Do night owls score higher? A small association and what the hour changes

0 The short answer

Evening types score slightly higher than morning types on average in several studies, but the association is small, it depends on how chronotype and ability are measured, and it does not show that staying up late raises or signals intelligence. A 2011 meta-analysis found a correlation of 0.08 between eveningness and cognitive ability and minus 0.04 for morningness, a 2021 update of 30 studies put eveningness at 0.056 and morningness at minus 0.008, and a systematic review of 65 studies found no main effect of chronotype in more than 80 percent of them. The 13.5 percent in the 2024 headlines is a regression coefficient of 0.1351 on a standardized scale, not a percentage of anything. The effect with the strongest experimental support is different: people tend to perform better at the time of day that matches their own rhythm, which is a reason to choose your test hour, not a reason to change your bedtime.

1 What Is a Chronotype, and How Do Studies Measure It?

Chronotype is a tendency toward earlier or later sleep and waking, and the studies behind the headlines measure it in three different ways that are not interchangeable. A 2019 paper in Chronobiology International defines morningness and eveningness, or chronotype, as the timing of sleep and waking patterns across a 24 hour day. Everything beyond that definition depends on what the researchers actually recorded.

The first measure is a self label. The 2024 UK Biobank analysis behind the "night owls are smarter" headlines used one touchscreen question, "Do you consider yourself to be?", with six answers: definitely a morning person, more a morning than an evening person, more an evening than a morning person, definitely an evening person, do not know, and prefer not to answer. The authors regrouped the answers into morningness, intermediate and eveningness, and excluded unclear responses (West and colleagues, 2024). A person who is vaguely "more evening than morning" is therefore an evening type in that paper, whatever their actual schedule.

The second measure is a questionnaire score. A 2007 experiment on adolescents used the Children's Morningness-Evening Preferences scale by telephone with 259 young adolescents aged 11 to 14, then invited those with extreme scores to the laboratory (Goldstein and colleagues, 2007). Comparing people at the two ends of a scale answers a different question from comparing three broad bins.

The third measure is behavior: clock times. The 2020 Mensa study used the Munich Chronotype Questionnaire, which records when people actually sleep on work days and on free days, and summarizes chronotype as the midpoint of free day sleep corrected for sleep debt (Ujma and colleagues, 2020). The 2009 Add Health analysis used self reported bedtimes and waking times, as that same paper describes it. Bedtime is partly a preference and partly a consequence of work, school, family and screens, so a study of bedtime and a study of preference can disagree without either being wrong.

Intelligence is measured just as unevenly. The UK Biobank paper used four brief touchscreen tasks: thirteen verbal and numerical reasoning questions, a card pairs memory game scored by errors, a snap game scored by reaction time, and a prospective memory task. Only about 10,000 of the 26,820 participants completed all four; the rest completed only pairs matching and reaction time. None of these is a full intelligence battery, and three of the four index memory or speed rather than reasoning. The page on what IQ measures and the page on the CHC model explain why a composite of two speed and memory games is a different quantity from a Full Scale IQ. Where this page says "cognitive ability", it means whatever each study actually measured, and it names the measure.

This page is about chronotype, a stable preference. It does not cover how many hours you sleep or what a sleepless night does to a score, which the page on sleep and IQ treats. A sibling page, stress and IQ, covers the separate question of state versus trait.

2 The Studies Side by Side

The studies differ so much in design, sample and measure that no single sentence summarizes them, and the table below keeps each result attached to the design that produced it. Every row is described in its own section further down; the correlation, coefficient or difference is the one the source prints.

Two patterns stand out before any reading of detail. First, the largest samples and the loosest measures tend to show small positive associations for evening types in adults. Second, the one design that manipulates time of day, as opposed to classifying people, produces an effect of a different kind: it depends on the match between a person and the hour.

3 Kanazawa and Perina, 2009: Where the Claim Started

The 2009 paper most often cited for the night owl claim reported a statistically detectable but very small association between childhood IQ and later sleep timing, and it read that association through an evolutionary hypothesis that simpler explanations fit equally well. Kanazawa and Perina build on what they call the Savanna-IQ Interaction Hypothesis, which says that more intelligent individuals are more likely to adopt evolutionarily novel values and preferences, while intelligence has no effect on evolutionarily familiar ones. Their reasoning about sleep, as their abstract gives it, runs as follows: ethnographies of traditional societies suggest that nocturnal activity was probably rare in the ancestral environment, so the hypothesis predicts that more intelligent individuals are more likely to be nocturnal. They report that their analysis of the National Longitudinal Study of Adolescent Health, known as Add Health, confirmed that prediction.

The size of the association is smaller than the confident framing suggests. Ujma and colleagues, in the introduction to their 2020 paper, summarize the Add Health result as a small but significant linear relationship between childhood IQ and self reported bedtimes and waking times between ages 18 and 28, with correlations of 0.013 to 0.053 in a sample of 15,197. A correlation of 0.053 squared is 0.0028, so on the most generous reading the childhood IQ score accounts for about 0.3 percent of the variation in later sleep timing (our arithmetic), and the smallest correlation accounts for less than 0.02 percent. In a sample this large, almost any nonzero association is statistically significant, which is why significance alone says little about importance.

The paper itself gives a concrete contrast. Using the childhood IQ classes of Herrnstein and Murray, Kanazawa and Perina report mean weeknight bedtimes of 23:41 for the 584 respondents with childhood IQs below 75 and 00:29 for the 468 with IQs above 125, a gap of 48 minutes (our subtraction), and they note that the absolute differences in minutes are small. A ZME Science report on the paper quoted the sleep researcher Hans Van Dongen, who cautioned that the analysis is based on how late individuals choose to go to bed and cannot rule out that adolescents with higher IQs choose to study late. That comment comes from the press account, so this page treats it as the commentator's view and not the authors' words.

Several limits apply.

• The evolutionary reading is an interpretation. The data show an association between an IQ score and a later bedtime. Whether bedtime reflects "evolutionary novelty", study habits, social life, work or something else is not something a bedtime variable can reveal.
• The temporal order helps and does not settle causation. Intelligence was measured before sleep timing, which rules out some reverse stories, but any third factor shared by intelligence and late bedtime, such as family circumstances, school type or later education, remains open. The page on is IQ genetic explains why shared family background is hard to separate from either genes or environment.
• Bedtime at ages 18 to 28 is not a chronotype. It reflects university schedules, early jobs and social life. This is exactly the issue the later Mensa study tested.
• The author's wider work is contested. The press account noted controversy over his broader evolutionary claims. That is a reason to read the interpretation cautiously, not an argument that the correlation is zero.

The honest summary of 2009 is that a large survey found a tiny positive association between childhood IQ and later bedtime, and that the headline "night owls are more intelligent" turned a correlation of at most 0.053 into a statement about types of people.

4 What Do the Meta-Analyses Say About Chronotype and Cognitive Ability?

The 2011 meta-analysis found that eveningness and cognitive ability correlate at 0.08, morningness and cognitive ability at minus 0.04, and that the academic achievement pattern runs the other way, and a 2021 update with more studies put the figures at 0.056 and minus 0.008. Preckel, Lipnevich, Schneider and Roberts (2011) reported four meta-analyses. Their abstract gives a total of 2,177 participants for morningness and cognitive ability, 1,519 for eveningness and cognitive ability, 3,220 for morningness and academic achievement and 700 for eveningness and academic achievement. Eveningness correlated positively with cognitive ability (0.08) and negatively with academic achievement (minus 0.14). Morningness showed the opposite pattern, minus 0.04 with cognitive ability and 0.16 with academic achievement.

Three points follow from those numbers.

• The effects are small. A correlation of 0.08 squared is 0.0064, so eveningness shares well under 1 percent of its variance with cognitive ability (our arithmetic). Ujma and colleagues, who cite the same meta-analysis, give the unrounded values as minus 0.042 and 0.075 and report that it found no evidence of publication bias.
• The morningness and eveningness correlations come from different groups of studies. They are not mirror images of one sample, which is why the sample sizes differ and why the two numbers do not combine into one.
• The paradox is the finding. Evening types did slightly better on ability measures and morning types did better on grades. The page on IQ and academic achievement covers the link between ability and grades in general. One reading of the chronotype pattern is that school schedules, which usually start early, reward a morning preference independently of ability. That is an interpretation, and the meta-analysis cannot test it.

What the meta-analysis can say is that the two outcomes pull in opposite directions, so a headline built on only one of them is incomplete.

The 2021 update by Ujma and Scherrer is the better current estimate for intelligence, because it pooled 30 studies and 11,160 participants and sampled ages from children to adults in late middle age. It found no significant link between morningness and intelligence overall (r of minus 0.008) and no evidence of publication bias. That overall figure hid an age pattern: the morningness correlation fell with mean sample age (R squared of 54 percent), from a nonsignificant positive trend in children and adolescents to a significant negative correlation after young adulthood. Eveningness correlated 0.056 with intelligence, but that finding rests on a more age restricted sample and reached significance only in some model specifications. The authors hypothesize that the age pattern reflects social effects, in which more intelligent individuals are better able to adjust their daily schedules to their natural rhythm. That is a hypothesis, not a result, and it fits the Mensa comparison described below.

Earlier single studies fit the same small, mixed picture. Roberts and Kyllonen (1999) studied U.S. Air Force recruits after six weeks of preparatory training that imposed a common sleep and waking schedule, which removes much of the lifestyle variation that confuses other samples. A later paper summarizes their result as better processing speed and working memory in evening recruits. The title of the original, which says that early to bed and early to rise "will likely make you anything but wise", shows how the finding was framed at the time. Two decades later, Zajenkowski, Jankowski and Stolarski (2019) reported in two studies (504 and 232 participants) that evening types rated their own intelligence higher, that this held after controlling for tested fluid and verbal intelligence, and that in the second study narcissism fully accounted for the self view. Their abstract takes as its premise that evening chronotypes show slightly higher intelligence than morning individuals, which agrees with the size of the meta-analytic correlation.

5 What Did the 2024 UK Biobank Study Actually Measure?

The 2024 paper was a cross sectional regression of brief touchscreen tasks on self reported sleep duration, chronotype and sleep quality in 26,820 UK Biobank participants aged 53 to 86, and its own abstract leads with sleep duration, not with night owls. The authors, West and colleagues at Imperial College London, split the sample into two cohorts. Cohort 1 had 10,067 participants (56 percent female) who completed four tasks: thirteen verbal and numerical reasoning questions, pairs matching, reaction time and prospective memory. Cohort 2 had 16,753 participants (56 percent female, mean age 72) who completed only pairs matching and reaction time. The two cohorts were analyzed separately.

Raw scores were converted to z-scores with the formula (raw score minus mean baseline) divided by the baseline standard deviation, which put the tasks on one scale. The outcome was a global cognitive z-score, analyzed by ordinary least squares regression with robust standard errors. The covariates named in the paper include sex, age, body mass index, smoking, alcohol intake and diagnoses such as diabetes and vascular or heart conditions.

The chronotype result comes as a set of coefficients. Intermediate types had higher z-scores than morning types (0.1061 in Cohort 1 and 0.0632 in Cohort 2), and so did evening types (0.1351 in Cohort 1 and 0.0750 in Cohort 2). The abstract says chronotype distinctions, particularly intermediate and evening types, were linked to superior cognitive function, and its conclusion emphasizes the role of sleep quality. The same table prints a coefficient for age of minus 0.0177 per year in Cohort 1 and minus 0.0264 in Cohort 2. Dividing, the evening coefficient is about the size of the difference associated with seven and a half years of age in Cohort 1 and under three years in Cohort 2 (our arithmetic on the printed coefficients, not a claim of the authors). In a sample spanning 33 years of age, that is a modest quantity.

Note also what the study did not do. It did not use a standardized intelligence test or compute an IQ. It did not follow anyone over time. It did not adjust for educational attainment, which the authors say was not possible because of incomplete data. And it did not control for the time of day at which the cognitive tasks were taken, a limitation the authors state themselves along with the exclusion of unclear chronotype answers and the cross sectional design.

The intermediate group deserves attention too. The group in the middle, which the headlines ignored, scored almost as far above the morning group as the evening group did (0.1061 against 0.1351 in Cohort 1, and 0.0632 against 0.0750 in Cohort 2). The 95 percent intervals overlap (Cohort 1: 0.0787 to 0.1335 for intermediate and 0.0898 to 0.1804 for evening types). A reader looking for a night owl advantage finds instead a contrast between the morning label and everyone else, which a single self rating question cannot interpret further.

6 Where Did "13.5 Percent Higher" Come From?

The percentages in the 2024 headlines are the paper's regression coefficients rewritten as percent, and a coefficient on a z-score scale is not a percentage difference in scores. Imperial College London's news release of July 11, 2024 said evening types scored about 13.5 percent higher in one group and 7.5 percent in another, relative to morning types, with intermediate types scoring about 10.6 and 6.3 percent higher. Neuroscience News and other outlets repeated the figures under headlines such as "Night Owls Have Higher Cognitive Aptitude."

In the full text of the paper that we searched, the number 13.5 does not appear. What appears is the table of coefficients, and set side by side the correspondence is exact to rounding.

So the press figures are the coefficients multiplied by 100 (our observation from the two documents; the release does not state its method). The multiplication changes the meaning. A coefficient of 0.1351 in a regression on a z-score outcome says that, holding the other covariates fixed, the average global z-score of evening types was 0.1351 units above that of morning types. If the z-score has a standard deviation near 1, that is roughly one seventh of a standard deviation, not 13.5 percent more of anything. "13.5 percent higher" invites the reader to picture a score of 100 becoming 113.5, and nothing in the paper supports that picture.

Independent experts said as much when the study was published. In reactions collected by the Science Media Centre, Dr. Jacqui Hanley of Alzheimer's Research UK noted that self reported sleep data are prone to error and that the design cannot show whether chronotype affects thinking or whether declining thinking changes sleep patterns. Professor Roi Cohen Kadosh of the University of Surrey said the study shows association, not causal effect, and called policy based on it premature. Dr. Jessica Chelekis of Brunel University London pointed out that the paper never states what time of day participants took the cognitive tests. The paper's own discussion says the cross sectional design limits causal inference. The release quotes the lead author, Dr. Raha West, saying these chronotypes "could impact" cognitive function, which is phrased as a possibility, and it also notes that the finding does not mean all morning people perform worse.

This is a design mismatch rather than a dispute about arithmetic. A cross sectional survey can support the phrases "is associated with" and "differs between." It cannot support "makes," "boosts" or "gives." The headline verbs outran the study.

7 Does the Difference Survive Age, Work Schedule, Education and Time of Testing?

The most direct test of whether the late bedtimes of high scorers reflect a night owl trait found no chronotype difference once work schedules were accounted for, and the other main confounds point the same way. Ujma, Baudson, Bódizs and Dresler (2020) compared 586 German Mensa members (IQ of 130 or higher) with 586 age and sex matched volunteers drawn from a large web database of the Munich Chronotype Questionnaire. Each group had 232 women and 354 men, with a mean age of 38.2 years. On work days, Mensa members fell asleep later and woke later, with midsleep at 3.40 hours against 3.21 (p equals 0.01). On free days there were no significant differences in the primary sleep timing measures. Corrected chronotype did not differ between the groups (p equals 0.75). Mensa members started work about 14 minutes later on average (8:35 against 8:21, p equals 0.02), and later work start times fully mediated the link between membership and work day midsleep.

The authors conclude that later sleep timing among high scorers is not due to physiological differences but to later work schedules. They note the limits: cross sectional design, a sample possibly underpowered to detect very small free day differences, and no IQ data for the control group. A group of people who joined a high IQ society is also not the general population. Even so, this is the cleanest test of the idea that the late bedtime of high scorers is a biological night owl trait, and it did not find one.

Four other confounds recur across the literature.

• Age. Chronotype shifts across the lifespan. In the 2007 adolescent study, mean preference scores declined steadily from age 11 to age 14, which the authors describe as a movement away from morningness. In the UK Biobank study, the participants were 53 to 86, and age had a strong coefficient of its own. Any comparison across ages mixes cohort, aging and chronotype.
• Education and occupation. The UK Biobank authors say they could not adjust for educational attainment. People with more schooling or flexible professional jobs may be freer to choose late schedules, and the 2020 Mensa result shows that schedule alone can generate an apparent difference.
• Children and the opposite sign. An analysis of 4,493 children from the Adolescent Brain Cognitive Development study (Yamashita, Shou and Mizuno, 2024) found the opposite direction: late chronotype was associated with lower vocabulary (standardized beta of minus 0.09), oral reading (minus 0.05) and picture sequence memory (minus 0.07), with no significant association for executive function or processing speed. That is a cross sectional result in children, and it shows that the sign depends on the age group.
• The hour of testing. If evening types are tested in the evening and morning types in the morning, or if the test falls at a mismatched hour for one group, the synchrony effect described below can create a chronotype difference by itself. The UK Biobank paper did not control for it.

A 2022 study from Western University in Canada points the same way for verbal ability. In a news summary, the senior author Stuart Fogel said that once bedtime and age were accounted for, morning types tended to have superior verbal ability (Gibbings and colleagues, 2022, read as an abstract fragment; news summary). The point is not that morning types are smarter. It is that the sign of the difference changes when a confound enters the model, which is what a confounded association looks like.

8 Can Eveningness Cause Higher Scores? Genes, Age and the Direction of the Arrow

No study read for this page randomized chronotype or shifted it and then measured intelligence, so none can say that eveningness causes higher scores, and the genetic evidence describes overlap rather than cause. Ujma and colleagues summarize large genome wide studies as finding a negative genetic correlation between cognitive ability and morningness (rg of minus 0.15 to minus 0.17), meaning that variants associated with an earlier chronotype tend also to be associated with lower cognitive ability. A genetic correlation says that two traits share some genetic influences. It does not say that one trait produces the other, and it is compatible with the same variants acting on both. The page on what heritability does and does not mean explains why such figures describe populations, not individuals.

Several explanations remain open, and each would leave a different fingerprint in the data.

The table shows why one correlation can have several honest explanations. The most direct tests, the Mensa comparison and the time of day experiments, point to schedules and hours rather than to a stable difference in ability. Age adds a developmental wrinkle: in the 2007 adolescent sample preferences moved toward evening between ages 11 and 14, while the 2024 children's analysis found late chronotype associated with lower vocabulary. If the "night owl advantage" were a stable trait of ability, it would be odd for its sign to depend this much on age group and measure.

9 What Is the Synchrony Effect, and Is It Real?

The synchrony effect, better performance at the time of day that matches a person's rhythm, is the best experimentally supported finding in this literature, and it varies by task, age and ability type. The clearest early test in intelligence measurement is Goldstein and colleagues (2007). From 259 young adolescents, they selected 80 (ages 11 to 14) with extreme preference scores and randomly assigned each to a morning session (8 to 10 am) or an afternoon session (1 to 3 pm), chosen to match the limits of a school day. Testing used three subtests of a Wechsler children's scale: Block Design, Vocabulary and Digit Span. The authors found a synchrony effect for the fluid measures, with better performance at times that matched the individual's preference, and a composite of the subtests that "computed to a 6 point difference in IQ estimates." Random assignment to session applies to the time of testing, not to chronotype itself, which was chosen by extreme scores. So the experiment supports the statement that, in this sample, scores were lower at a mismatched hour, and it does not make chronotype a randomized factor.

Later studies refine the picture without erasing it. Jankowski, Díaz-Morales and Vollmer (2023) tested 198 Polish students aged 17 to 20, 92 in a class at 8:00 and 106 in an afternoon class. They used a 32 item synonym task for crystallized intelligence and a 50 item nonverbal scale for fluid intelligence. In the morning class, morning types outperformed evening types on the crystallized test (R squared of .11, Cohen's d of .94), and in the afternoon class the difference was gone (d of .11). The effect came from evening types performing worse in the morning that recovered in the afternoon, while morning types held steady. There was no significant synchrony effect for fluid intelligence, and the crystallized pattern was independent of self reported sleepiness. The two studies therefore disagree about which ability type shows the effect, which is itself informative: the effect depends on the task.

The broadest check is a 2025 systematic review of 65 studies. It found that more than 80 percent reported no main effect of chronotype on cognitive function. For adults aged 18 to 45, 29 of 64 studies (45.31 percent) showed a synchrony effect, mostly in attention, inhibition and memory. For older adults, 10 of 12 studies (83.33 percent) did, especially on tasks involving fluid abilities. The authors add that exogenous factors such as task complexity and lighting also moderate the effect, and that ignoring synchrony may exaggerate apparent cognitive deficits, especially in older adults.

Read together, these findings reverse the headline. A main effect of chronotype is mostly absent. An interaction between chronotype and hour is common, though not universal. For a person taking a test, the useful variable is when, not which type.

10 How Many IQ Points Is the Difference Worth?

Treated as IQ points, the chronotype associations are worth about one or two points at most, while the best estimate of a mismatched test hour is larger and applies only to the sitting, not to the person. The table converts each reported quantity using the IQ standard deviation of 15, which the page on the standard deviation of 15 explains. The multiplication is our arithmetic. It treats each correlation as the change per standard deviation of chronotype, and each coefficient as a group difference, and it assumes the outcome is on an IQ-like scale, which several of these are not.

Three cautions apply. First, a UK Biobank composite of memory and speed games is not an IQ, so the conversion is an illustration of scale only. Second, a one or two point difference is small beside the uncertainty around any single score, which the page on reliability and validity explains and which a report states as a confidence interval. Third, the Goldstein figure is measured on adolescents, on three subtests, in a laboratory with an 8 to 10 am slot against an afternoon slot, so it should not be stretched to adults at home.

On the IQ scale, one or two points is roughly a tenth of a standard deviation (our arithmetic: 1.5 divided by 15). The page on how an IQ converts to a percentile shows what a shift of that size does to rank, and the IQ percentile calculator lets a reader check a specific case.

11 What Does This Not Say About You?

Group averages in a population study cannot tell you your own IQ, your own chronotype advantage or whether your schedule helps or harms you. Nothing on this page implies that a night owl is smarter, or that a morning person is less capable. Even the largest association, 0.1351 in Cohort 1, sits inside a distribution where individuals in every group span the whole range of scores. The 2024 release says as much, noting that the findings do not mean all morning people perform worse. These are statements about averages in samples, and the page on what the research says about signs of high intelligence keeps the same conclusion for individuals: being a night owl is not a usable marker of ability.

Self perception complicates the picture further. In the 2019 questionnaire studies, evening types rated their own intelligence higher than morning types did, and the effect persisted after controlling for tested intelligence. In the second study, narcissism fully accounted for the self view. A belief that night owls are smarter, in other words, may be held most firmly by night owls. The page on IQ and personality covers how traits relate to measured ability more broadly.

Nor does the evidence say that moving your bedtime will change your score. The studies here classified people or manipulated the hour of a test. None changed anyone's chronotype and measured IQ afterward. The page on whether you can increase your IQ covers what training and interventions have been tested.

The practical use is narrower and more concrete. If you are taking an online test, the evidence on synchrony supports choosing a window in which you are ordinarily alert, and it argues against a window that forces an extreme night owl into an 8:00 start or an extreme lark into the late evening. The page on how to prepare for an IQ test gives the same rule in practical terms, and the page on sleep and IQ covers rest before a sitting. ACIS is an online, self administered assessment without a proctor, with forms of about 45 minutes (Quick), about 110 minutes (Optimized) and about 175 minutes (Full Scale), breaks allowed and 30 days to complete, so the timing of each sitting is in the test taker's hands. It is not a clinical or diagnostic instrument, and it is not for hiring, school accommodations or admission to high IQ societies. The page on what is my IQ explains how to choose a test and read its result.

12 What the Evidence Supports, Stated Narrowly

Evening types score slightly higher on average on some cognitive measures in some adult samples, the association is small and sensitive to confounds, and the effect of testing at a mismatched hour is better supported than any trait difference. Each statement below is limited to the design behind it.

• Supported: In a 2011 meta-analysis, eveningness correlated 0.08 with cognitive ability (N of 1,519) and morningness minus 0.04 (N of 2,177), while the academic achievement pattern reversed.
• Supported: In a 2021 update of 30 studies and 11,160 participants, eveningness correlated 0.056 with intelligence and morningness minus 0.008 overall, and the morningness correlation turned negative in samples past young adulthood.
• Supported: In the 2024 UK Biobank cross sectional analysis of 26,820 adults aged 53 to 86, evening and intermediate types had higher global z-scores than morning types, with coefficients of 0.1351 and 0.0750 for evening types.
• Supported: The 13.5 percent headline figure corresponds to the coefficient 0.1351 multiplied by 100. A coefficient on a z-score scale is not a percentage difference in IQ.
• Supported: In a matched comparison, Mensa members did not differ from controls in corrected chronotype, and later work start times accounted for their later work day sleep.
• Supported: Performance depends on the match between a person's chronotype and the hour of testing in many studies, though not in all, and for different ability types in different studies.
• Not supported: That staying up late raises IQ, that night owls are more intelligent as a type, that a morning person scores lower, or that bedtime is a sign of ability.
• Not supported: Any individual inference. A population association of this size cannot be applied to one person.

The Standards for Educational and Psychological Testing (AERA, APA and NCME, 2014), published jointly by the three organizations and described on the Standards website, frame the same limit. Evidence has to support each intended interpretation of a score for a specified use, and the conditions of administration are part of what a score means. A chronotype label is not an intended interpretation of any intelligence test, so a score cannot be read as evidence of it, and a score gathered at a bad hour for the person should be read with that in mind. The APA Ethical Principles of Psychologists and Code of Conduct, Standard 9.02(a), says psychologists administer, score and interpret assessments in a manner and for purposes appropriate in light of the evidence on their proper use. For a general reader, that means treating a night owl headline as a hypothesis about groups and an IQ score as an estimate that depends on the hour and the conditions in which it was gathered. The technical manual documents what the ACIS report shows and does not show.

13 Sources Behind This Page

These sources span a meta-analysis, a systematic review, an experiment, matched and cross sectional comparisons, a longitudinal survey and press and expert commentary, so that a result from one design is not mistaken for another. Correlations, coefficients and sample sizes keep the context of the papers that report them. Where only an abstract or a news summary was read, the entry says so. Arithmetic marked as ours, such as conversions to IQ points, is not attributed to any author. Publication records were checked against Crossref on October 6, 2026.

• West R, Wong R, Park J, Lee S, Ekanayake Mudiyanselage D, Liu Z and Ma D. Sleep duration, chronotype, health and lifestyle factors affect cognition: a UK Biobank cross-sectional study. BMJ Public Health, 2024, volume 2, issue 1, article e001000 (full text read via Europe PMC).
• Imperial College London. Being a night owl is associated with mental sharpness, study shows. News release, July 11, 2024 (page opened October 6, 2026).
• Science Media Centre. Expert reaction to study on sleep, chronotype and cognition. Science Media Centre, 2024 (page opened October 6, 2026).
• Kanazawa S and Perina K. Why night owls are more intelligent. Personality and Individual Differences, 2009, volume 47, issue 7, pages 685 to 690 (full text read for the sample of 15,197 and the bedtimes; the correlations of 0.013 to 0.053 as summarized by Ujma and colleagues).
• ZME Science. Night owls versus early birds, a press account of the 2009 paper. ZME Science(page opened October 6, 2026 and rechecked October 7, 2026; used only for the Van Dongen comment and the note on controversy).
• Preckel F, Lipnevich A, Schneider S and Roberts R. Chronotype, cognitive abilities, and academic achievement: a meta-analytic investigation. Learning and Individual Differences, 2011, volume 21, issue 5, pages 483 to 492 (abstract read).
• Ujma P and Scherrer V. Circadian preference and intelligence, an updated meta-analysis. Chronobiology International, 2021, volume 38, issue 8, pages 1215 to 1229 (abstract read).
• Roberts R and Kyllonen P. Morningness-eveningness and intelligence: early to bed, early to rise will likely make you anything but wise. Personality and Individual Differences, 1999, volume 27, issue 6, pages 1123 to 1133 (record verified; result as summarized by Jankowski and colleagues).
• Goldstein D, Hahn C, Hasher L, Wiprzycka U and Zelazo P. Time of day, intellectual performance, and behavioral problems in morning versus evening type adolescents: is there a synchrony effect? Personality and Individual Differences, 2007, volume 42, issue 3, pages 431 to 440 (full text read).
• Ujma P, Baudson T, Bódizs R and Dresler M. The relationship between chronotype and intelligence: the importance of work timing. Scientific Reports, 2020, volume 10, issue 1, article 7105 (full text read via PubMed Central).
• Jankowski K, Díaz-Morales J and Vollmer C. Chronotype, time of day, and performance on intelligence tests in the school setting. Journal of Intelligence, 2023, volume 11, issue 1, article 13 (full text read via PubMed Central).
• Chauhan S, Vanova M, Tailor U, Asad M, Faßbender K, Norbury R, Ettinger U and Kumari V. Chronotype and synchrony effects in human cognitive performance: a systematic review. Chronobiology International, 2025, volume 42, issue 4, pages 463 to 499 (abstract read).
• Zajenkowski M, Jankowski K and Stolarski M. Why do evening people consider themselves more intelligent than morning individuals? The role of Big Five, narcissism, and objective cognitive ability. Chronobiology International, 2019, volume 36, issue 12, pages 1741 to 1751 (abstract read).
• Yamashita M, Shou Q and Mizuno Y. Association of chronotype with language and episodic memory processing in children: implications for brain structure. Frontiers in Integrative Neuroscience, 2024, volume 18 (page read).
• Gibbings A, Ray L, Smith D, van den Berg N, Toor B, Sergeeva V, Viczko J, Owen A and Fogel S. Does the early bird really get the worm? How chronotype relates to human intelligence. Current Research in Behavioral Sciences, 2022, volume 3, article 100083 (abstract fragment and a news summary only).
• American Educational Research Association, American Psychological Association and National Council on Measurement in Education. Standards for Educational and Psychological Testing, 2014, with the APA Ethical Principles of Psychologists and Code of Conduct, Standard 9.02(a). Citing them does not imply that their authors endorse ACIS or any other product.

14 Frequently Asked Questions