Various theorists have tried to provide models to explain the Dunning–Kruger effect's underlying causes. The original explanation by Dunning and Kruger holds that a lack of metacognitive abilities is responsible. This interpretation is not universally accepted, and many alternative explanations have been proposed. Some of them focus only on one specific factor, while others see a combination of various factors as the cause.
Metacognitive The
metacognitive explanation rests on the idea that part of acquiring a skill consists in learning to distinguish between good and bad performances of the skill. It assumes that people of low skill level are unable to properly assess their performance because they have not yet acquired the discriminatory ability to do so. This leads them to believe that they are better than they actually are because they do not see the qualitative difference between their performance and that of others. In this regard, they lack the metacognitive ability to recognize their incompetence. This model has also been called the "dual-burden account" or the "double-burden of incompetence", since the burden of regular incompetence is paired with the burden of metacognitive incompetence. The metacognitive lack may hinder some people from becoming better by hiding their flaws from them. This can then be used to explain how self-confidence is sometimes higher for unskilled people than for people with an average skill: only the latter are aware of their flaws. Some attempts have been made to measure metacognitive abilities directly to examine this hypothesis. Some findings suggest that poor performers have reduced metacognitive sensitivity, but it is not clear that its extent is sufficient to explain the Dunning–Kruger effect. Another study concluded that unskilled people lack information but that their metacognitive processes have the same quality as those of skilled people. An indirect argument for the metacognitive model is based on the observation that training people in logical reasoning helps them make more accurate self-assessments. Many criticisms of the metacognitive model hold that it has insufficient empirical evidence and that alternative models offer a better explanation.
Statistical and better-than-average effect A different interpretation is further removed from the psychological level and sees the Dunning–Kruger effect as mainly a statistical artifact. It is based on the idea that the statistical effect known as
regression toward the mean explains the empirical findings. This effect happens when two variables are not perfectly correlated: if one picks a sample that has an extreme value for one variable, it tends to show a less extreme value for the other variable. For the Dunning–Kruger effect, the two variables are actual performance and self-assessed performance. If a person with low actual performance is selected, their self-assessed performance tends to be higher. Most researchers acknowledge that regression toward the mean is a relevant statistical effect that must be taken into account when interpreting the empirical findings. This can be achieved by various methods. Some theorists, like
Gilles Gignac and Marcin Zajenkowski, go further and argue that regression toward the mean in combination with other cognitive biases, like the
better-than-average effect, can explain most of the empirical findings. This type of explanation is sometimes called "noise plus bias". According to the better-than-average effect, people generally tend to rate their abilities, attributes, and personality traits as better than average. For example, the average
IQ is 100, but people on average think their IQ is 115. The better-than-average effect differs from the Dunning–Kruger effect since it does not track how the overly positive outlook relates to skill. The Dunning–Kruger effect, on the other hand, focuses on how this type of misjudgment happens for poor performers. When the better-than-average effect is paired with regression toward the mean, it shows a similar tendency. This way, it can explain both that unskilled people greatly overestimate their competence and that the reverse effect for highly skilled people is much less pronounced. This can be shown using simulated experiments that have almost the same correlation between objective and self-assessed ability as actual experiments. Some critics of this model have argued that it can explain the Dunning–Kruger effect only when assessing one's ability relative to one's peer group. But it may not be able to explain self-assessment relative to an objective standard. A further objection claims that seeing the Dunning–Kruger effect as a regression toward the mean is only a form of relabeling the problem and does not explain what mechanism causes the regression. Based on statistical considerations, Nuhfer et al. arrive at the conclusion that there is no strong tendency to overly positive self-assessment and that the label "unskilled and unaware of it" applies only to few people.
Science communicator Jonathan Jarry makes the case that this effect is the only one shown in the original and subsequent papers. Dunning has defended his findings, writing that purely statistical explanations often fail to consider key scholarly findings while adding that self-misjudgements are real regardless of their underlying cause.
Rational The rational model of the Dunning–Kruger effect explains the observed regression toward the mean not as a statistical artifact but as the result of prior beliefs. If low performers expect to perform well, this can cause them to give an overly positive self-assessment. This model uses a psychological interpretation that differs from the metacognitive explanation. It holds that the error is caused by overly positive prior beliefs and not by the inability to correctly assess oneself. For example, after answering a ten-question quiz, a low performer with only four correct answers may believe they got two questions right and five questions wrong, while they are unsure about the remaining three. Because of their positive prior beliefs, they will automatically assume that they got these three remaining questions right and thereby overestimate their performance.
Distribution of high and low performers Another model sees the way high and low performers are distributed as the source of erroneous self-assessment. It is based on the assumption that many low performers' skill levels are very similar, i.e., that "many people [are] piled up at the bottom rungs of skill level". This would make it much more difficult for them to accurately assess their skills in relation to their peers. According to this model, the reason for the increased tendency to give false self-assessments is not a lack of metacognitive ability but a more challenging situation in which this ability is applied. One criticism of this interpretation is directed against the assumption that this type of distribution of skill levels can always be used as an explanation. While it can be found in various fields where the Dunning–Kruger effect has been researched, it is not present in all of them. Another criticism holds that this model can explain the Dunning–Kruger effect only when the self-assessment is measured relative to one's peer group, but that it may fail when it is measured relative to absolute standards.
Lack of incentive A further explanation, sometimes given by theorists with an economic background, focuses on the fact that participants in the corresponding studies lack incentive to give accurate self-assessments. In such cases, intellectual laziness or a desire to look good to the experimenter may motivate participants to give overly positive self-assessments. For this reason, some studies were conducted with additional incentives to be accurate. One study gave participants a monetary reward based on how accurate their self-assessments were. These studies failed to show any significant increase in accuracy for the incentive group in contrast to the control group. == Practical significance ==