IQ and age

What does change

Raw performance on speeded and novel reasoning tasks does change across adulthood. At the same time, knowledge and vocabulary build. Many tasks people care about at work and at home draw on both, which is part of why an IQ score is only one input among many.

Genetics matter too, in a way that changes with age. Twin and family studies find that the heritability of intelligence rises from about 20% in infancy to perhaps 80% in later adulthood (Plomin & Deary, 2015). Heritability describes what explains differences across a population; it does not say how much any single person's ability is fixed, and it does not mean environment stops mattering.

Children and IQ

Age norms are especially important for children, whose abilities change quickly from year to year. A child's IQ score only means something when it compares them with children of the same age, on a test built and normed for that age. If you have a question about a child's development or schooling, a licensed psychologist can give an age-appropriate standardized assessment.

How stable is your rank over a lifetime?

Fairly stable, but not fixed. In the Lothian Birth Cohort 1921, scores on the same test taken at age 11 and again at 90 correlated .54, or .67 after correction for range restriction (Deary et al., 2013). That is strong continuity across almost 80 years, with plenty of room for individuals to move. More in intelligence across the lifespan.

Reading an old score

Many people have a number from a childhood assessment and wonder how it compares with a score today. A few things are worth checking before you compare:

• Which test and which scale. Most modern tests use a standard deviation of 15, but older Stanford-Binet forms used 16 and the Cattell Culture Fair III scale uses 24 (Wikipedia: IQ classification). The same percentile prints as a different number on each. The percentile to IQ converter shows all three.
• When the test was normed. Average performance rose through much of the 20th century, by about 2.8 points per decade across many countries (Pietschnig & Voracek, 2015). A score from a test with old norms can read higher than the same performance would against current ones.
• Measurement error. Clinical scores carry a confidence interval, often around plus or minus 5 points at 95% (Wechsler, 2008). Two scores a few points apart may not differ at all.
• Practice. Each time you take a cognitive test, you get a little more familiar with the format. Retakes score about a quarter of a standard deviation higher on average (Hausknecht et al., 2007).

Put together, a childhood score and an adult score are best compared as rough percentiles, not as exact numbers.

What this means for IQ Octopus

IQ Octopus is designed for adults. It does not ask your age and does not apply age norms. Its puzzles are mostly fluid-reasoning tasks, the kind whose raw performance peaks earliest. Two practical consequences follow:

• If you are well past your twenties, your IQ Octopus estimate may read lower than an age-normed clinical score would, because you are not being compared only with people your own age.
• Pace does not change your estimated IQ on IQ Octopus. Only which puzzles you answer correctly counts, and the time limit is generous, so slower, careful work is not penalized.

Read the full method on how it works, and for the population-level story of how average scores have shifted across generations, see the Flynn effect.