More Recent Comments

Wednesday, March 07, 2018

Can the Dunning-Kruger effect be reversed?

The Dunning-Kruger Effect was first proposed in a classic 1999 paper (Kruger and Dunning, 1999).1 People suffering from this effect show one of two characteristics. If they are not knowledgeable about a subject they tend to overestimate their ability. If they are experts in a subject they tend to underestimate their ability (see figure).

The phenomenon is more significant in people who overestimate their ability because it includes a large number of people who are making decisions on subjects that they know little about. Because of the Dunning-Kruger effect, they are confident that their decisions are based on facts and evidence. That's bad enough, but there's another aspect to this problem—why do these people seem to be incapable of recognizing that they are suffering from the Dunning-Kruger effect? Here's how Kruger and Dunning explain this ...
We argue that when people are incompetent in the strategies they adopt to achieve success and satisfaction, they suffer a dual burden: Not only do they reach erroneous conclusions and make unfortunate choices, but their incompetence robs them of the ability to realize it. ... they are left with the mistaken impression that they are doing just fine. ... as Charles Darwin (1871) sagely noted over a century ago, "ignorance more frequently begets confidence than does knowledge."
We all know examples of people who are overly confident and we are familiar with the other end of the spectrum; namely, experts who worry about how little they know. We, ourselves, are very likely guilty at both ends of the Dunning-Kruger effect.

We would all like to believe that we can correct our ignorance with facts and evidence but if we are truly incompetent then we might be resistant to evidence. Here's what Kruger and Dunning have to say about that,
One puzzling aspect is how the incompetent fail, through life experience, to learn that they are unskilled. This is not a new puzzle. Sullivan, in 1953, marveled at "the failure of learning which has left their capacity for fantastic, self-centered, delusions so utterly unaffected by a life-long history of educative events." With that observation in mind, it is striking that our student participants overestimated their standing on academically oriented tests as familiar to them as grammar and logical reasoning. Although our analysis suggests that incompetent individuals are unable to spot their poor performances themselves, one would have thought negative feedback would have been inevitable at some point in their academic career. So why have they not learned?
The authors propose several reasonable explanations.
  1. Most people don't get negative feedback because their friends are reluctant to point out how incompetent they are.
  2. In some situations here's no opportunity to recognize that you are incompetent.
  3. Even if people receive negative feedback, they may not accept it. This is because they usually blame someone, or something, else for their failure.
  4. Incompetent individuals may be unable to benefit from constructive criticism because they lack the ability to change. The very flaws that cause them to be incompetent in the first place are what prevent them from improving their reasoning skills and knowledge.
That last point is important. It's what the authors were referring to in the first quotation.

There's one aspect of the problem that isn't covered in the work on the Dunning-Kruger effect. That's the problem faced by experts when they fail to make an impression on the unknowledgeable. It must be very frustrating to try and teach people who suffer from unjustified overconfidence in their abilities. For example, imagine that you are an expert on international trade, the global economy, and macroeconomics. You are hired by a wealthy businessman to teach him about these subjects but you find it impossible to get anywhere because the businessman is incompetent but, at the same time, extremely confident that he knows everything there is to know about these subjects. He suffers from the Dunning-Kruger effect and "fantastic, self-centered, delusions." Our hypothetical student has resisted all previous attempts to educate him for all four of the reasons listed above.

If I were the teacher in such a situation, I'd quit my job. According to Kruger and Dunning, I'd be wasting my time.

1. The concept wasn't new. The contribution of Kruger and Dunning was to document the effect and explain it in a concise, easy-to-understand, manner.

Kruger, J., and Dunning, D. (1999) Unskilled and unaware of it: how difficulties in recognizing one's own incompetence lead to inflated self-assessments. Journal of personality and social psychology, 77:1121. [doi: 10.1037/0022-3514.77.6.1121]


Robert Byers said...

I agree with this whole concept. i think about it in origin issues and other issues of mankind.
Yet i don't think its complicated.
i think its simply a issue of intelligence.
Its a curve in the class and then comparing classes.
Error or accuracy in contentions is not a roll of the dice for individual.
its about identities or groups or any thing that influences people.
Recently we saw in the winter olympics Norway, a tiny populated country, beat everyone in medals. beating high winter nations and all nations.
This shows a curve of intelligence in winter sports.

In North America one always finds social/political/religious/, and even oRIGIN subects to follow closely demographics.
i know evolutionists throw at us about education levels and acceptance of evolution etc etc.
there is a point here.
WE throw at them that the common middle class people are more likely to question things and think for themselves as opposed to upper classes/upper middle classes who accept anything they are told as if to justify their own status.

Yes I think the more intelligent are less likely to be so confident about things they know little about or things they know about. Because humans still know little about anything.
Less intelligent people are more sure complicated things are easy to understand and less able to see errors and errors in their own thinking.

if its a curve on human intelligence it could only be as this Dunning- Krueger chart indicates.

Its like adults and children. The kids are more likely to get things wrong and fail to correct them even after instruction. The kids are dumber and so consistent.
Its a curve.

Larry Moran said...

I note that you seem to know very little about evolution. I’m amazed at how confident you are when you lecture the experts on all the flaws in their field of expertise.

steve oberski said...

Robert Byers and Joe G, the Dunning-Kruger poster boys.

CrocodileChuck said...

Good post.

Too bad you blew it at the end, invoking 'macroeconomics': the most 'BS-ie' of BS social sciences.

dean said...

"WE throw at them that the common middle class people are more likely to question things and think for themselves as opposed to upper classes/upper middle classes who accept anything they are told as if to justify their own status."

An amazing assertion with zero evidence to support it. Par for the course for you.

DGA said...

I am not an expert, but I infer that this post is a reference to Trump and Cohn.

Robert Byers said...

I know enough and quite a lot about the main points of evoltionary biology. i learn things , great and small, like learning from this blog about random drift being a prioriity/or important in evolution as opposed to histircial conclusions on natural selection alone working on mutations. That would not be in the public domaine and so i know more then the average educated public.
I also know YEC/ID ideass which brings in more awareness of evolutionary ideas.

I deny I lecture experts. Nobody wants that tag.
Whats an expert? where is the experts?
a expert is just someone with knowledge in their subject and a complete knowledge.
anyone can master the conclusions from experts as they write them down.
so anyone can think about them and take them on.
i address only certain conclusions and yES am intellectually confident .
No excuse for thinking i know my tuff and can correct error.
Your thread was about people in thev weong more likely to be slow to correct their error or recognize it.
one could also be in the right and more likely know they are in the right after dealing with contentions for a long time.
its a curve on intelligence.
Its more then knowing the subject. everyone has this issue.
I always watch, Mr Moran, carefully for how evolutionists make their cases.
I find flaws and find them not corrected or addressed!
yes i think creationists are more accurate as we deal with opponents a great deal more in society. its harder to be a creationist unless we are confident we are right.
Its easy to be a evolutionist and miss important flaws.
its not about experts but curves in the population as the Dunning/Krueger curve suggests.

Robert Byers said...

Whoops my computer was lying to me about my posts.
The last one is the one i wish to be quoted on for history!

colnago80 said...

It would appear that Cohn has taken Prof. Moran's advice and voted with his feet.

colnago80 said...

Re Larry Moran:

With regard to booby Byers, rarely have I heard anyone speak knowledgeably from such a vast fund of ignorance.

Joe Felsenstein said...

It is interesting to look at this from the point of view of statistics. No one will of course be perfect in estimating their abilities, so we expect some departure from a perfect correlation. That would lead to people who were actually good at task X underestimating their ability, and people who were bad at it overestimating their ability. That would simply be the consequence of none of us being perfect at estimating.

Is the figure above just showing that? Or is there a further bias causing incompetent people to be make exaggeratedly high estimates of the their own ability? and causing people who are very competent to make exaggeratedly low estimates of their ability?

I can't tell from the figure. I have heard that 75% of drivers think that they are better than average drivers. If it were just the result of inaccurate estimation then that should be close to 50% instead, as a result of the good drivers being too modest about their abilities.

Alex SL said...

As far as I understand, that is one of the main points of D-K. People don't just randomly over- or underestimate themselves, but most people assume they are slightly above average. This leads the best to slightly underestimate their own competence and the worst to vastly overestimate theirs.

(The second main point, which although addressed by this post unfortunately rarely gets mentioned by most of those who discuss D-K, is the systematic difference in the ability to reassess one's own competence relative to the average competence level when presented with evidence of everybody else's competence. The best see how bad everybody else is and now realise that they are better than they thought, but the worst lack the ability to recognise that other people's performance is better than theirs, and thus their self-assessment does not change.)

Robert Byers said...

How did a new point of view change the stats here?
The stats showed the experts questioned themselves and the unexperts did not at all.
it would be that way because a expert would know things are complicated as a result of his expertness.
Athe other folks would not see things as complicated and then think they got it pretty good.

It could only be this way. otherwise non experts would understand how a subject nuances make it more complicated then a quick read.
Its the non experts here who are in the wrong. the experts are probably right about themselves.
Unlikely they are wrong about their own subject. by the way where asre these humble experts??
Its a issue of intelligence and so a curve relative to that based on the persons knowledge of any subject.

Joe Felsenstein said...

@Alex SL: Yes, I think that the D-K effect is not in the slope of the line. The slope would be less than 1 if people simple were inaccurate in assessing their ability. It is in the shift of that line upwards, so that it does not pass through the means of the X and Y variables.

ealloc said...

I'm not sure that fits, since it implies Gary Cohn is an "expert on international trade, the global economy, and macroeconomics". Most articles about him emphasize how he's the opposite of a scholar: He's the first Fed chair in 30 years without a PhD, has no patience for "eggheads", and barely ever reads anything. His "aggressive personality" is typically cited as his major qualification.

I can believe he's an expert in markets, investments and financial products though. He was pretty smart to bet against his clients in the subprime mortgage crisis.

ealloc said...

Apologies, he's economic advisor not fed chair.

Dale Hoyt said...

Another personality trait that might interact with D-K is the locus of control - external or internal. Students with an external locus explain their poor performance on the unfair or trick questions on the test. Those with an internal locus explain their performance on inadequate knowledge or poor study habits. It's easy to generalize this to the larger population, I think.

LEH said...

Ehrlinger et al's extension of K-D has some interesting implications for how to teach. One of the things I've been contemplating is how K-D interacts with cognitive development. Young adults tend to be dualistic thinkers who are also looking to others as sources of information - what Dale Hoyt refers to as 'external locus'. If the developmental folks models reflect something akin to reality, then moving away from K-D mind-set is more than growing expertise, but also growing cognitively (Ehrlinger et al 2008, Why the unskilled are unaware: Further explorations of (absent) self-insight among the incompetent. Organizational Behavior and Human Decision Processes, 105(1), 98-121.)

Donald Forsdyke said...


I reported a similar curve to that shown here in a 1978 paper (1). There are three groups to consider. 1. Those so incapable that they cannot be cued towards a correct answer. 2. Those less capable, but able to be cued. 3. Those, so capable that cues play less of a role. Generally only groups 2 and 3 serve as subjects for this type of enquiry (e.g. college students). To test cueing we have a 1:5 multiple choice. Group 2 do better than expected here. On the other hand, the brilliant student (often smarter than the professor who set the examination) sees more subtleties in the question and does not do so well as expected.

The group 2 folk have found, over the years, that quite often they can wing-it in tests. Their hunches are more often right than wrong, so they are overconfident about their own ability. By the same token, the group 3 folk, knowing that they are being tested on their ability to tune in to an examiner's foibles, are less confident. This may explain why the curve I obtained matches that of D-K (2).

Whereas (2) has been cited 1435 times since it was published,(1) has been cited 5 times over a 40 year time span!

1. Forsdyke DR (1978) A comparison of short and multiple choice questions in the evaluation of students of biochemistry. Medical Education 12:351-356.

2. Dunning D, Kruger J (1999)Unskilled and unaware of it: how difficulties in recognizing one's own incompetence lead to inflated self-assessments Journal of Personality and Social Psychology 77:1121-1134.

Unknown said...
This comment has been removed by the author.
Larry Moran said...

There's nothing wrong in theory with using biological evolution as a possible explanation of modern human behavior. This could be a valid explanation as long as; (1) the behavior was known to have a significant genetic component, (2) the allele(s) arose in the relatively recent past, and (3) the allele(s) conferred a selective advantage that caused them to be fixed by natural selection (i.e. an adaptation). Alternatively, the alleles could have been deleterious and fixed by random genetic drift (nonadaptive).

The problem with the field is that it is not scientifically rigorous with respect to biological evolution. The work that has been published rarely addresses the questions that should be answered before jumping to an adaptive explanation of behavior. In general, publications are not skeptical and don't address alternative explanation that are either nonadaptive or have no known genetic component to be selected.

My main criticisms of the field is that it is nonscientific and hyperadaptive. I don't rule out other flaws but those are the ones I know of.

Unknown said...

What I find most interesting about the Dunning-Kruger effect is how people tend to think the effect describes people other than themselves. The effect describes all people. What we call ‘sophomoric’ is a subset of this behavior.
Of course one supposedly can’t recognize one’s own stupidity.
But I know this is true about me— I over estimate my knowledge in areas where I am ignorant and I under estimate my knowledge in areas where I am not.
I chalk it up to the fact I am human and I can’t know what I don’t know, so I am ignorant about how ignorant I am on most subjects.

Truth said...

Lol yes it can

Donald Forsdyke said...

An extended version of this comment has been posted at the PubPeer site.