A Meta-Analysis on the Validity of Self-Assessed Intelligence through the Lens of the Multiverse

Do people actually know how smart they are? A large 2024 meta-analysis, which pooled 278 results from 115 independent samples and more than 36,000 participants, found that people’s own estimates of their intelligence do track their measured IQ, but only moderately (a correlation of about 0.30).

The result held up under thousands of alternative ways of analyzing the data, with 96% of them still showing a positive link. The match was not equally strong everywhere, though. It was tighter for numerical ability than for verbal or spatial skills, which the authors suggest may be because everyday life, such as splitting a bill or calculating a tip, gives us frequent feedback on our math but rarely on our spatial reasoning. It was also tighter among students than in the general population, possibly because students have more experience with tests and grades.

Sex, neuroticism, and the way the question was asked made little difference to the correlation. One caveat is that a correlation only shows that self-estimates and IQ rise and fall together. It says nothing about whether people over- or underestimate themselves in absolute terms, and other research suggests women tend to underestimate more. Overall, we have a decent but imperfect sense of our own abilities, and regular, varied feedback seems to sharpen it.

Link to article: 10.3390/jintelligence12090081