Can Your IQ Change Over Time? What Retesting Shows

Two claims circulate about IQ and both are stated with total confidence. The first is that your IQ is fixed at birth and nothing moves it. The second is that IQ is fluid, changes constantly, and therefore means nothing. There is an unusually direct way to test them, because a group of Scottish children sat an intelligence test in 1932 and a substantial number of them sat the same test again nearly seventy years later.

The test that was taken twice, 68 years apart

On a single day in June 1932, the Scottish Mental Survey gave the Moray House Test to almost every eleven-year-old attending school in Scotland. Decades later, researchers in Edinburgh realised the records had survived and tracked down surviving participants living in the Lothian region. Those born in 1921 sat the identical test again at the age of 79, and later at 87 and 90. It is one of the longest follow-up intervals in the history of psychology, using the same instrument at both ends.

The raw correlation between the age-11 and age-79 scores was about 0.66. Once the analyses correct for the imprecision in each measurement, the estimate rises further, into the region of 0.7 and above. Squared, a correlation near 0.7 implies that something approaching half the variation in scores in old age was already present in the same people as children — a result that is genuinely remarkable, and that is routinely reported as though it settled the argument in favour of fixity. It does not.

• A correlation of 0.66 is a long way from 1.0, and the gap is not noise. Real people moved substantially up and down the ranking over their lifetimes.
• It describes rank order within a group, not the absolute level of anyone's ability. Everyone can decline while the ordering between them is preserved.
• Part of the imperfection is measurement error rather than genuine change, which is why the corrected figures are higher — and why our explainer on why a score is a range, not a number is the necessary companion to any claim about change.
• The finding is about a cohort followed through one particular century in one particular country. It is strong evidence about stability in general and not a law.

Three different questions wearing one hat

Most disagreements here happen because people are answering different questions. Separating them makes the evidence line up cleanly:

• Does the reported number move between two sittings? Yes, routinely, and mostly for uninteresting reasons — measurement error, sleep, the room, and the practice effect described in our piece on what a retest actually measures. This is movement in the score, not in the person.
• Does your position relative to your peers move over years? Yes, but less than people expect. This is the question the Lothian correlation answers, and the answer is substantial stability with real individual movement inside it.
• Does the underlying ability itself change with age? Yes, and differently for different abilities — reasoning under time pressure and accumulated knowledge follow different trajectories, which our piece on when intelligence actually peaks sets out.
• Can deliberate intervention move it? Partly. Education is the intervention with the strongest evidence, at roughly 1 to 5 points per additional year in pooled analyses, which we cover in our piece on average IQ by education level.

The teenage years look less settled

Stability across a lifetime does not imply stability across every stretch of it. In 2011 a team publishing in Nature retested 33 adolescents who had first been assessed at ages 12 to 16, returning to them around three years later. Both verbal and non-verbal scores moved considerably in individual participants, in both directions, and the changes corresponded to differences in grey matter in regions related to the abilities concerned — verbal change tracking a speech-related area, non-verbal change tracking a motor-related one. The pairing of behavioural change with structural change is what made the study notable.

What this means for a score you are holding

If you took a test years ago, the sensible reading is that the result still carries information about you and carries less of it than it did. Rank order is largely preserved, so a result well above or below average then is more likely than not to be so now. But the confidence band around the original score has effectively widened with time, and if the number matters for a decision, the answer is a fresh sitting rather than an argument about the old one.

If you are hoping to change the number, the evidence supports a narrow and unglamorous set of things. Education has the strongest causal case. Sleep, health and testing conditions affect what you can demonstrate on the day, as our piece on what sleep loss does to test performance describes. Repeated practice on the test itself raises the score without raising the ability, which is the least useful kind of change available and the easiest to buy. Whatever your result, convert it to a percentile — the share of people who scored lower — with our IQ percentile calculator, because a percentile survives translation between scales in a way a bare number does not.

The honest answer to whether IQ can change is that it changes less than the optimists claim and more than the fatalists claim, and that the interesting quantity is not whether it moves but how much. Around two-thirds of a correlation between eleven and seventy-nine is the best single fact available, and it is a fact with two halves. Half of what you are measured as in old age was visible in childhood. The other half was not, and nothing in the data says it was determined in advance. If you want to know where you sit today rather than where you sat, you can take the assessment and find out.

Common questions