Explainers

Twelve myths about IQ tests

Einstein's IQ, the 10 percent brain, left-brain thinkers, Mensa cut-offs and the idea that IQ is fixed at birth. Sorted one at a time.

1. “Einstein had an IQ of 160”

Einstein was never given an IQ test. The figure is an invention, repeated so often that it acquired the texture of a fact, and it is usually quoted alongside a specific number for Marilyn vos Savant, Stephen Hawking or whoever else the list-maker reached for.

Hawking was asked about his IQ in a 2004 interview and replied that people who boast about theirs are losers. The other figures attached to historical figures are retrospective estimates produced by researchers who scored biographical evidence of childhood precocity — an exercise with no test, no norm group and no way to be wrong.

2. “We only use 10 percent of our brains”

Not a finding, never was. Functional imaging shows activity throughout the brain over the course of a day, and the metabolic cost of maintaining unused neural tissue would make such an arrangement biologically absurd. Damage to almost any region produces deficits, which is not what you would expect if nine-tenths were spare.

The myth's origin is murky and probably a garbling of early observations that a large fraction of brain cells are glia rather than neurons, combined with some very old self-help marketing.

3. “Some people are left-brained, some right-brained”

Lateralisation is real: language processing is left-dominant in most people, some spatial functions are right-dominant. Personality types built on top of that are not.

A large imaging study looking for individuals with globally stronger left or right networks found no such pattern. People are not analytical or creative because of which hemisphere they favour, and no IQ test measures anything of the kind.

4. “Your IQ is fixed from birth”

Rank order is fairly stable from late childhood, but individual scores move. Longitudinal studies of adolescents have found shifts of more than ten points in either direction across a few years, accompanied by structural brain changes.

Heritability estimates for adult IQ are substantial, commonly cited between 0.5 and 0.8, and heritability is routinely misread. It describes how much of the variation in a particular population tracks genetic variation. It does not mean a trait is unchangeable, and the same trait can have a different heritability in a different environment.

5. “Mensa requires an IQ of 140”

Mensa's requirement is the 98th percentile of the general population. On a scale with a standard deviation of 15, that is about 130 — not 140.

The confusion comes partly from the Cattell scale, which uses a standard deviation of 24, where the 98th percentile is around 148. This is a good illustration of why quoting an IQ figure without saying which scale it is on carries almost no information.

6. “IQ tests are completely culturally biased” and 7. “IQ tests are culture-free”

Both overstate. Modern test construction includes formal differential item functioning analysis, which flags items that behave differently across groups matched on overall ability, and flagged items get revised or removed. Crude cultural bias of the kind found in early twentieth-century tests — items about tennis courts and commodores — has largely been engineered out.

What has not been engineered out is the deeper layer. Comfort with the testing situation, familiarity with abstract classification, schooling in the language of the test and experience with timed assessment all vary systematically and all affect performance. “Culture-fair” tests such as Raven's matrices reduce the language component; they do not eliminate the rest, and the Flynn effect being largest on exactly those tests is the clearest evidence that they are not measuring something culture-independent.

8. “There are multiple intelligences, so IQ is meaningless”

Gardner's theory of multiple intelligences is widely taught and widely liked. It has not fared well empirically: the proposed intelligences either correlate with each other — in which case they are not independent — or they are abilities that most researchers would classify as talents or personality rather than intelligence.

That IQ tests miss a great deal about a person is true and important. It does not follow that what they do measure is meaningless, and the phrase “multiple intelligences” is usually used to reach that conclusion without arguing for it.

9. “A child's IQ test predicts their future”

Scores before about age six are notoriously unstable, and predictive power for adult outcomes from that age is modest. Scores become more stable through later childhood.

A single childhood score should never be treated as a prognosis. Clinical guidance is consistent on this and is routinely ignored by everyone except clinicians.

10. “A free online test can tell you your real IQ”

It can tell you your approximate band. It cannot tell you your number, for reasons covered at length in our accuracy piece: no administration control, self-selected norms, short length, unlimited retakes.

The tell is precision. A site that returns “your IQ is 127” after ten questions has chosen a satisfying output over an accurate one.

11. “140 is the genius threshold”

There is no official genius threshold on any modern scale, because “genius” is not a psychometric category. Terman used the word for scores above 140 in 1916; publishers abandoned that vocabulary decades ago precisely because it implied a verdict rather than a measurement.

What exists today are descriptive bands — extremely high, very high, high average, and so on. Nothing in the modern literature draws a line and calls one side of it genius.

12. “You cannot prepare for an IQ test”

You can, and it works, which is precisely why the claim persists — it would be convenient if the test measured something practice could not touch.

Practice effects on reasoning tests are well documented and worth several points, concentrated on the first repeat. This is why clinicians use alternate forms and observe minimum retest intervals, and why comparing your fourth online attempt with someone else's first tells you nothing.