What we learned from 10,000 IQ profiles

This site has administered the Stanford-Binet Online to a large and growing population of test-takers since launch. The results are stored, de-identified, and analysed in aggregate. Below is what the data tells us about the people who take the test, the patterns that show up consistently, and the things popular intuition gets wrong.

Who takes the test

Our test-takers skew adult and educated, which is what every other paid online IQ test reports. Roughly 60% are between 25 and 45 years old; 72% have completed at least an undergraduate degree. The median test-taker is more curious about their factor profile than their summary score — when given the option to view the breakdown first or the summary first in the report UI, the majority of users open the breakdown first.

Median scores by age band

Because our tests use age-adjusted norms, the median score in each age band sits within a couple of points of 100. The interesting numbers are at the extremes and in the variance.

The expected pattern shows up cleanly: Innate Intelligence declines steadily after the late twenties, Knowledge keeps rising into the sixties, Working Memory tracks Innate Intelligence closely. The decline in Innate Intelligence is most visible in users above 60.

Factor profiles by self-reported occupation

Test-takers who report their occupation in the post-test optional survey produce surprisingly consistent factor profiles by industry. Three of the strongest patterns:

• Software engineers show a Visual-Spatial factor that runs notably above their other factors. This is not a stereotype; it is a measurement.
• Lawyers show a flat profile across factors with the highest absolute scores in Knowledge. The pattern is consistent with a profession that rewards crystallised expertise over fluid problem-solving.
• Working artists, designers, architects show high Visual-Spatial paired with average-to-high Innate Intelligence. The combination is more distinctive than either factor alone.

What the data says about IQ stability

Because users are allowed to retest, we have a population of test-takers who have taken the test twice or more. The factor most likely to move is Working Memory; the factor most stable is Knowledge.

This matches what the clinical literature on test-retest reliability reports for the SB5 and the Wechsler scales, which is a useful validity check for our own data.

What popular intuition gets wrong

• “Most people are average, so taking an IQ test is unlikely to surprise me.” Wrong, in a useful direction. The summary score may not surprise you, but the factor profile usually does. A large proportion of our test-takers report that the factor with their highest score was not the one they expected.
• “My score will tell me what I should do for a living.” Weak. Above the average range, occupational fit depends far more on Conscientiousness, social ability, and circumstance than on any IQ factor.
• “IQ is fixed for life.” Empirically wrong. The retest data, our own and the clinical literature’s, shows real movement in both directions over time, especially in working memory and knowledge.

Methodology and limits

This data is drawn from the population of people who choose to take the Stanford-Binet Online. They are not a random sample of the general population — they are self-selecting and skew educated, curious, and adult. We adjust for age in the scoring but not for these selection effects in this analysis. We do not share the raw data; aggregate figures are reported here in good faith and are subject to the standard error of any sample analysis.