What is the average IQ?

The average IQ is 100. Not approximately 100, and not 100 as a finding — exactly 100, because the scale is built so that the average of the reference population lands there.

100 is a definition, not a measurement

When a test is normed, a large sample sits the battery and their raw performance is converted to a scale with a mean of 100 and a standard deviation of 15. The conversion is chosen to produce those numbers. So asking what the average IQ is is a bit like asking what the average percentile is: the answer is fixed by the arithmetic.

The interesting question is not the average but the spread. Fixing the standard deviation at 15 is what gives the scale its meaning: it sets how far apart two scores have to be before the difference is worth anything.

How the spread works out

Because scores are normally distributed, the standard deviation tells you exactly how the population divides:

85 to 115about 68% of people(within one standard deviation)
70 to 130about 95% of people(within two)
55 to 145about 99.7% of people(within three)

The practical consequence: most people are average, and being average is the unremarkable default rather than a verdict. A 30-point band holds two thirds of everyone.

Why averages drift over time

Raw performance on IQ tests rose steadily through the twentieth century across many countries, a pattern named the Flynn effect. Because the scale is pinned to 100, this shows up not as rising scores but as repeated re-norming: tests are periodically restandardised so the average returns to 100.

This is why a score from a test normed in 1985 is not comparable to one from a test normed last year. The yardstick moved.

National and group averages

Published tables of average IQ by country circulate widely and are worth treating with suspicion. They typically combine small, unrepresentative samples collected decades apart, using different instruments normed on different populations, and then present the result to one decimal place. The precision is fictional.

A score is only interpretable against the population it was normed on. Comparisons across populations that were never jointly normed are not measuring what they appear to measure.

Where your own score sits

If you want your position on this scale rather than the population’s, the band breakdown and the percentile table convert a number into a rank. This test reports both, plus the confidence interval around the estimate.

Keep reading

Find out where you actually land.

70 calibrated items, about thirty minutes, no account.

Start the test