When Ignorance Begets Confidence: The Classic Dunning-Kruger
Henry Kang , Editor | Jul 3, 2014
Topic category: Zen Enlightenment

The Dunning-Kruger effect describes a cognitivebias in which people perform poorly on a task, but lack the meta-cognitive capacity to properly evaluate their performance. As a result, such people remain unaware of their incompetence and accordingly fail to take any self-improvement measures that might rid them of their incompetence. In my extended version of the Dunning-Kruger effect, this also leads to extensive Fremdscham in others, but this is not covered by the original research...

Dunning and Kruger reported their seminal experimental findings more than ten years ago in the Journal of Personality and Social Psychology, and the Dunning-Kruger effect has since become a popular culture item, similar to inattention blindness, or cognitive dissonance. For most of my readers therefore, the general concept might be very old news indeed. Nonetheless, I think it is original research, well worth reconsidering:

Finding the Effect

The first of the original Dunning Kruger experiments featured a group of undergraduate students who were asked - just as they walked out of an exam - to rate their performance for the class just completed. In particular they were asked how well they had mastered the course material, and what they predicted their raw score to be for the test just taken.

After comparing the student's own impressions with their actual performance, a clear pattern emerged in Dunning and Kruger's data: Worse students grossly overestimated their own performance, while top students somewhat underestimated theirs. You get a clear sense of the extremety of the poor student's penchant to overestimate their own performance, when you consider these results: For the bottom quartile, while their actual performance may have

"put them in the 12th percentile, they estimated their mastery of the course material to fall in the 60th percentile and their test performance to fall in the 57th".

Bottom performers tended to overestimate their performance by roughly 30%; a general pattern that has been replicated many times over since.


"Participants taking tests in their ability to think logically, to write grammatically, and to spot funny jokes tend to overestimate their percentile ranking relative to their peers by some 40 to 50 points, thinking they are outperforming a majority of their peers when, in fact, they are the ones being outperformed."

As Dunning and Kruger point out in an overview of the literature that followed their research:

"This pattern also emerges in more real-world settings: among debate teams taking part in a college tournament and hunters quizzed about their knowledge of firearms just before the start of hunting season; among medical residents evaluating their patient-interviewing skills; and among medical lab technicians assessing their knowledge of medical terminology and everyday problem-solving ability in the lab."

Dunning-Kruger as a Vicious Cycle

Dunning and Kruger often refer to a "double curse" when interpreting their findings: People fail to grasp their own incompetence, precisely because they are so incompetent. And since overcoming their incompetence would first require the ability to distinguish competence from incompetence, people get stuck in a vicious cycle.

"The skills needed to produce logically sound arguments, for instance, are the same skills that are necessary to recognize when a logically sound argument has been made. Thus, if people lack the skills to produce correct answers, they are also cursed with an inability to know when their answers, or anyone else's, are right or wrong. They cannot recognize their responses as mistaken, or other people's responses as superior to their own."

How to Escape the Vicious Cycle?

More important maybe than recognizing the intrinsic dilemma in the Dunning Kruger effect, is the issue of how people may overcome this incompetence driven vicious cycle. After all, it might be flattering to think of you and I as so smart that we are never the incompetents in the Dunning-Kruger story, but alas, this is the very problem of it all: Our own judgment of competence isn't worth much, when it is possible that we are merely too incompetent to provide good judgment.

One possible way to improve people's ability to discriminate between poor and great performances on a particular task domain, is of course to teach them additional discrimination skills. And indeed, one of Dunning and Kruger's experiments aims to find out how well this method works. As in the previous experiments, students first participated in a logic test and then rated their perceived performance. (Again, poor performers grossly overestimated their performance on the test, and high performers erred towards the side of modesty.) This time, in a second phase of the experiment, half of the participants were now given a mini-lecture on how to solve the type of logic questions they had just seen on their test. This was an attempt to provide them with systematic tools for distinguishing accurate from inaccurate answers.

"When given their original test to look over, the participants who received the lecture, and particularly those who were poor performers, provided much more accurate self-ratings than they had originally. They judged their performance quite harshly- and even lowered their confidence in their own general logical reasoning ability, even though, if anything, the mini-lecture had strengthened that ability, not weakened it."

In light of the above the result, one might view the Dunning-Kruger effect a little less as a vicious cycle, as it can be broken fairly easily by external communication of meta-cognitive skills. Such communication seems to significantly improve people's self-assessment ability and thus lay the groundwork for self-improvement.

What About the Cream of The Crop?

So far, I have mostly mentioned the poor performers in the Dunning Kruger effect, but there are of course also the top performers, who tend to underestimate their performance? For them, the cognitive bias in the Dunning Kruger studies is of a different kind than the one described for the poor performers:

"Top performers tend to have a relatively good sense of how well they perform in absolute terms, such as their raw score on a test. Where they err is in their estimates of other people-consistently overestimating how well other people are doing on the same test".

Not surprisingly, an easy way of providing top students in the Dunning Kruger study a perspective on the true exceptionality of their performance was to simply show them some samples of other people's answers. Given these students general ability to discern a good from a poor performance, such comparison opportunities were sufficient for high performing students to revise their self-assessments and rate temselves more accurately.

Quite notably, providing poor students with sample answers of their better performing peers did nothing to improve their relative self-assessment. In line with Dunning and Kruger's hypothesis, poor performing students seemed to lack the ability to identify other student's answers as superior to their own, and therefore were unable to use this information as a benchmark for re-evaluating their own relative performance.

Self-View Not IQ

To be clear, the main reason for the Dunning Kruger effect should not be viewed as lying in a person's general IQ. Much rather the Dunning Kruger effect seems to arise from the general top-down approach in which people estimate their own performances: In evaluating ourselves, we tend to start with preconceived notions about our general skill and then we integrate these notions into how well we think we are doing on a task.
A great illustration of this can be found in another experiment by - again - Dunning and Kruger. These guys are prolific: In this experiment students were first asked to report their views about their own "abstract reasoning ability" as well as their perceived knowledge and skill regarding the use of computers.

After assessment of these self-related views, participants were administered a test which was claimed to be either a logic quiz or a computer quiz, but was in both cases actually the exact same quiz.

In general people were far more confident about their logical reasoning ability than they were about their computer skills, and quite tellingly, when it came to rating their performance on the subsequent test, overestimation in poor performers - the Dunning Kruger effect - was much stronger for those participants who believed they were taking a logic quiz, than for those who believed they were taking a computer quiz.

To a substantial degree, it was participants preconceived notions that seemed to drive the observed overestimation.

Why it Matters

As Dunning and Kruger point out

"The top-down nature of performance estimates can have important behavioral consequences. [...]Starting inadolescence, women tend to rate themselves as less scientifically talented than men rate themselves. Because of this, women might start to think they are doing less well on specific scientific tasks than men tend to think, even when there is no gender difference in performance. Thinking they are doing less well, women might become less enthusiastic about participating in scientific activities".

In a final clever experiment, Dunning and Kruger tested this hypothesis regarding women's motivation to participate in scientific endeavors as follows: Male and female students were given a pop-quiz on scientific reasoning, and results were compared to previously obtained self-views on scientific skills. Although there was no gender difference in actual performance, women both reported lower self-views regarding their skill, as well as lower self-ratings regarding their performance on the pop-quiz.
Quite illustratively, later, when asked if they would like to participate in a

"science competition for fun and prizes , the women were more likely than the men to decline the invitation".

Statistical analysis revealed this reluctance to be correlated strongly with the women's perceived performance on the test.

"Perception of performance, not reality, influenced decisions about future activities".

[Click Link Below for Full Article]

Daniel R. Hawes, Rachael Grazioplene, and Kimberlee D'Ardenne, Ph.D |Jun 6, 2010
Tags: TypeZen, Type A, Type A Personality, Henry Kang, Daniel R. Hawes, Rachael Grazioplene, and Kimberlee D'Ardenne, Psychology Today
comments powered by Disqus