Does practice beat IQ? What the 10,000-hour rule and the meta-analyses found

0 The short answer

Practice and IQ are not rivals: the best evidence says deliberate practice matters a great deal in some domains, explains well under half of the differences between people in every domain studied, and works alongside cognitive ability instead of replacing it. The 2014 meta-analysis of 88 studies found that practice explained 26 percent of the variance in performance in games, 21 percent in music, 18 percent in sports, 4 percent in education and under 1 percent in professions, and its authors concluded that practice is important but not as important as has been argued. The researcher who proposed deliberate practice answered in 2019 that those studies counted far looser kinds of practice than his original definition, so part of the dispute is about definitions and not about whether practice matters. The 10,000-hour figure was never a threshold that the original research found.

1 Where Did the 10,000-Hour Rule Come From, and What Did the Violin Study Find?

The 10,000-hour rule is a popular simplification of a 1993 study of violin students, and the researcher behind that study has said in print that the number was never a magic threshold. The study is the paper by Ericsson, Krampe and Tesch-Römer in Psychological Review. It compared students at a music academy in West Berlin who had been grouped by their teachers as the best violinists, good violinists or less accomplished, and it asked each student to estimate how many hours per week they had practiced alone in every year since they started. Summing those estimates produced accumulated hours of solitary practice, and the groups differed in them. The authors concluded that individual differences in ultimate performance could largely be accounted for by differential amounts of past and current practice, a sentence that every later critic and defender quotes.

Several things are easy to miss in that design. The estimates were retrospective, so each student reconstructed years of practice from memory. The best and good groups had ten students each in the original analysis, as described in the 2019 replication by Macnamara and Maitra. And the number that matters for the popular rule is lower than 10,000: the replication paper reports that the best violinists in the original study had a mean of 7,410 hours of practice alone by age 18, against 5,301 hours for the good violinists. In 2014 Ericsson published the 95 percent confidence interval around the best group's mean, which the replication reports as 2,894 to 11,926 hours, an interval wide enough to contain the good group's mean.

The popular rule arrived 15 years later. According to Ericsson and Harwell, Malcolm Gladwell's 2008 book Outliers devoted a chapter to a 10,000-hour rule and cited the 1993 paper as its main evidence, calling the figure the magic number for true expertise. Ericsson and Harwell reject that reading in their own words: their research showed "no evidence for a magical number." They add that the first author estimated about 25,000 hours as the more accurate figure for winning international piano competitions, that Gladwell never used the term deliberate practice, and that the examples in the book included kinds of practice that did not meet the original criteria. They also report, citing other work, that performers competing at the international level have often accumulated between 3,000 and 40,000 hours before reaching their highest level. A rule with a spread from 3,000 to 40,000 is not a rule about any one person.

What the original definition required matters for everything that follows. In the 1993 paper's own wording as quoted by Ericsson and Harwell, the activity has an explicit goal of improving performance, the person should receive immediate informative feedback, and the person should repeatedly perform the same or similar tasks. Ericsson and Harwell stress that in the music academy the practice that fit this description was solitary practice, often shaped by a teacher. Group practice, which Ericsson and Harwell exclude in their reanalysis, and simply logging years in a field do not fit it. That strict definition is the hinge of the dispute covered in the fourth section, and it is also why the question of practice versus intelligence as a general ability cannot be settled by a single hour count. Hours of practice measure exposure to training. They do not measure the knowledge that results from it, which is a different construct from intelligence.

2 What Did the 2014 Meta-Analysis of Practice and Performance Find in Each Domain?

Across 88 studies, deliberate practice explained a substantial share of the differences in performance in games, music and sports, a small share in education, and almost none in professions. The meta-analysis by Macnamara, Hambrick and Oswald in Psychological Science covered every major domain in which deliberate practice had been studied. Its abstract reports that deliberate practice explained 26 percent of the variance in performance for games, 21 percent for music, 18 percent for sports, 4 percent for education and less than 1 percent for professions, and concludes that deliberate practice is "important, but not as important as has been argued." The Association for Psychological Science's summary of the paper counts 88 studies and puts the overall share at about 12 percent of individual differences in performance. Ericsson and Harwell, writing five years later, describe the same paper as finding 14 percent. The two readings differ, and the domain figures in the abstract are the ones both sides use.

The table puts those figures beside the reanalysis that Ericsson and Harwell published, which kept only the effect sizes that met their criteria for deliberate practice, and beside the kind of IQ evidence this page can offer for each domain.

The table shows three features that the headline numbers hide. The first is that both counts put professions last, so the camps disagree about size more than about whether practice matters at the bottom of the list. The second is that the stricter reanalysis does not simply raise every domain by the same amount: in it, music comes out highest of the three domains it covers and games lowest, whereas the 2014 meta-analysis ranked games first. The third is that the reanalysis could keep only 14 effect sizes in total, which is our arithmetic on the counts it lists, and that is a small basis for a claim about all expertise. The papers cited on this page discuss both the definition of practice and the restricted range of skill in many samples as reasons the numbers differ between domains, and neither point is settled.

A further point comes from the sports meta-analysis by the same research group. In the 2016 meta-analysis by Macnamara, Moreau and Hambrick, deliberate practice accounted for 18 percent of the variance in sports performance overall, but for only 1 percent among elite-level performers. The same paper found that athletes who reached a high level of skill did not begin their sport earlier in childhood than lower-skill athletes. The first result fits the idea that once everyone in a sample has practiced a lot, practice no longer separates them. The second challenges a popular corollary of the rule, that the future elite are the ones who started earliest.

3 What Does "Percent of Variance Explained" Mean, and What Does It Not Mean?

A figure such as 26 percent describes how much of the spread between people in a sample lines up with their spread in practice, and it says nothing about how many hours any one person needs. The statistic comes from squaring a correlation. If people who practiced more also performed better, and the correlation between the two measures is moderate, the squared correlation is the share of the variation in performance that moves with the variation in practice. It is a statement about a group at one moment, measured with error, and it can be small even when the average effect of practice on a person is large, as when everyone in the sample practiced a lot and only the remaining differences are being explained. It can also be large without proving that practice caused the difference, because people who are good at something tend to practice it more.

Three misreadings are common. The first is that 26 percent of a person's skill comes from practice and 74 percent from talent. A variance share is not a recipe, and the remaining share is not a single thing called talent. It contains everything the studies did not measure, including measurement error in the practice estimates, age of starting, coaching quality, health, motivation, opportunity and cognitive ability. The second misreading is that a small share means practice is unimportant. The 2014 authors say the opposite, and the critical commentary by Hambrick and colleagues says that deliberate practice matters but that most of the variance in performance is explained by other factors. The third misreading is the one most relevant to the original rule: that the share of variance tells you how many hours a person needs.

The study that shows why the third is wrong is the chess study by Gobet and Campitelli, which asked 104 Argentinian players, from weak amateurs to grandmasters, about their practice, starting age and handedness. It reaffirmed the importance of practice for reaching a high level, and it also found large variability: the slower player needed 8 times as much practice to reach master level as the faster player. The authors read the result as showing that practice is necessary but not sufficient. A rule that says "about 10,000 hours" and a data set in which the hours to reach one level differ eightfold between individuals cannot both be taken as a statement about what one person will need.

Measurement adds a second layer. Retrospective practice estimates are not perfectly reliable, and a correlation computed from an unreliable measure understates the relationship. The group around Hambrick and the group around Ericsson both discuss correcting for that, and they reach different numbers. Hambrick and colleagues report that after their correction practice accounted for about one-third of the reliable variance in each of chess and music, and for 30 percent in music, which is the figure they defended against a critique of the music figure. Ericsson and Harwell report 29 percent for their narrower set of studies and about 61 percent after correcting for attenuation. The size of the correction depends on what is assumed about the reliability of practice estimates, which is the same concept the page on reliability and validity explains for test scores. Both groups therefore agree that correcting for measurement error raises the estimates. They disagree about which studies count as deliberate practice, and that choice moves the answer more than any statistical adjustment does.

4 What Did Ericsson Answer, and Did the Violin Study Replicate?

Ericsson's answer is that the 2014 meta-analysis counted the wrong things as deliberate practice, and the replication of his 1993 study found a smaller effect than the original, which is why neither side can claim the number. In their 2019 paper, Ericsson and Harwell state that the meta-analysis used a definition that differs from the original and call what it measured structured practice. They then apply three criteria: the outcome must be reproducibly superior performance in a recognized domain of expertise, the practice counted must have been designed to improve that performance, and it must have been solitary or under individualized supervision, not team or group practice. After excluding the effect sizes that failed those criteria, 14 remained by our count: 5 in games, 3 in music, 5 in sports and 1 in education, with none in professions. Their random-effects average correlation was 0.54, with a 95 percent confidence interval of 0.44 to 0.63, which they describe as about 29 percent of the variance, and about 61 percent after correcting for attenuation.

The music figure has its own history. A 2014 meta-analysis of music by Platz and colleagues pooled 13 studies with 788 participants and found a corrected correlation of 0.61, with a 95 percent confidence interval of 0.54 to 0.67, between accumulated task-relevant practice and musical achievement. Hambrick and colleagues, in their commentary, report that Platz and colleagues objected to reading that correlation as 36 percent of variance, on the argument that meta-analytic effects should be reported as correlations and not squared. Hambrick and colleagues call that criticism puzzling and keep their reading that practice accounted for 30 percent of the variance in music performance. So the music dispute includes a statistical convention as well as a definition.

The most direct test is a replication. Macnamara and Maitra, in the 2019 paper in Royal Society Open Science, repeated the violin study at the Cleveland Institute of Music with changes meant to reduce bias: a double-blind procedure in which the interviewers did not know the students' skill group and the students were not told the purpose, analyses suited to a three-group design, and a new measure of teacher-designed practice. The best group had 13 students, against 10 in the original. The authors report that they did not replicate the core finding that accumulated deliberate practice corresponded to each skill level. The effect was substantial but considerably smaller than in the original: for accumulated practice alone to age 18 the effect size was 0.26, with a 95 percent confidence interval of 0.03 to 0.44, set against 0.48 for the original study. Teacher-designed practice, the stricter form of the concept, did not explain more variance than practice alone. With samples this small the interval is wide, and the authors present the result as a challenge to a seminal finding, not as a verdict on practice.

Read together, these papers support a modest and defensible conclusion. All parties agree that long-term practice is required and that it is associated with higher performance. They disagree about whether the studies that find smaller associations measured the right thing, about whether small samples and retrospective estimates make the original result look stronger than it is, and about which statistic to report. That is a real scientific dispute, and it should make anyone suspicious of a page that gives one number as settled. It also leaves a gap that none of these practice papers fills. They compare practice with performance, and almost none measures cognitive ability, which is why the rest of this page turns to studies that do.

5 What Does the Meta-Analysis of Cognitive Ability and Chess Skill Show?

Chess is the domain where practice and cognitive ability have been measured together most often, and the pooled correlations between cognitive tests and chess skill are positive, small to medium, and weaker among adults and ranked players. The meta-analysis by Burgoyne and colleagues in the journal Intelligence pooled 19 studies with 82 effect sizes and 1,779 participants. The players covered a wide range of strength: the average Elo rating was 2018, and ratings ran from 1311, an amateur level, to 2607, an elite level. In the five studies that reported full-scale IQ, the weighted mean was 120.5, with an average standard deviation of 14.8. That is a selected group, and the authors note that restricted ranges of both ability and skill limit how large a correlation can be.

The abstract reports that chess skill correlated positively and significantly with fluid reasoning (Gf, r = 0.24), comprehension-knowledge (Gc, r = 0.22), short-term memory (Gsm, r = 0.25) and processing speed (Gs, r = 0.24), with a meta-analytic average of 0.24. The authors translate that into about 6 percent of the variance in chess skill explained on average, and 5 percent for Gc. A model that included only full-scale IQ tests gave a non-significant correlation of 0.10, with a 95 percent confidence interval from minus 0.19 to 0.38, which the authors say means full-scale IQ explained less than 1 percent of the variance in chess skill, and which they call somewhat surprising given the other results. The moderators matter more than the averages. The Gf correlation was 0.32 in youth samples and 0.11 in adult samples, and 0.32 in unranked samples against 0.14 in ranked samples. Among the ability types, numerical ability correlated 0.35 with chess skill, verbal ability 0.19 and visuospatial ability 0.13. The authors published a corrigendum in 2018 that changes some of these estimates, and the page on chess and IQ gives the corrected values and the instruction studies in detail. The figures here are those of the 2016 paper as first published.

The paper's introduction quotes a 2007 Harvard Business Review article by Ericsson, Prietula and Cokely, which claimed that there is no correlation between IQ and expert performance in fields such as chess, music, sports and medicine. The pooled data contradict that claim for chess, since most of the correlations (79 percent) were positive. But Ericsson and Harwell offer a reading of the pattern that deserves a fair hearing. They describe the 2016 finding as a substantial correlation for beginners and less-skilled players that was no longer significant for highly skilled players, and they argue that there is an accumulating body of evidence for a gradual disappearance of correlations between ability tests and performance as domain-specific mechanisms are acquired. Burgoyne and colleagues add a caution of their own: samples of strong players are range restricted in Elo rating, which pushes the correlation down mechanically. Both explanations predict a weaker correlation among experts, and the data cannot separate them.

Two other chess papers put the pieces together. In a reanalysis of a key data set, Burgoyne, Nye, Macnamara, Charness and Hambrick modeled latent variables for serious chess activity and formal instruction, plus a variable for starting age, and found that together they accounted for 63 percent of the variance in peak chess rating, with a later serious starting age still associated with a lower peak rating after the training variables were included. That is much larger than the 26 percent for games in the 2014 meta-analysis, and the contrast is instructive: a latent variable captures what several measures of training have in common, and a model with instruction and starting age includes more than hours alone. The model has no cognitive variable in it, so it cannot say how intelligence would change the picture.

Vaci and colleagues did include one. In a longitudinal study that tracked chess players through their careers, both intelligence and practice positively affected the acquisition and retention of chess skill, and the two interacted nonlinearly: more intelligent individuals benefited more from practice. With the same amount of practice they acquired chess skill more quickly, reached a higher peak performance and slowed the decline at older ages. The authors conclude that their research shows "the futility of scrutinizing the relative importance of highly intertwined factors." That sentence is the best short answer to the question in this page's title. Because the effects interact, a statement such as "practice explains X percent and IQ explains Y percent" is not additive, and the chess papers cannot be summed into one pie. The meta-analysis and the longitudinal study also come from different samples with different measures, so their percentages are not directly comparable. The cognitive abilities named here are explained on the pages about fluid and crystallized intelligence, the CHC model, working memory and processing speed.

6 What Do Music Studies Say About Practice, Working Memory and Genes?

In music, practice is a strong predictor of performance, working memory capacity adds to it, and a large twin study found that the amount of practice people choose is itself partly heritable, which complicates any claim that hours cause skill. The piano study by Meinz and Hambrick tested participants with a wide range of piano skill, from novice to expert, on sight-reading. Deliberate practice accounted for nearly half of the total variance in sight-reading performance. Working memory capacity added an incremental positive effect on top of that, and there was no evidence that practice reduced the effect. The authors note that working memory capacity is highly general, stable and heritable, and conclude that their results call into question the view that expert performance is solely a reflection of deliberate practice. In the vocabulary of the CHC model, this is evidence that a broad ability, Gwm, contributed to a task that looks like a pure skill. It is one task and one sample, and it concerns sight-reading and not performance at an elite level.

The twin studies ask a different question: what do the practice hours reflect? Mosing and colleagues examined associations between music practice and music ability, measured as rhythm, melody and pitch discrimination, in 10,500 Swedish twins. The correlations ranged from 0.18 to 0.36. Music practice was substantially heritable, at 40 to 70 percent. The association between practice and ability was predominantly genetic, and within identical twin pairs that differed in amount of practice there was no difference in ability. The authors conclude that music practice may not causally influence music ability and that genetic variation affects both ability and the inclination to practice.

The Ericsson side contests these results on definitional grounds. Ericsson and Harwell reanalyzed a twin data set of high school students that had been used to argue for genetic effects on music achievement, after defining the achievement to match the level of the students at the West Berlin academy, and report that the estimate of the additive genetic effect was then no longer significant. They also argue that the genetic effects found so far account for remarkably small amounts of variance, except for height and body size. The twin samples come from the general population, in one case from high school students, while the violin study sampled a conservatory. Whether the two sets of findings can be reconciled depends on where in the range of skill each one sampled, which is an open empirical question and not a rhetorical one.

For the reader who asks what this means for IQ, the answer is indirect. The music studies did not administer full IQ batteries, and the one cognitive measure here, working memory capacity, is one of six broad domains that a full battery reports. The page on IQ and musical ability covers aptitude and training directly, the page on what heritability actually means explains why a heritable influence on practice does not mean practice has no effect, and the page on IQ and memory covers the memory side.

7 What Do Sports Studies Say About Practice and Cognitive Ability?

In sports, practice explains 18 percent of the differences in performance overall but only 1 percent among elite athletes, and the cognitive measures that separate higher-skilled from lower-skilled athletes are mostly sport-specific decision tasks and not IQ. The sports figures were given above: 18 percent overall and 1 percent among elite-level performers in the 2016 meta-analysis, and no evidence that high-skill athletes started earlier. The question that fits this page is what cognitive tests show. The 2021 meta-analysis by Kalén and colleagues in Psychological Bulletin screened 9,433 records and included 136 reports containing 142 studies, 1,227 effect sizes and 8,860 participants, all of competitive athletes. Higher-skilled athletes scored better than lower-skilled athletes on cognitive functions and skills, with a Hedges g of 0.59 and a 95 percent confidence interval of 0.49 to 0.69, which the authors call a medium effect.

The breakdown is more informative than the average. The effect was larger for tests of cognitive decision-making skills (g = 0.77) than for basic cognitive functions (g = 0.39) or higher cognitive functions (g = 0.44), and larger for sport-specific stimuli than for general ones. Only 11 of the studies used a prospective design, so the evidence is mostly a comparison of athletes who already differ, and the authors report insufficient evidence to determine whether cognitive functions can predict future sport performance. They also found no evidence to support claims that tests of general cognitive functions, such as executive functioning, should be used for talent identification or player selection. The measures in these studies are not intelligence quotients in the sense used elsewhere on this site, so the result should not be read as a statement about IQ and athletic success.

One reading, which fits the larger effect for sport-specific stimuli, is that experience builds domain-specific cognition. The design cannot show that, because most studies compare athletes who already differ. The page on average IQ by sport covers the one league whose players it found measured, and explains why those scores describe a selected group of professionals. In sports as in chess, the weakest link is the one that the headline numbers hide: the samples compare people who already differ, so neither shows what practice or ability would do for a given beginner.

8 Why Are the Education and Professions Figures So Small?

Deliberate practice explained 4 percent of performance differences in education and under 1 percent in professions in the 2014 meta-analysis, and the two sides read those numbers in opposite ways. The figures come from the abstract of the Macnamara, Hambrick and Oswald paper. Their reading is the plain one: in these domains most of the variation in performance is explained by something other than practice, which is part of why they conclude that practice is important but less important than has been argued. The second reading comes from the reanalysis by Ericsson and Harwell, whose three criteria left only one effect size from education, a study of Spelling Bee performance, and none from professions. On that reading the studies in those two domains mostly measured something other than deliberate practice, so their small figures say little about the concept itself.

Neither reading tells the reader how large the role of cognitive ability is in school or at work. That is a different body of research with its own meta-analyses, and the site's pages on IQ and academic achievement, IQ and job performance and average IQ by profession cover it with their own sources, so this page does not repeat their numbers. The point to take from the comparison is structural, and it is our reading and not a finding of the papers. In games, music and sports the outcome is a rating, a performance or a competitive result that can be tied to a defined practice routine, while in education and professions the outcome is a grade or an evaluation influenced by many other things. Practice that fits the original definition is harder to isolate there, and a pooled correlation that mixes very different kinds of activity can be small whether or not the best-defined practice matters.

This is also why the domain table has to be read with the sample in mind. A study of preparation for a spelling contest and a study of how long people have held a job are not the same experiment, and averaging them hides the fact that the word practice stands for a different activity in each. The honest summary is that the evidence is strongest where practice can be defined and counted, which is also where Ericsson's original definition applies best, and weakest where it cannot. For the same reason, the article does not claim that practice matters little at work. It claims that these particular studies do not show how much.

9 Do Practice and Intelligence Add Up, Interact, or Get Mixed Together?

The studies show three different relations between practice and ability, and a single percentage for either one cannot describe all three. The first relation is additive. In the piano study by Meinz and Hambrick, working memory capacity predicted sight-reading above and beyond practice, so each contributed something the other did not. The second is interaction. In the chess careers followed by Vaci and colleagues, the benefit of practice depended on intelligence: with the same amount of practice, more intelligent players gained skill faster, peaked higher and declined later. The third is correlation between the two. In the twin data of Mosing and colleagues, the amount of practice itself was partly heritable, which means that the people who practice more are not a random draw from the population.

Each relation has a different consequence for a question like "practice or IQ." If the two add, one can in principle ask what each contributes, though only for a defined sample. If they interact, the contribution of practice is not one number but depends on the person. If they are correlated, then comparing people with different amounts of practice also compares people who differ in other traits, so the practice difference cannot be read as a pure effect of training. The statistics that Macnamara and her colleagues reported are consistent with any of these. They do not distinguish between them.

This is also why the term intelligence needs care. The measure in the chess meta-analysis was a set of broad abilities named in the CHC vocabulary, among them fluid reasoning (Gf), comprehension-knowledge (Gc), short-term memory (Gsm) and processing speed (Gs), and the strongest content-specific link was with numerical ability. The measure in the piano study was working memory capacity (Gwm in the CHC terms ACIS uses). A statement about "IQ" in an expertise study is therefore usually a statement about one or two broad abilities, which is the reason the page on g and the page on types of intelligence distinguish the general factor from the broad abilities. It is also why the difference between aptitude and intelligence matters: an aptitude for music or chess is a narrower thing than a Full Scale IQ, and the two should not be swapped when reading a study.

Gobet and Campitelli added a further variable that belongs in any combined picture: starting age. In their sample of 104 players, skill correlated with starting age, and the authors concluded that practice is necessary but not sufficient and that starting age is important. The 2019 reanalysis of another chess data set found that a later serious starting age was still associated with a lower peak rating after training was accounted for. But the sports meta-analysis found that athletes who reached a high level did not begin earlier than lower-skill athletes. The age at which people start matters in some domains and not in others, and the evidence does not support a rule that applies to all of them.

10 Study by Study: Design, Sample, Measure and Result

The evidence on practice and IQ comes from five kinds of designs, and each one can answer only some of the questions people ask. The table lists the studies used on this page, with what each one actually compared. Results are as the papers report them, and any conversion is flagged as our arithmetic.

Five designs appear here. Retrospective group comparisons and their replication test whether more practice goes with higher level in one conservatory. Meta-analyses pool correlations across samples and so inherit the quality of the practice estimates. Reanalyses test whether the answer changes with the criteria. Longitudinal studies follow people and can see interaction. Twin studies separate genes from environment but measure practice and ability in populations that may differ from experts. None of the designs randomizes people to practice or no practice for years, so none supports a firm causal claim about hours. Where a verb such as "predicted" or "accounted for" is used on this page, it refers to a statistical relation within a sample and not to what practice does to a given person.

11 What the Evidence Supports, Stated Narrowly

The evidence supports a short list of claims, and each is narrower than the headlines on either side. The claims below hold across the studies reviewed, with the strength noted.

• Practice is associated with performance in games, music and sports, with 18 to 26 percent of the variance explained in the 2014 meta-analysis, and no paper reviewed here argues that practice is irrelevant. The disagreement is about how much and in what form.
• The 10,000-hour figure is not a finding of the 1993 study. The best violin group's mean was 7,410 hours by age 18 as the replication reports it, the proposer of deliberate practice calls the number a misreading, and international-level performers have been reported to range from 3,000 to 40,000 hours.
• The share of variance explained by practice depends on how practice is defined and counted. Estimates run from 12 or 14 percent across domains in the 2014 meta-analysis, to 26 percent for games in that analysis, to 29 percent and about 61 percent after correction in the reanalysis with stricter criteria.
• In chess, cognitive abilities correlate positively with skill, more among young and unranked players than among adults and ranked players, and intelligence and practice interact in longitudinal data.
• In music, working memory capacity added to practice in one sample, and twin studies show that practice itself is partly heritable, a result that the Ericsson side disputes on the grounds of how achievement was defined.
• In sports, practice explained 1 percent of variance among elite athletes, and cognitive differences between skill levels are largest for sport-specific decisions and not for general functions.

The evidence does not support the following claims. It does not show that practice does not matter or that talent is fixed. It does not show that IQ determines who becomes an expert, because the correlations are small to moderate and are weaker among strong players. It does not give a number of hours for a person, a skill or a domain. It does not show that the percent of variance is the share of skill that practice builds in one person. And it does not show that practice can or cannot compensate for a lower score on a cognitive test, because no study reviewed here compared people with the same practice and different IQ scores in a way that would answer that question for an individual.

12 What This Does Not Say About You, and How to Read a Score

Nothing in this literature can tell you how many hours you need, what your ceiling is, or what an IQ score implies about a skill you are building. Every figure above describes a group of people measured at one point in time. The standard that applies to such evidence is validity: the Standards for Educational and Psychological Testing (AERA, APA and NCME, 2014) frame validity as the degree to which evidence and theory support the interpretation of test scores for their proposed use. On that standard, a cognitive test score supports an interpretation about the broad abilities it was designed to measure, and a variance share from a meta-analysis supports an interpretation about groups. Neither supports the interpretation that a person will or will not reach a level of skill. The American Psychological Association's Ethical Principles of Psychologists and Code of Conduct, in the version effective January 1, 2017, adds two requirements. Standard 9.02 asks psychologists to use assessments in a manner and for purposes appropriate in light of the evidence on their usefulness, and Standard 9.06 asks them to take into account the purpose of the assessment and the characteristics of the person assessed that might reduce the accuracy of their interpretations, and to indicate significant limitations.

Applied to this topic, those standards give four practical rules. A score on a test of broad abilities is not a measure of how much you have practiced, and a practiced skill is not a measure of general ability, so the two should not be used as proxies for each other. A correlation of 0.24 between ability and chess skill is a statement about covariation in a sample, and it does not predict one player's rating. Differences in practice are confounded with differences in other traits, as the twin data show, so neither a low nor a high score should be used to talk yourself out of training. And a score read without its confidence interval reads more precision than the instrument has, which is why the page on how IQ scores are normed, the page on what an IQ score means and the page on reliability and validity matter more than any single number.

ACIS reports a Full Scale IQ and six index scores for fluid reasoning, verbal comprehension, quantitative reasoning, visual spatial, working memory and processing speed, on a scale with a mean of 100 and a standard deviation of 15, with percentiles and a 95 percent confidence interval, and adult norms for ages 16 to 90. The prices read on October 20, 2026 were 15 dollars for the Quick form (6 subtests, about 45 minutes), 30 dollars for the Optimized form (13 subtests, about 110 minutes) and 50 dollars for the Full Scale form (all 20 subtests, about 175 minutes), each a one-time payment, with a free trial that needs no card. The test is online and unsupervised, it is not a clinical or diagnostic instrument, it is not suitable for hiring decisions, school accommodations or admission to high IQ societies, and it is in English only. The technical manual documents how the scores are built. What the report can show is a profile of broad abilities. It cannot show how far practice will take you in a particular domain, and this page does not claim that it can. Related pages in this series apply the same discipline to other topics, including math ability and IQ and the quirks attributed to high-IQ people.

13 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. Percentages, correlations and effect sizes are quoted as the papers report them, and any count or conversion that is ours is labeled as our arithmetic. Journal records were confirmed with Crossref on October 20, 2026, and the abstracts of the journal articles were read through Europe PMC or the authors' open versions. The open access texts of the replication, the commentary and the reanalysis, the full text of the 2016 chess meta-analysis and the 2019 chess reanalysis, the Standards and the APA code were read in full or searched on October 20, 2026. The 2014 meta-analysis was read through its abstract and the press summary below, and its overall figure is reported there and in the reanalysis with different values, as the page says.

• Ericsson K A, Krampe R T and Tesch-Römer C. The role of deliberate practice in the acquisition of expert performance. Psychological Review, 1993, volume 100, issue 3, pages 363 to 406.
• Macnamara B N, Hambrick D Z and Oswald F L. Deliberate practice and performance in music, games, sports, education, and professions: A meta-analysis. Psychological Science, 2014, volume 25, issue 8, pages 1608 to 1618.
• Association for Psychological Science. Becoming an expert takes more than practice, press summary of the 2014 meta-analysis, opened October 20, 2026.
• Macnamara B N, Moreau D and Hambrick D Z. The relationship between deliberate practice and performance in sports: A meta-analysis. Perspectives on Psychological Science, 2016, volume 11, issue 3, pages 333 to 350.
• Macnamara B N and Maitra M. The role of deliberate practice in expert performance: Revisiting Ericsson, Krampe and Tesch-Römer (1993). Royal Society Open Science, 2019, volume 6, issue 8, article 190327.
• Ericsson K A and Harwell K W. Deliberate practice and proposed limits on the effects of practice on the acquisition of expert performance: Why the original definition matters and recommendations for future research. Frontiers in Psychology, 2019, volume 10, article 2396.
• Hambrick D Z, Altmann E M, Oswald F L, Meinz E J and Gobet F. Facing facts about deliberate practice. Frontiers in Psychology, 2014, volume 5, article 751.
• Platz F, Kopiez R, Lehmann A C and Wolf A. The influence of deliberate practice on musical achievement: A meta-analysis. Frontiers in Psychology, 2014, volume 5, article 646.
• Gobet F and Campitelli G. The role of domain-specific practice, handedness, and starting age in chess. Developmental Psychology, 2007, volume 43, issue 1, pages 159 to 172.
• Burgoyne A P, Sala G, Gobet F, Macnamara B N, Campitelli G and Hambrick D Z. The relationship between cognitive ability and chess skill: A comprehensive meta-analysis. Intelligence, 2016, volume 59, pages 72 to 83; with the authors' corrigendum, Intelligence, 2018, volume 71, pages 92 to 96 (not opened).
• Burgoyne A P, Nye C D, Macnamara B N, Charness N and Hambrick D Z. The impact of domain-specific experience on chess skill: Reanalysis of a key study. The American Journal of Psychology, 2019, volume 132, issue 1, pages 27 to 38.
• Vaci N, Edelsbrunner P, Stern E, Neubauer A, Bilalić M and Grabner R H. The joint influence of intelligence and practice on skill development throughout the life span. Proceedings of the National Academy of Sciences, 2019, volume 116, issue 37, pages 18363 to 18369.
• Meinz E J and Hambrick D Z. Deliberate practice is necessary but not sufficient to explain individual differences in piano sight-reading skill: The role of working memory capacity. Psychological Science, 2010, volume 21, issue 7, pages 914 to 919.
• Mosing M A, Madison G, Pedersen N L, Kuja-Halkola R and Ullén F. Practice does not make perfect: No causal effect of music practice on music ability. Psychological Science, 2014, volume 25, issue 9, pages 1795 to 1803.
• Kalén A, Bisagno E, Musculus L, Raab M, Pérez-Ferreirós A, Williams A M, Araújo D, Lindwall M and Ivarsson A. The role of domain-specific and domain-general cognitive functions and skills in sports performance: A meta-analysis. Psychological Bulletin, 2021, volume 147, issue 12, pages 1290 to 1308.
• American Educational Research Association, American Psychological Association and National Council on Measurement in Education. Standards for Educational and Psychological Testing, 2014; and American Psychological Association, Ethical Principles of Psychologists and Code of Conduct, Standards 9.02 and 9.06, version effective January 1, 2017.

14 Frequently Asked Questions