Does the Dunning-Kruger effect exist (whereby lower IQ people tend to overestimate their IQ), and do higher IQ people tend to underestimate their IQ?

Note: This is a section of a longer article. To go to the start, click here.

It is true in our data that those with lower measured IQ tended to overestimate their measured IQ, and those with higher measured IQ tended to underestimate it. However, whether or not this should be interpreted as a genuine Dunning-Kruger effect is up for debate. See our in-depth discussion of this question in our Dunning-Kruger analysis report here.

Here is what we did to investigate this question. First, after each set of intelligence related tasks participants took, we asked them to estimate in what percentile they think they scored on each such  task. Thus, for each person we were able to calculate the mean of their actual, objective performance on the set of intelligence  tasks (the mean score) and the mean of their subjective, self-estimated performance on those same tasks (the estimated score). That means that, for each participant, we took the mean of self-estimated performance on the tasks that specific person took, as well as the mean of the objective performance on the tasks that specific person took. We next converted all of these scores to so-called z-scores to have them all on the same scale (with mean 0 and standard deviation 1 across people). Finally, we fitted the line of best fit describing the relationship between actual objective scores and self-estimated subjective scores, together with the line that shows what the perfect relationship between these two scores would look like (i.e. if everybody was perfect at estimating their average scores). 

Here is the graph that shows these two lines. The curved line represents the actual relationship between objective and subjective scores while the straight dashed diagonal line represents what perfect self-estimations would look like. What can be seen on it is that, on the left side of the graph among lower levels of objective scores, the curved line is above the diagonal line and is flat. This means that in general people with lower scores tended to overestimate their performance.

On the contrary, on the right side of the graph among higher levels of objective scores, the curved line is below the diagonal line, meaning that people with higher IQ tended to underestimate themselves. However, this time the line is going upward which means that, although in general they underestimated their performance, higher scoring people estimated their performance to be somewhat better (meaning that the higher their objective score was, the higher their subjective estimation was too). 

One thing to note is that recently the Dunning-Kruger effect was questioned and presumed to be mostly a statistical artefact, i.e., to appear because wrong kinds of statistical analyses were used (e.g. Gignac & Zajenkowski, 2020). We will not go into details about this here, but two valid statistical approaches were suggested for testing the Dunning-Kriger effect: Glejser heteroskedasticity test and the test of quadratic effects. We analyzed our data using these two approaches too and the results confirmed the existence of the effect. On the other hand, there are some doubts about the interpretation of all of the results and their connection to a "real" Dunning-Kruger effect. As mentioned, you can find more details in our Dunning-Kruger analysis report here. 

What do the other studies say?

Recent research challenges the generality of the Dunning-Kruger effect in intelligence self-assessment. While some studies found limited support for the effect (Hofer et al., 2021; Dunkel et al., 2023), others argue it's primarily a statistical artifact (Gignac & Zajenkowski, 2020).In conclusion, recent literature suggests that the effect's magnitude may be minimal, affecting only a small portion of the population with very low IQ scores (Gignac, 2024).

Takeaways

• Our study found evidence for the existence of the Dunning-Kruger effect, though the interpretation of such an effect is complicated, and may not mean what it is generally believed to mean, as we discuss in our report here.

If you'd like to read the full report, of which this is a section, as one long PDF, you can download it here.

And if you'd like to understand where your intellectual strengths and weaknesses lie, try the cognitive assessment tool that we developed out of this research:

Launch the Cognitive Assessment!