The results indicated that children under the age of 10 cannot provide useful ratings of their own intelligence. A major reason is that younger children may not have the level of abstract thought needed to understand how intelligence would look in daily life, and they may struggle to see that abstract quality in themselves.
The authors also measured the children’s self-esteem. Measured IQ, self-esteem, and self-rated intelligence were all positively correlated, but there seems to be no causal relationship impact of self-esteem and IQ. Self-esteem had very little incremental validity over IQ when predicting IQ 2 years later.
The age 10 threshold for accurate self-assessment is really interesting. Before that, kids lack the metacognitive ability to evaluate their own thinking or compare themselves meaningfully to peers. The graph shows younger kids’ self-ratings are basically random relative to actual IQ, but by grade 6-8, there’s a clear positive relationship. What’s important is that self-reported intelligence only added about 1% to predicting future IQ beyond what actual IQ already predicted. This confirms that self-perception of intelligence is mostly just reflecting actual ability plus some self-esteem noise, not measuring something independently useful.
The correlation of 0.38-0.41 between self-rated and measured intelligence is actually pretty weak, meaning people are only moderately accurate at assessing their own cognitive abilities even in adolescence. The self-esteem finding is important because it shows that confidence in your intelligence doesn’t cause higher IQ or vice versa. They’re just correlated because smarter people tend to recognize they’re smarter (once they’re old enough to understand what that means). For practical purposes, this means asking people how smart they think they are is a poor substitute for actual testing, especially for anyone under 10 years old.
