Why the Average IQ Score Is Always 100

The average IQ score is 100 because the people who built the scale decided it would be. It is a unit, in the same way that zero degrees Celsius is a unit, and it is re-imposed every time a test is renormed against a fresh sample. Most readers meet the figure as though it were a finding about humanity. It is not; it is the origin of a coordinate system.

Taking that seriously changes several things that otherwise look puzzling — including why a whole century of rising raw performance never moved the average by a single point.

The mean is set, not found

Reported IQ scores use a standard-score metric with a mean of 100 and a standard deviation of 15 — the convention the psychometric literature states outright, including a 2026 meta-analysis of Wechsler profiles that standardised its comparisons against a control with a mean of 100 and a standard deviation of 15.

Getting there takes one conversion. A publisher administers the battery to a standardisation sample, records how that sample performed at each age, and builds tables that map raw performance onto the fixed metric. The centre of the sample lands on 100 because the table sends it there. So "the average IQ is 100" is not a claim that can be false. It is a description of the tables, and it would be equally true on a scale centred on 500 or on 0.

Anything genuinely informative about a score lives in the comparison the tables encode — who was in the sample, when, and how they performed. That machinery is set out in our explainer on how an IQ score is calculated.

What actually moved during the twentieth century

Raw performance on these tasks rose, substantially, across generations. The best single summary is the 2014 meta-analysis by Trahan, Stuebing, Hiscock and Fletcher in Psychological Bulletin, which pooled 285 studies with 14,031 participants, all of them involving people who took two intelligence tests built on different normative bases.

Their headline figure was a mean of 2.31 standard score points per decade, with a 95 per cent confidence interval of 1.99 to 2.64. For a subset of modern comparisons the estimate was 2.93 IQ points per decade (95 per cent confidence interval 2.3 to 3.5), which matched the long-standing rule of thumb of roughly three points a decade. The authors reported the pattern as robust across age groups, measures, samples and performance levels.

This is the effect named after James Flynn, who documented it across many countries. The finding is about raw performance, and it is genuinely striking — but it is not a finding about the average score, because the average score cannot move.

Renorming: where the gain goes

Here is the mechanism that connects the two facts, and it is the part most explanations skip.

When a publisher renorms, it recruits a new standardisation sample and rebuilds the tables so that the new sample's average lands on 100. If the new sample performs better in raw terms than the old one did, then the same raw performance now converts to a lower reported score than it used to. The rising tide is absorbed into a stricter conversion rather than a rising mean.

Pearson describes the Wechsler Adult Intelligence Scale, Fifth Edition — published in 2024 and covering ages 16:0 to 90:11 — as carrying updated norms reflecting a contemporary population. That is renorming in its ordinary form, and it is why a score has a shelf life. A result from an earlier edition was a rank against a different group of people, and quoting it years later without saying which edition produced it leaves out the only thing that made it interpretable.