The Average Rationalist IQ is About 122

In The Mystery Of Internet Survey IQs, Scott revises his estimate of the average LessWrong IQ from 138 to 128. He doesn’t explicitly explain how he arrived at this number, but the figure appears to be an average of the demographic-norm method (123) and the SAT method (134). However, using the information in his post, the SAT method doesn’t actually yield 134, but rather 123.

Here’s the breakdown: IQ has a mean of 100 and a standard deviation of 15, while the SAT has a mean of 1020 and a standard deviation of 194¹. A median SAT score of 1490 (from the LessWrong 2014 survey) corresponds to +2.42 SD. Using an SAT-IQ correlation of +0.80, the expected IQ z-score is $0.80 \times 2.42 = 1.93$, which corresponds to an IQ of 129. Subtracting 6 points (since according to the ClearerThinking test, people who took the SAT and remember their score have IQs about 6 points higher than the group average) brings the adjusted IQ estimate to 123.

The ClearerThinking test also provides a way to adjust self-reported IQs. If we apply the same correction to LessWrong self-reports, then based on a self-reported average of 138, subtracting 17 points (since people who report having taken an IQ test claim an average score of 131, but their tested average is only 114—a difference of 17 points) gives an adjusted IQ of 121.

Aggregating the data across all LessWrong and SSC surveys² with available information, the estimates—whether calculated from self-reported IQs, or self-reported SAT scores—consistently cluster around an average IQ of 122.

Scott also cites Spencer Greenberg’s ClearerThinking sample, which included both crowdworkers and people referred through Spencer’s social media. The crowdworkers had an average normed-IQ of 100, while the social media referrals had an average normed-IQ of 120. Scott notes that 120 is a plausible figure for Spencer’s friends and friends-of-friends (given Spencer’s background in math at Columbia). I expect Spencer’s social network to be similar in intelligence to rationalists, especially given the overlap between those populations. Thus, we have a piece of corroborating evidence for the estimate of 122.

I also have a small independent data point: a certain Discord channel (that was, at the time, full of rationalists and rationalist-adjacent people) take an IQ test (N = 12). I’m aware of the issues with internet IQ tests, but what makes this one useful is that there is a study in which participants took both this test and the WASI-II (an actual well-validated IQ test), so I can translate the Open Psychometric scores into actual IQ scores. Doing this yields a median IQ of 122.5 and a mean of 122.1, yet another corroborating piece of evidence for the 122 figure.

While some might think an average IQ of 122 sounds too low, it’s worth noting the average IQ of PhD holders is around 122.