Home › Personality Types › What an IQ Test Measures and What It Includes › Average IQ Score and What It Means

The Average IQ Score, and What Being Average Actually Means

The average IQ score is 100, and it is 100 because the scale is defined that way rather than because anyone measured it. That one fact explains the Flynn effect, re-norming, and why the average IQ by country tables circulating online do not hold up.

6 sources cited10 min read
Average IQ Score and What It Means: Seven adults of different heights stand facing a sand-colored wall, the middle one glancing back.

In short

The average IQ score is 100 on every major test, not because that is what people turned out to score but because the scale is built to put the reference group's average at 100 with a standard deviation of 15. Raw performance on the same questions rose through most of the twentieth century, by roughly three points a decade across a meta-analysis of 271 samples from 31 countries, each new edition of a test resets the average back to 100 to absorb that rise, and in several countries the rise has stalled or reversed. Being near average predicts very little about one person's life, and the average IQ by country tables that circulate online rest on data that has not survived review. Psychology.com's IQ Test is a free reasoning quiz and does not produce an IQ score.

The average
100, set by definition on every major test
Why it never moves
Each new edition is re-normed to put the average back at 100
Twentieth-century rise
About 3 points a decade in raw performance
Recent trend
Stalled, and reversed in some countries
Country rankings online
Built on data that has not survived review
Test
IQ Test: Free Reasoning Quiz

The average is 100 because someone decided so

When a publisher builds an intelligence test, it gives the questions to a large sample chosen to resemble the population in age, sex, region, education and other characteristics. Everybody's raw performance is ranked. The ranks are then converted onto a scale whose midpoint is set to 100 and whose standard deviation is set to 15. The average comes out at 100 because the conversion puts it there.

That makes the question what is the average IQ test score slightly different from how it sounds. The answer is not an empirical finding about human beings. It is a statement about the arithmetic: on any properly normed test, half the reference sample scores above 100 and half below, and about two thirds of them land between 85 and 115.

It also means that an average score is the most ordinary thing a test can report, and that is worth saying plainly because most people who look this question up are quietly asking whether they are all right. The middle of the scale is where most people are. A score of 97 and a score of 103 describe performances that nobody could tell apart, and both of them describe the large majority of competent, capable adults.

Re-norming: why the average never moves

Every decade or two, publishers collect a fresh reference sample and re-norm the test. The questions may be updated, but the important change is in the comparison: your raw performance is now ranked against a newer group of people. Because the scale is redefined so that this new group averages 100, any improvement in raw performance between editions disappears from the reported numbers.

This has a counterintuitive consequence that catches people out. The same raw performance can earn a lower score on a newer edition of a test than it did on the older one, because the comparison group got harder to beat. Scores from different editions are not the same currency, and a competent report always names the instrument and the edition for that reason.

Re-norming is also why nobody can answer the question has the average IQ gone up by looking at reported scores. By construction, the answer is always no. To see whether performance moved you have to compare raw results on the same items across cohorts, which is exactly what the research on the Flynn effect does.

The Flynn effect: raw performance really did rise

In 1987 James Flynn published an analysis of data from 14 nations showing substantial gains in test performance across generations, large enough that the standardization samples of the 1930s would score well below average against those of the 1980s. The finding was replicated widely enough that it took his name.

The best current estimate of its size comes from a 2015 meta-analysis by Jakob Pietschnig and Martin Voracek, pooling 271 independent samples covering almost four million participants in 31 countries from 1909 to 2013. They found annual gains of about 0.41 points for fluid reasoning, 0.30 for spatial ability, 0.28 for full-scale scores and 0.21 for crystallized knowledge, which is roughly three points a decade overall. Reasoning gained faster than knowledge.

What the effect does not show is that people became more intelligent in any general sense. The authors themselves argued that gains of that size over a century are better read as improvements in specific abilities than as a rise in general cognitive ability. Better nutrition, smaller families, more schooling, more years of test-like tasks and far more exposure to abstract visual material are all plausible contributors, and none of them require the underlying capacity to have changed.

The stall, and the reversal

The same meta-analysis found that gains have been smaller in recent decades. In several high-income countries they appear to have stopped, and in some they have gone into reverse, which has produced a run of headlines claiming that people are getting less intelligent.

The most careful study of a reversal is Bernt Bratsberg and Ole Rogeberg's 2018 analysis of Norwegian military conscription records for men born between 1962 and 1991. Scores rose, peaked and then declined across those cohorts, and the researchers were able to recover the whole pattern, rise and fall alike, from differences between brothers raised in the same family. That rules out explanations based on changing family composition or inherited factors and points firmly at the environment.

Taken together, the rise and the reversal say the same thing about test scores. They move with the conditions people grow up in, which is informative about societies and close to useless as a measure of the worth of any generation. They also undercut the idea of an IQ score as a fixed personal quantity, because a quantity that drifts several points a decade with schooling and circumstance is not fixed.

What an average score actually predicts

At the level of large groups, measured ability does correlate with later outcomes. Tarmo Strenze's meta-analysis of longitudinal studies found corrected correlations of .56 with educational attainment across 59 samples and 84,828 people, .43 with occupational attainment, and .20 with income. Those are real associations and they have been replicated.

Hold that next to the question this page is about. Being average on a measure that predicts education at around .56 across tens of thousands of people is not a finding about your prospects, because the same meta-analysis found parental background and school grades predicting education nearly as strongly. Three predictors of roughly equal size do not add up to a verdict on anybody.

Translate that down to one person and almost nothing is left. A correlation around .5 across tens of thousands of people means the two measures move together loosely, and that people fifteen points apart routinely finish in the reverse order on anything you would care about. An average score forecloses nothing: not a degree, not a profession, not a field that interests you. It tells you that you handled a set of tasks the way most people handle them.

The average IQ by country tables

Search for an average IQ and you will quickly meet a table ranking countries by a single number each. Those tables trace back to a small number of compilations that assembled published and unpublished test results from many countries and estimated a figure for the rest. They are presented as data and they do not hold up.

The problems start with the samples. Many of the underlying studies used small groups, or groups never intended to represent a country: schoolchildren in one region, people recruited at a single institution, or samples whose selection the compilation did not describe. A systematic review of African test data by Jelte Wicherts, Conor Dolan and Han van der Maas found that one widely cited estimate below 70 for sub-Saharan Africa depended on which studies were left out, and that reviewing the published literature systematically produced an average closer to 82 relative to British reference data.

Then come translation and test bias. A reasoning test built and normed in one country, written in one language, assuming familiarity with a particular kind of schooling and with abstract pictorial puzzles, is not automatically measuring the same thing somewhere else. Wicherts and colleagues ended their review by calling for exactly the psychometric work that would be needed to establish whether these instruments measure comparably across populations, work that has largely not been done.

Why those tables have been withdrawn from the literature

This is not just a disagreement between researchers. In March 2024 the journal Proceedings of the Royal Society B retracted a 2010 paper that had used one of these national datasets, with the editors noting that the data on between-country variation in intelligence had been the subject of several critiques. A retraction on the basis of the underlying data is as clear a signal about data quality as academic publishing gives.

There is also a structural reason to distrust a country-level number even if the sampling were fixed. The Flynn effect means a national figure is dated the moment it is collected, and gains have been largest in countries that were catching up on schooling and nutrition, so comparing a sample taken in 1995 with one taken elsewhere in 2015 compares two decades as much as two places. If you came here wanting to know how your country ranks, nobody has the data to tell you.

If you mostly want to know whether you are normal

Most people who search for the average are not doing comparative psychometrics. They have a number from somewhere, or a worry, and they want to know whether they are fine. Almost everybody reading this is fine, and a test score is a poor instrument for settling the question anyway.

If the worry is about a specific thing you have noticed, that detail is far more useful than a score. Losing the thread in meetings, rereading paragraphs, taking much longer on work that used to be quick, forgetting conversations: those are observations a clinician can do something with. Bring the examples, not the number. If what you have noticed is a change from your own usual functioning, that matters regardless of how any quiz turned out.

If the worry is really about comparison, it helps to notice that the scale was designed to produce a crowd in the middle. Ten points either side of 100 covers a very large share of the population, and the people in that span include most of the competent adults you know. There is no version of this topic in which being near the average is a finding about your worth.

What the research says

The well-supported parts of this topic are the scaling and the group-level associations. Setting a reference sample's mean to 100 is a transparent statistical choice, modern batteries are highly reliable, and links between measured ability and education, occupation and health outcomes have held up in large prospective studies. The Flynn literature is solid too, and it is the most useful corrective available, because it shows the same tests producing different raw performance across cohorts and so limits how far a single score can be treated as a fixed personal property.

The weak part is the country-level comparison industry, and it is weak in a specific way rather than merely contested: the sampling is undocumented or unrepresentative in many of the source studies, measurement comparability across languages and education systems has not been established, and at least one journal has withdrawn a paper over the data. A number with that provenance should not be quoted, and it is not quoted here.

How to find out where you land

Psychology.com's IQ Test is 30 original text-only questions, untimed, with no speed bonus, and it reports an estimated range with a breakdown across five categories of reasoning rather than an IQ figure. It is self-administered, unproctored and unnormed, which means it cannot place you against other adults and we do not pretend otherwise. An optional in-depth report is available if you want a longer write-up.

For a real comparison against a reference group you need an individually administered battery given by a qualified examiner, which is what the Stanford-Binet and the Wechsler scales are for. If a decision depends on the result, that is the route. If curiosity does, a quiz is a reasonable afternoon and the IQ scale page will tell you how to read whatever number you end up with.

FeatureWhat a population average can supportWhat one person's score can support
Comparison over timeYes, if the same items and raw scores are usedOnly against the same edition, with a margin
Comparison between placesOnly with documented, comparable samplingNot at all
Explaining group outcomesPartly, alongside income, schooling and healthNo
Predicting one lifeNoNo
Spotting a change in someoneNoOnly against that person's own earlier result
Deciding what someone may attemptNoNo

Key takeaways

  • The average IQ is 100 by construction: the reference sample's midpoint is set to 100 and one standard deviation to 15, so the average cannot drift.
  • Raw performance on the same tests rose through the twentieth century by roughly three points a decade, pooled across 271 samples in 31 countries, and re-norming hides that rise from reported scores.
  • Gains have slowed and in some countries reversed; Norwegian conscript data recovered both the rise and the fall from differences between brothers, pointing at environment.
  • Being near average predicts very little about one person: parental background and school grades predicted educational attainment almost as strongly as measured ability in the same meta-analysis.
  • Average IQ by country tables rest on undocumented or unrepresentative samples and untested cross-language comparability, and a journal retracted a 2010 paper that relied on them.

Want to see how you handle the reasoning?

Our free IQ Test reports a range and a five-category breakdown, and deliberately never converts either into a number you could compare with 100. It is a reasoning quiz, not an IQ score.

Take the free IQ Test

Cite this source

Fontane Pennock, S. (2026, November 8). Average IQ Score and What It Means. Psychology.com. https://psychology.com/types/average-iq-score

Frequently asked questions

What is the average IQ score?

100 on every major test, because the scale is defined that way. The reference sample's midpoint is set to 100 and one standard deviation to 15 points. On that scale about 68 percent of people fall between 85 and 115. The figure is a statement about how the scale was built rather than a measurement result, which is why it is the same on every properly normed instrument.

Has the average IQ gone up over time?

Reported scores have not, because every new edition is re-normed to put the average back at 100. Raw performance on the same questions did rise: a meta-analysis of 271 samples from 31 countries between 1909 and 2013 found gains of roughly three points a decade, larger for reasoning than for knowledge. In several countries those gains have now stalled or reversed.

Is an IQ of 100 good?

It is the most ordinary result the test can give, which also makes it the most common. The scale is designed to put a crowd in the middle, so a score in the nineties or low hundreds describes a very large share of capable adults. It closes off nothing, and a score five points either side of it describes a performance nobody could distinguish from yours.

Are the average IQ by country rankings real?

They are not data you should rely on. Many of the underlying studies used small or unrepresentative samples, some selection procedures were never described, and whether the same test measures comparably across languages and school systems has not been established. In 2024 a journal retracted a 2010 paper that used one of these datasets, citing the critiques of it.

What does it mean if my score is slightly below average?

On its own, very little. A single result is affected by sleep, mood, pain, medication, distraction, unfamiliar language and how much written arithmetic you have done recently, and a professional report always states a confidence interval, usually about five points either side of the score. If you have noticed a real change in how you function day to day, bring that observation to a clinician rather than the number.

Why is the Flynn effect important if scores still average 100?

Because it shows the same test producing different raw performance from one cohort to the next, which limits how far a score can be treated as a fixed personal quantity. It also explains why scores from different editions are not comparable, and why a performance that scored above average against a 1970s reference sample may be average against a current one.

Can I find my own percentile from a free online test?

No. A percentile requires a representative reference sample, a controlled administration and a qualified examiner. Our own quiz is untimed, unproctored and unnormed, so it reports an estimated range of quiz performance and deliberately does not convert that into an IQ figure. For a genuine comparison you need an individually administered battery.

Related types

References

  1. Flynn JR. Massive IQ gains in 14 nations: What IQ tests really measure. Psychological Bulletin. 1987;101(2):171-191. doi.org
  2. Pietschnig J, Voracek M. One Century of Global IQ Gains: A Formal Meta-Analysis of the Flynn Effect (1909-2013). Perspectives on Psychological Science. 2015;10(3):282-306. doi.org
  3. Bratsberg B, Rogeberg O. Flynn effect and its reversal are both environmentally caused. Proceedings of the National Academy of Sciences. 2018;115(26):6674-6678. doi.org
  4. Wicherts JM, Dolan CV, van der Maas HLJ. A systematic literature review of the average IQ of sub-Saharan Africans. Intelligence. 2010;38(1):1-20. doi.org
  5. Retraction: Parasite prevalence and the worldwide distribution of cognitive ability (2010), by Eppig, Fincher and Thornhill. Proceedings of the Royal Society B. 2024;291(2017):20240463. doi.org
  6. Strenze T. Intelligence and socioeconomic success: A meta-analytic review of longitudinal research. Intelligence. 2007;35(5):401-426. doi.org
Important: A personality type, attachment style, or result pattern is an educational way of describing tendencies, not a diagnosis, a fixed identity, or a substitute for professional care. If you are struggling, reach out to a licensed mental-health professional.