific form called test anxiety, and it is answered by experiments and by studies that measure anxiety and performance in the same session. The second question is whether people who are more anxious in general, or who have an anxiety disorder, differ in measured intelligence from people who are not, and in which direction. That is a question about trait anxiety and diagnosis, and it can only be answered by cohorts that measured ability before the anxiety developed.

The distinction is the same one the page on what IQ measures draws between performance and ability. A test score is a performance on a day, and anything that competes for attention on that day, anxiety among them, lowers it; the ability the test is designed to estimate is not changed by an anxious morning. The page on reliability and validity explains how the standard error of measurement expresses the expected wobble of a score from occasion to occasion, and state anxiety is one of the things inside that error. The page on IQ and mental health sets out the broader landscape of ability and psychiatric outcomes within which the second question sits, and the page on IQ and depression treats the disorder that most often travels with anxiety.

Keeping the questions apart also explains why the literature seems to contradict itself. A study that finds anxious students scoring lower on a test is measuring the state effect and the trait association at once, and cannot tell how much of the gap is anxiety on the day and how much is a difference that was there before. The meta-analyses in the next sections are the state evidence; the cohorts later on are the trait evidence; and the Mensa surveys, which are the source of the claim that intelligence brings anxiety, belong to neither category cleanly, which is their problem.

2 Anxiety and Working Memory: The 2016 Meta-Analysis

Working memory is the capacity an IQ test draws on most directly, and the largest synthesis of the evidence found that anxious people have measurably less of it available, on every kind of task used to measure it. Moran's 2016 meta-analysis pooled 177 samples with 22,061 individuals in which self-reported anxiety, trait or state, was related to performance on working memory capacity tasks. The overall effect was g = 0.334 in the harmful direction, highly significant. It held across the three families of task: complex span tasks such as the operation span, at g = 0.342 over 30 samples and 3,196 people; simple span tasks such as digit span, at g = 0.318 over 127 samples and 17,547 people; and dynamic tasks such as the n-back, at g = 0.437 over 20 samples and 1,318 people. The narrative review that accompanied the meta-analysis found that anxiety, whether self-reported or induced in the laboratory, was related to poorer performance across a wide variety of other tasks as well.

The mechanism the field works from is attentional control theory, set out by Eysenck, Derakshan, Santos and Calvo in Emotion in 2007. On that account, anxiety diverts attention toward threat and toward worry, which impairs the inhibition of irrelevant material and the shifting of attention between tasks, and it does so more reliably to the efficiency of processing, how much effort a result costs, than to its effectiveness, the result itself. Anxious people can often reach the same answer by working harder, until the task is timed or the load is high, at which point the extra effort is no longer available and performance falls. That prediction fits the pattern in the meta-analysis, where the dynamic tasks that demand continuous updating show the largest effect.

For a battery, the implication is specific. The page on the six cognitive domains describes the Working Memory Index, and in ACIS the digit span and letter number tasks that build it are exactly the simple and complex span tasks in the meta-analysis. A third of a standard deviation on those tasks is about five points on an index scaled with a standard deviation of 15, by our arithmetic, which is larger than the standard error of the index and would be visible in a profile as a working memory score sitting below the reasoning scores. The page on IQ and memory explains why that pattern can arise for several reasons, and state anxiety is one that a retest under calmer conditions can distinguish from the others.

3 Test Anxiety and Test Scores: Sixty Years of Studies

Test anxiety is the one form of anxiety that has been studied against test scores directly for six decades, and the two large syntheses agree that it lowers them, that the effect is small to moderate, and that reducing it raises scores. Hembree integrated 562 studies in the Review of Educational Research in 1988 and drew four conclusions that have held: test anxiety causes poor performance; it relates inversely to self-esteem and directly to fear of negative evaluation, defensiveness and other forms of anxiety; ability, sex and grade level are among the conditions that produce different levels of it; and a variety of treatments reduce it, with improved test performance and grade point average consistently accompanying the reduction, contrary to the prior view that treated anxiety and performance as separable. The last finding is the one that establishes direction, because an intervention that lowers anxiety and raises scores is evidence that the anxiety was doing the lowering.

Von der Embse, Jester, Roy and Post updated the literature in the Journal of Affective Disorders in 2018 with 238 studies published since Hembree. Test anxiety was significantly and negatively related to standardized tests, university entrance examinations and grade point average, most strongly in the middle grades. Self-esteem was a strong predictor of test anxiety, and perceived difficulty and the high stakes of a test were related to higher anxiety. The pooled relationships were in the small to moderate range, which is the honest size of the effect: a real cost, not a catastrophe, and one that varies with the person and the stakes.

The page on how to prepare for an IQ test applies this in practice, and the page on time limits explains why timed subtests are where anxiety costs most, since the clock removes the option of compensating with effort that attentional control theory predicts. The stakes finding has a practical corollary: a test taken for one's own information, with the option to retake, carries less of the high stakes anxiety that the meta-analysis associates with the largest effects than an entrance examination does, which is one reason a self-administered battery can produce a cleaner estimate for an anxious person than a single supervised session.

4 Choking Under Pressure: The Ablest Lose the Most

Pressure does not lower everyone's performance equally, and the experimental evidence shows it costs most to the people with the most working memory to lose, on the problems that need it most. Beilock and Carr reported in Psychological Science in 2005 on mathematical problem solving under pressure induced by incentives, observation and evaluation. Only participants high in working memory capacity were harmed by the pressure, and their decrements were confined to the problems with the highest working memory demands. The account is that high capacity individuals solve hard problems by strategies that rely on working memory, and pressure consumes exactly that resource, so that the advantage they hold in calm conditions is the first thing to go. Low capacity participants, who were already using strategies that did not depend on it, had nothing to lose.

The best known intervention built on that account did not survive replication. Ramirez and Beilock reported in Science in 2011 that ten minutes of writing about testing worries immediately before an examination improved the scores of habitually anxious students in a classroom. Camerer and colleagues' replication project in Nature Human Behaviour in 2018 attempted the classroom experiment twice with well powered samples and found no benefit of expressive writing on test performance, with effect sizes slightly negative. The finding sits beside the smartphone presence effect and the literary fiction effect as a widely repeated result that a direct test did not reproduce, and it should not be offered as a remedy.

What survives is the choking result and Hembree's broader finding that treatments which reduce test anxiety over time do raise performance. The page on processing speed explains why speeded tasks are the most sensitive to any competing demand, and the page on the working memory test describes the span tasks on which pressure operates. A person with a strong reasoning profile who scores oddly low on a timed working memory task under stressful conditions is showing the Beilock and Carr pattern, not a hidden deficit.

5 Math Anxiety: A Domain Where the Cost Is Measured

Anxiety about a specific domain lowers measured performance in that domain by an amount that two recent meta-analyses agree on, and the size is a guide to what test anxiety does to any quantitative subtest. Namkung, Peng and Lin pooled 131 studies with 478 effect sizes in the Review of Educational Research in 2019 and found a correlation of 0.34 in the negative direction between mathematics anxiety and mathematics performance among school aged students. Barroso, Ganley, McGraw, Geer, Hart and Daucourt pooled 747 effect sizes in Psychological Bulletin in 2021 and found a small to moderate negative correlation of 0.28. Both are correlations in populations, and both mix the state effect of anxiety on the day with the trait association between avoiding mathematics and being anxious about it, but their agreement fixes the size of the relationship: a fifth to a third of a standard deviation, which on a quantitative index scaled with a standard deviation of 15 is three to five points, by our arithmetic.

The mechanism proposed for math anxiety is the same working memory competition described above, and the direction runs both ways. A child who is anxious about arithmetic performs worse, avoids practice, falls behind and becomes more anxious, which is the anxious analogue of the reading spiral described on the page on reading and IQ. A quantitative subtest cannot separate the two, and a person who has avoided mathematics for years will score lower on it for reasons of practice as well as of anxiety.

The page on IQ and academic achievement explains how quantitative reasoning contributes to a full scale score, and the page on the arithmetic subtest describes what a mental arithmetic task under time pressure requires. A person who knows that their anxiety is specific to numbers has, in these meta-analyses, a reason to expect the quantitative index to underestimate their reasoning by a few points, and a reason to read the nonverbal and verbal indices as the better guide to it.

6 Does High Intelligence Bring Anxiety? The High IQ Society Surveys

The claim that intelligent people are more anxious comes almost entirely from surveys of people who joined a high IQ society, and both of the surveys that produced it warn that the society, not the intelligence, may be what they measured. Karpinski, Kinase Kolb, Tetreault and Borowski surveyed 3,715 members of American Mensa and reported in Intelligence in 2018 that self-reported diagnosed and suspected mood and anxiety disorders, attention deficit disorder, autism spectrum conditions, allergies, asthma and autoimmune disease were more prevalent than in national comparison data, and proposed a hyper brain and hyper body theory in which overexcitabilities predispose the highly intelligent to psychological and physiological conditions. Fries, Baudson, Kovacs and Pietschnig ran a preregistered survey of 615 Mensa members in Austria, Germany, Hungary, Switzerland and the United Kingdom and reported in Frontiers in Psychology in 2022 elevated prevalences of autism spectrum conditions, chronic fatigue, depression and generalized anxiety, the last at a risk ratio of 3.82, but no elevation of asthma, allergies or autoimmune disease, and little evidence for the immunological explanation.

Both papers state the alternative that undoes the inference. People who join a high IQ society are not a random sample of people with high IQs; they are people who sought out a society organized around a test score, which is more likely for those who want an explanation for feeling different, who have time and reason to pursue it, and who are willing to fill in a health survey. The comparison groups were national samples that had not been selected on any of those things. A risk ratio computed between the two measures the selection as much as the trait, and neither survey measured the IQ of a representative population against its anxiety. The page on high IQ society requirements describes who qualifies and how, and the page on signs of high intelligence addresses the folk belief the surveys appear to confirm.

Smaller studies are sometimes cited alongside them. Coplan and six colleagues reported in Frontiers in Evolutionary Neuroscience in 2012 that within a small group of patients with generalized anxiety disorder, IQ correlated positively with worry, and Penney, Miedema and Mazmanian reported in Personality and Individual Differences in 2015 that verbal intelligence correlated with worry and rumination in a sample of undergraduates. Both are small, cross-sectional and drawn from populations selected on anxiety or on enrolment, and the page on IQ and personality explains why a within group correlation of that kind does not describe the population relationship, which the cohorts below measure directly.

7 What the Representative Cohorts Show

When ability is measured in childhood in a representative sample and anxiety is diagnosed decades later, higher ability predicts less anxiety, and the finding repeats across New Zealand, the United States and Sweden with samples from a thousand to a million. Koenen, Moffitt, Roberts, Martin, Kubzansky, Harrington, Poulton and Caspi followed the Dunedin birth cohort of 1,037 people born in 1972 and 1973, with WISC-R IQ measured at 7, 9 and 11 and research diagnoses made at 18, 21, 26 and 32 with 96 percent retention, and reported in the American Journal of Psychiatry in 2009 that lower childhood IQ was associated with increased risk of schizophrenia spectrum disorder, adult depression and adult anxiety, with greater comorbidity and with persistence of depression; the association with persistence of generalized anxiety disorder was nearly significant, and higher childhood IQ predicted increased risk of mania. The authors framed the result as support for cognitive reserve: lower ability is an antecedent of several common disorders.

Martin, Kubzansky, LeWinn, Lipsitt, Satz and Buka analysed 689 members of the Providence site of the National Collaborative Perinatal Project, followed from birth into their mid thirties, and reported in the International Journal of Epidemiology in 2007 that a 15 point advantage in cognitive performance at age seven was associated with a 50 percent reduction in the risk of lifetime generalized anxiety disorder, and with reductions of 89 and 57 percent in the risk of onset in childhood and adolescence, after adjustment for socioeconomic status and parental history of mental health problems, and unaffected by childhood behavioural inhibition or learning disabilities. Gale, Batty, Tynelius, Deary and Rasmussen followed 1,049,663 Swedish men from their conscription intelligence tests for a mean of 22.6 years and reported in Epidemiology in 2010 that the risk of hospital admission rose for every category of mental disorder with each point decrease on the nine point IQ scale, with age adjusted hazard ratios per standard deviation of 1.51 for neurotic disorders, 1.50 for mood disorders, 1.60 for schizophrenia and 1.75 for personality disorders, changing little on adjustment for confounders.

Wraw, Deary, Der and Gale added a nuance in Intelligence in 2016 from 5,793 members of the American 1979 youth cohort: higher intelligence at 15 to 23 predicted fewer depressive symptoms, fewer sleep difficulties and better mental health status at 50, but a slightly higher probability of reporting a diagnosis of depression, an odds ratio of 1.11 that rose to 1.32 after adjustment for adult socioeconomic status. The likeliest reading, which the authors discuss, is that more able people are more likely to seek and receive a diagnosis for the same symptoms, and that a diagnosis count can therefore run opposite to a symptom count. Brown, Wai and Chabris's analysis of 48,558 people in four representative cohorts from 1957 onward, published in Perspectives on Psychological Science in 2021, tested the general claim that ability stops helping or begins to harm beyond some threshold across occupational, educational, health and social outcomes, and found most effects linear or null, nearly all nonlinear effects negligible or unreplicated, and no support for any downside to higher ability. The page on IQ and success and the page on IQ and longevity set those results in the wider cognitive epidemiology.

8 Direction and Mechanism: Reserve, Reporting and Reverse Causation

The cohort associations are consistent, but a consistent association is not a mechanism, and three explanations share the evidence. The first is cognitive reserve, the hypothesis Koenen and colleagues tested by name: higher ability provides more resources for appraising threat, solving problems and regulating emotion, so that the same stressors produce less disorder, and the Martin finding that the protective association is strongest for childhood onset fits an account in which ability shapes how a child interprets ambiguity. The second is confounding by shared causes: early adversity, poverty and poor health lower both childhood test scores and later mental health, and although the cohorts adjust for socioeconomic status and family history, no adjustment is complete. The page on what heritability actually means explains why genetic overlap between ability and liability to anxiety would also produce the association without either causing the other.

The third is reverse causation at the time of testing, and it is the one that joins the two halves of this page. A child who is anxious at seven performs below their ability on the day, as the working memory meta-analysis predicts, and if that anxiety is the beginning of a disorder, the low score and the later diagnosis share a cause that the score did not measure. The Providence study addressed this by showing that behavioural inhibition in childhood did not alter the result, and the Dunedin study by averaging three assessments across four years, which reduces the weight of any one anxious morning. Neither can exclude it entirely, and a reader should hold the cohort findings as associations between measured ability and later anxiety that run partly through reserve, partly through shared causes and partly, perhaps, through anxiety's effect on the measurement itself.

The reporting artefact in the American cohort deserves separate weight, because it explains the Mensa surveys as well. A group that is more likely to seek assessment, to know the names of conditions and to answer a survey about them will report more diagnoses at the same level of symptoms, and a high IQ society is such a group by construction. The page on IQ and personality reviews the modest negative correlation between measured ability and neuroticism in general population samples, which is the trait level version of the cohort finding and the opposite of the folk claim.

9 The Evidence by Design

Read by design rather than by headline, the anxiety literature sorts into meta-analyses of state effects on performance, experiments on pressure, self-selected surveys of high IQ societies, and representative cohorts, and the last two disagree because only one of them sampled the population.

The column to read is the sample. The meta-analyses of state effects sampled people taking tasks and found anxiety lowering performance. The surveys sampled members of a society and found more reported anxiety, which the society's selection can explain. The cohorts sampled populations and found the reverse of the surveys.

10 What the Evidence Supports, Stated Narrowly

Stated as narrowly as the designs allow, the evidence supports six sentences about anxiety and IQ, and the sixth is the one that the popular claim gets backwards. First, anxiety at the time of testing is associated with lower working memory performance by about a third of a standard deviation across 22,061 people, on the span tasks that an IQ battery uses, and the effect is largest on tasks that demand continuous updating. Second, test anxiety lowers scores on standardized tests, entrance examinations and grades by a small to moderate amount, the effect is causal in the sense that treatments which reduce anxiety raise performance, and it is largest where stakes are high and in the middle school years. Third, performance pressure harms the people with the most working memory to lose, on the hardest problems, and the brief writing intervention proposed to counter it did not replicate. Fourth, anxiety specific to mathematics correlates with mathematics performance at about 0.3 in the negative direction, a relationship that runs in both directions through avoidance and practice. Fifth, members of high IQ societies report more anxiety and mood disorders than national samples, and both of the studies that found this state that membership rather than intelligence may explain it. Sixth, in representative cohorts that measured ability in childhood or youth and diagnosed disorder decades later, higher ability predicts less anxiety and less mental disorder of almost every kind, with no evidence of any threshold beyond which ability stops helping.

What the evidence does not support is the claim that intelligent people are, in the population, more anxious than others, because no representative study finds it and several find the opposite. It does not support the claim that anxiety lowers intelligence, as opposed to lowering a score, because the state effects are effects on performance under load and are reversible by conditions. It does not support the writing intervention or any other quick fix for test anxiety, because the one tested at scale failed, while it does support the slower treatments that Hembree's synthesis found effective. And it does not support the reading of a low working memory index in an anxious person as a deficit, because the meta-analysis predicts exactly that pattern from state anxiety alone.

The narrow statement leaves a clear division of labour. For a score, anxiety is a condition of measurement to be managed, and the standard error is the tool for reading how much a single session can be trusted. For a person's standing over a lifetime, anxiety is not a marker of ability in either direction, and the cohorts say that if anything the correlation runs toward calm.

11 What This Does Not Say About You

No finding on this page tells an anxious reader what their IQ is, and the ones that come closest tell them how much a score taken while anxious may understate it. A pooled effect of a third of a standard deviation is an average across 22,061 people; a given person may lose nothing on a calm day and a good deal on a bad one, and the page on the standard deviation of 15 puts a third of a standard deviation, about five points, beside the standard error of a single index to show that the two are of the same order. A hazard ratio in a million conscripts describes a population gradient; a person with generalized anxiety disorder is not told their score by it, and a person with a high score is not told they will be calm. The cohorts and the meta-analyses are population statements and the page on score versus percentile explains how to read an individual result against them without importing either.

For a person who wants a score that reflects their ability rather than their morning, the evidence points to conditions rather than to worry about the finding. A normed battery taken when rested, without a clock running on more than the subtest in front of them, and with the option of spreading the work over several sittings removes most of the state effects this page describes. ACIS runs twenty timed subtests that a person can take in sessions across thirty days rather than in one supervised block, scores them against an adult reference frame of 3,243 records aged 16 to 90 described in the technical manual, and prints the standard error of every index, so that a working memory score sitting below the reasoning scores can be read as the profile it is rather than as a verdict. The retake policy exists for the case in which a session was clearly compromised, and a second administration under calmer conditions is the only instrument that separates an anxious day from a real pattern.

Two cautions complete the picture. A person with a diagnosed anxiety disorder should expect the working memory and speed indices to be the most sensitive to their state and should weight the reasoning indices accordingly, which is the same advice the page on ADHD and IQ gives for a different reason. And a person who scores highly and reads the Mensa surveys as an explanation for their anxiety should know that the surveys measured the society, not the score, and that the population evidence offers no such explanation and no such consolation. The relationship between a person's anxiety and their ability is, on the evidence, mostly a relationship between anxiety and the day of the test.

12 Sources Behind This Page

Every figure above is traceable to one of the following, and each is linked at the point where it is used. Effect sizes, hazard ratios, risk ratios and sample sizes are quoted from the abstracts and results of the papers as published, and the conversion of a standardized effect to index points is our arithmetic and is labelled as such where it appears.

• Moran T P. Anxiety and working memory capacity: A meta-analysis and narrative review. Psychological Bulletin, 2016, volume 142, issue 8, pages 831 to 864.
• Hembree R. Correlates, causes, effects, and treatment of test anxiety. Review of Educational Research, 1988, volume 58, issue 1, pages 47 to 77.
• von der Embse N, Jester D, Roy D and Post J. Test anxiety effects, predictors, and correlates: A 30-year meta-analytic review. Journal of Affective Disorders, 2018, volume 227, pages 483 to 493.
• Eysenck M W, Derakshan N, Santos R and Calvo M G. Anxiety and cognitive performance: Attentional control theory. Emotion, 2007, volume 7, issue 2, pages 336 to 353.
• Beilock S L and Carr T H. When high-powered people fail: Working memory and "choking under pressure" in math. Psychological Science, 2005, volume 16, issue 2, pages 101 to 105.
• Ramirez G and Beilock S L. Writing about testing worries boosts exam performance in the classroom. Science, 2011, volume 331, issue 6014, pages 211 to 213.
• Camerer C F and colleagues. Evaluating the replicability of social science experiments in Nature and Science between 2010 and 2015. Nature Human Behaviour, 2018, volume 2, issue 9, pages 637 to 644.
• Namkung J M, Peng P and Lin X. The relation between mathematics anxiety and mathematics performance among school-aged students: A meta-analysis. Review of Educational Research, 2019, volume 89, issue 3, pages 459 to 496.
• Barroso C, Ganley C M, McGraw A L, Geer E A, Hart S A and Daucourt M C. A meta-analysis of the relation between math anxiety and math achievement. Psychological Bulletin, 2021, volume 147, issue 2, pages 134 to 168.
• Karpinski R I, Kinase Kolb A M, Tetreault N A and Borowski T B. High intelligence: A risk factor for psychological and physiological overexcitabilities. Intelligence, 2018, volume 66, pages 8 to 23.
• Fries J, Baudson T G, Kovacs K and Pietschnig J. Bright, but allergic and neurotic? A critical investigation of the "overexcitable genius" hypothesis. Frontiers in Psychology, 2022, volume 13, article 1051910.
• Koenen K C, Moffitt T E, Roberts A L, Martin L T, Kubzansky L, Harrington H, Poulton R and Caspi A. Childhood IQ and adult mental disorders: A test of the cognitive reserve hypothesis. American Journal of Psychiatry, 2009, volume 166, issue 1, pages 50 to 57.
• Martin L T, Kubzansky L D, LeWinn K Z, Lipsitt L P, Satz P and Buka S L. Childhood cognitive performance and risk of generalized anxiety disorder. International Journal of Epidemiology, 2007, volume 36, issue 4, pages 769 to 775.
• Gale C R, Batty G D, Tynelius P, Deary I J and Rasmussen F. Intelligence in early adulthood and subsequent hospitalization for mental disorders. Epidemiology, 2010, volume 21, issue 1, pages 70 to 77.
• Wraw C, Deary I J, Der G and Gale C R. Intelligence in youth and mental health at age 50. Intelligence, 2016, volume 58, pages 69 to 79.
• Brown M I, Wai J and Chabris C F. Can you ever be too smart for your own good? Comparing linear and nonlinear effects of cognitive ability on life outcomes. Perspectives on Psychological Science, 2021, volume 16, issue 6, pages 1337 to 1359.

13 Frequently Asked Questions