What Is the Average IQ, and Why Is It Always 100?

The average IQ is 100. It was 100 fifty years ago, and it will be 100 fifty years from now. That is not a remarkable fact about humanity holding steady — it is a consequence of how the scale is built. Understanding why explains several things that otherwise look paradoxical.

100 is set, not found

An IQ test does not measure intelligence in absolute units and then happen to average out at 100. The process runs the other way around.

When a test is developed, it is administered to a large standardisation sample chosen to represent the target population by age, sex, region, education and other characteristics. Whatever that sample scores becomes the reference. The raw scores are then transformed so that the sample's mean lands on 100 and its standard deviation on 15.

So the average is 100 for the same reason that water freezes at 0 °C: someone chose where to put the zero. If the convention had been mean 500 with SD 100, as the SAT once used, the same people would receive different numbers describing exactly the same relative standing.

This also means that "average IQ" is always relative to a particular reference group at a particular time. The phrase has no meaning in the abstract.

The Flynn effect: raw performance kept rising

Here is where it gets interesting. If the average is pinned to 100 by definition, can performance change at all? Yes — and it did, substantially.

In the 1980s the political scientist James Flynn documented a pattern that had been visible in test publishers' data for decades without anyone drawing the general conclusion: when people took an older version of a test, normed on an earlier generation, they scored well above 100. Successive generations were answering more items correctly than the generations the tests were normed on.

Across many countries and much of the twentieth century, the gain ran at roughly three points per decade on tests normed to a fixed reference. Over several decades that compounds into a striking figure: a person scoring at the average today would have scored well above average against a norm sample from the mid-twentieth century.

Test publishers respond by re-norming every so often. New standardisation data is collected, and the scale is recentred so the current population averages 100 again. This is why the average never moves in the published numbers even though raw performance did — the ruler is periodically rescaled to match.

A practical consequence: taking a test with outdated norms inflates your score. Some tests circulating online use norms that are decades old, which is one reason internet IQ results skew high.

Why did raw scores rise?

Genetic change cannot explain a shift this fast — a few generations is far too short a window. The rise was environmental, and the leading explanations are complementary rather than competing:

• More schooling, and different schooling. Both the length of education and its emphasis changed. Modern curricula spend far more time on abstraction, classification and hypothetical reasoning — precisely the habits of thought these tests sample.
• Better nutrition and health. Reduced childhood illness, better early nutrition and the removal of environmental toxins such as leaded petrol all plausibly raised cognitive development at the population level.
• Familiarity with abstraction. Flynn's own emphasis. Symbolic, rule-based thinking is now ordinary — in work, in media, in daily technology. Tasks that once felt alien to most adults are now routine, and test familiarity rose along with it.
• Smaller families and more parental investment. Fewer children per household generally means more individual attention during development.

Notably, the gains were not uniform across question types. They were largest on abstract, non-verbal reasoning tasks such as matrix problems, and much smaller on tests of vocabulary and general knowledge. That pattern fits the "familiarity with abstraction" account better than a story about general cognitive improvement.

Has it stopped?

In several developed countries, the gains appear to have slowed, halted, or in some datasets reversed since around the 1990s — sometimes called a negative Flynn effect. In other countries, particularly those where educational access expanded more recently, gains have continued.

The reversals are debated, and the explanations offered range from changes in education and media habits to shifts in the composition of the samples being tested. There is no settled consensus. It is safest to say the twentieth-century pattern was real and well documented, and that the current picture varies by country.

National average IQ comparisons

Tables ranking countries by average IQ circulate widely. They should be read with considerable caution, and mostly not relied upon at all.

The underlying datasets are assembled from studies that differ enormously in quality. Sample sizes are sometimes tiny and rarely nationally representative; the tests used differ between countries; the years of collection span decades without adjustment; and where data is missing, values have in some compilations been estimated from neighbouring countries rather than measured. These methodological problems have been the subject of sustained criticism.

Beyond the data quality issue, there is a deeper interpretive one. Test performance is strongly influenced by schooling, nutrition, health, literacy and familiarity with testing itself — all of which vary enormously between countries for reasons that have nothing to do with the capability of the people being tested. A cross-country score gap mostly measures differences in circumstances.

What this means for your own score

• "Average" means average for a reference group. A score is only interpretable alongside the question: compared to whom, tested when?
• Old norms inflate scores. If a test's norms are decades out of date, its results run high. Recency of norming matters more than most people realise.
• Self-selected samples are not the general population. This applies to every online test, including this one. People who choose to take an IQ test online are not a random sample of humanity, and any statistics drawn from them reflect that group, not the world.
• Scoring 100 means you are typical. That is what the number was designed to say. Half of everyone is below it, and the design guarantees that will always be true.