What the Dunning-Kruger effect is and isn’t (2010)
talyarkoni.org
talyarkoni.org
- 3 months ago 68 comments [https://news.ycombinator.com/item?id=25551624]
- Jan 2020 [https://news.ycombinator.com/item?id=21984435]
- May 2018 [https://news.ycombinator.com/item?id=17069386]
- Aug 2015, 25 comments [https://news.ycombinator.com/item?id=10145480]
- Jul 2010, 23 comments [https://news.ycombinator.com/item?id=1498136]
- Jul 2010 [https://news.ycombinator.com/item?id=1494732]
- Often referenced in other high-repost DK posts like "Dunning-Kruger and other memes" and "Wikipedia: Dunning–Kruger Effect"
That's an interesting result, but I think it would be pretty surprising if your link was discussed since your link is from 2017 and the post is from 2010 and doesn't appear to have a recent update.
"Our results further confirm that experts are more proficient in self-assessing their abilities than novices and that women, in general, self-assess more accurately than men. The validity of interpretations of data depends strongly upon how carefully the researchers consider the numeracy that underlies graphical presentations and conclusions. Our results indicate that carefully measured self-assessments provide valid, measurable and valuable information about proficiency"
Because of ceiling and floor effects (it's hard to massively underrate yourself when you're already in the bottom decile of actual competency, and vice versa), non-experts have a mean self-assessment higher than their actual competency, but that doesn't mean that non-experts as a group tend to overrate themselves.
Doesn't it mean exactly that? It doesn't matter if it's caused by ceiling and floor effects, the end result is that the less competent tend to overrate themselves.
Random noise is a primary reason that we see regression to the mean in most contexts (the exception being actual deterministic factors, such as seasonality, the restoring force of a spring, etc.)
The chain of reasoning goes like:
Random noise (Nuhfer)
-> can cause regression to mean (covered in the article)
-> calls into question the Dunning-Kruger effect
This is a convoluted way of saying men may just be overconfident, but importantly not because we assume we k ow everything, but because we are more confident we can learn it even if we don’t know it.
Maybe me writing all this while self-assessing myself is yet more male self-assessment hubris...
>...we are more confident we can learn it even if we don’t know it.
I’m confident women can learn things they aren’t familiar with, and as quickly as a man can. Your whole comment reeks of bias, actually.
Do you mean "identical"?
I wonder what makes you think that men and women would function precisely the same? (Apart from physical things)
People can be different, but still be equal (as in equality)
Also true story: if I am solving a problem and ask dude whether he knows the solution, he says yes I do, takes away my keyboard, starts googling and then refuses to return keyboard back, it is serious wtf annoying.
Also, people react different to overconfidence from woman - in my repeated experience, if woman claims she can do something technical she then starts figuring it on the fly, she will be butt of jokes. I mean, I cant figure technical stuff out on the fly in presence of some people, cause they will seriously mess into it.
If I am asking for my keyboard back, I should not have to physicality touch or fight you to get it back. Regardless of what you think about me.
EDIT: In my head, I feel like you've given up on the idea that I can help you, and you're probably going to go find someone smarter and more competent than me. In this context, a guy who doesn't want to give the keyboard back is afraid of admitting failure.
If you don't know answer, the normal response is no. I don't want you to highjack my computer and completely prevent me from working. That is ridiculous. And it is not solution either - it is literally preventing me to solve it.
> However, men are hearing "Can you solve this problem for me". Giving you back the keyboard means admitting that I failed to fix it.
Then men should fcking learn how to parse human language. Because based on this, betweem them not being able to interpret direct speech and then supposedly being unable to interpret hints, there is not much space. How exactly am I supposed to communicate? And none of this is believable, because above interaction rarely happens against other men.
The whole "he is trying to fix it while women just want to talk" is insulting in situation when he is preventing me to solve thing, when he is not solving anything and I was not chatty at all.
It was long ago so I don't really remember if the phrasing was "do you know..." or not, but yeah now I agree that just give people direct answers and maybe an open offer like "do you need another set of eyes?" or just leave them alone.
If they asked about cooking or cleaning, you'd probably get the opposite reaction.
Of course men aren't inherently better at math nor are women better at cleaning. It's just that we socialize them to think they are.
That's one of the central dogmatic feminist talking points that are just spouted off uncritically, along with other things men are supposedly socialised for, like wage negotiation and preferences for violence, taking up space and technical subject matter. Included is the starting position that women and men are the same and if women are perceived to be behind, it's because something was done to them irrespective of their own agency and accountability as individuals and adults. It's a conspiracy theory, and an infantilising one at that.
In social sciences, this becomes social and/or (modern) historical determinism, leaning heavily towards assuming everything's a construct and caused by historical precedent, ignoring biology (other animals, hormones, sexual attraction, evolution), long-term history and global invariability of gender roles. Sure, that's a feel good narrative, but is it a rational stance supported by evidence?
> If they asked about cooking or cleaning, you'd probably get the opposite reaction.
If this is about socialisation, the simpler explanation is that men are socialised to project confidence, because that's what romantically/sexually rewarded. The link towards granular, per-topic confidence is more tenuous and requires you accept a lot of feminist baggage at face value or provide an alternate explanation.
> Of course men aren't inherently better at math nor are women better at cleaning.
No, but perhaps they are more interested in maths by virtue of the topic itself, or due to ancillary effects like future status and wages, like women are more interested in nesting than men. Similarly, men are generally more drawn to the concepts of service and duty than women are, which is why vastly more men work logistics, manufacturing, law enforcement, fire departments and the military.
Personally, I find feminist axioms to be a very poor predictor of reality because they fail to account for mating strategies, personal choice, responsibility and preference - which is ironic for a movement that was initially about women's choices in the reproductive sphere.
Identifying those factors and working to minimize or eliminate their affect is a noble goal. But there are some factors that can't be eliminated, and we wouldn't want to live in a world where they were.
Feminists are right that there is a problem, this is a demonstrably unfair society. It doesn't matter what the reason is as much as it matters what the solution is. People spend a huge amount of time arguing about the problem, who's fault is it, what factors come into play.
I don't want to live in a world where women are men. Where they can't take time to start a family and properly care for their children, because keeping the money flowing is a more immediate need.
I want to live in a world where you can't just take time off to raise a child, but also to do art, or to travel, or to simply be a human who exists outside their office.
TLDR: Why matters very little, the income gap exists and we should fix it.
When people tout the "30% less" statistic, it often refers to lifelong earnings, not hourly wage, in a study that didn't control for experience, sector, location or education.
It's 18% less when just comparing median wages. When controlled for the same job and qualifications, women earn 98 cents on the dollar (0). Women (as a group) work more part-time and in less profitable sectors like education, NGOs, government and nursing. They are less interested in high-profile and/or high status roles. Again, it feels good to say that women earn less due to discrimination and less so when it's because of their own choices.
You could argue (and I would agree) that some professions should be paid more. I think Covid has shown all of us which professions provide actual value, amongst whom definitely nursing and teaching.
> Women have all the same needs that men have
Women have the same basic needs, but there are differences in preferences, needs unique to women, needs unique to men and gender-specific dreams/wants that aren't needs.
> the necessity of being present at, key, moments of the child rearing process
Sure, but there is also a greater preference towards staying at home and having the man provide. It's simplistic to say "women are being forced" or "men aren't taking up their fair share" - this is a multivariate analysis. Specifically when it comes to a family, two thirds of the divorces are initiated by women. So not only are women staying home more often, they are choosing not to have a man be there at all. That might be for legitimate reasons, or it might not - but a divorce will impact your life balance.
> Feminists are right that there is a problem, this is a demonstrably unfair society
The problem, according to modern feminism, is different outcomes. As I see it, that's not a problem as long as opportunity is equal. Where women earn less on average per annum, they also have almost none of the workplace related deaths, lower suicide rates, homelessness, depression, incidence of burnouts, etc. In the younger generation, women outperform men both in wages and education. To me, it's not so clear if there is a better deal, and if so who has it.
> It doesn't matter what the reason is
I couldn't disagree more. If the claim is the problem lies with men or "the patriarchy", then the onus is on the claimant to prove that position, starting with a solid definition and falsifiable demonstration of patriarchy. That said, if you don't understand the reason for your outcome, you're powerless to change it. In a very concrete sense, you won't know what policy to implement when you topple the status quo and get to power. Among third wave feminists, I'm not hearing about how women architect their own fates and the importance of choices. I do among second wave feminists.
Would you say "it doesn't matter what the reason is" if all your relationships are short-lived, if you keep getting fired or if you keep failing your driving exam?
> this is a demonstrably unfair society
Yes, because of unequal access to money and genetics. Therefore there are class issues first and foremost, some sexism, some racism and the other forms of discrimination. To claim all (or even most) of women's problems are due to sexism, systemic or not, is a reductio ad absurdum.
> I don't want to live in a world where women are men. Where they can't take time to start a family and properly care for their children, because keeping the money flowing is a more immediate need.
Neither do I. But if that's what you want, you can't be opposed to earning less, either. Those are the consequences of your choices. Child rearing is unpaid, unless you're nannying as a service.
> I want to live in a world where you can't just take time off to raise a child, but also to do art, or to travel, or to simply be a human who exists outside their office.
Going out on a limb, I'm going to assume you mean where you can take time off. I'm fully with you there. That comes with a trade-off: you're going to be less "successful" in the conventional, square, monetary sense. You're sacrificing compensation for added fulfillment. That might entail any of: being less resistant to economic downturns, a smaller house, less of a pension, no or fewer kids, less social credit, fewer available/compatible dating partners, more limited career chocies, etc. It might also mean more laughing wrinkles, good moments, average happiness, life expectancy, better relationships, and so on.
That's why I said modern feminism is infantilising: it refuses to acknowledge that life is a struggle, filled with compromise and sacrifice. It's the Disney princess that won't grow up and the college kid that won't stop shouting from the barricades long enough to get on with writing their final dissertation. You have to pick and choose, mostly because no one owes you a damn thing. That is, until we've achieved fully automated luxury space communism à la Roddenberry. We don't live in a post-scarcity world yet by any stretch of the imagination.
> as much as it matters what the solution is
TL;DR: If the cause/reason is unimportant and the solution is key, what do you think is/are some good next step(s)?
> For 2018 (the most recent data available), the dollar value of women’s unpaid work in the U.S. was equal to 86% of all the economic activity recorded in the state of New York. In other years—say, the late 1990s and late 2000s—the value of women’s unpaid work even surpassed New York state GDP. And keep in mind this value is at the low end of the possible range because we use the federal minimum wage and not, for example, higher state minimum wages let alone market wages that correspond to the specific work being done.
> The UNDP Women and Development Report of 1995 conducted a time-use study that analyzed the amount of time women and men spend on paid and unpaid household and community work in thirty-one countries across the world, including countries classified as 'industrial, 'developing' and 'transition economies.'[12] They found that in almost every country studied women worked longer hours than men but received fewer economic rewards. The study found that in both the 'developing' and 'industrialized world', men received the "lion's share of income and recognition" for their economic inputs, while women's work remained "unpaid, unrecognized, and undervalued."[12]
The fact that we don't pay as much for the things women tend to do is the problem. We've created a world where if you choose to spend your life sitting in a cubicle, you can support yourself. But if you spend your life caring for the people around you, you cannot, your labor still has value, it's simply not compensated. This is a really bad incentive scheme. We want parents to spend time with their children, not just because children with present parents perform better, but because of course we do. We want smart, capable people to become social workers and teachers and pediatricians without sabotaging their finances. To put another way, the world would be made worse, if 20% of the people who are currently working in their homes, decided to become software engineers instead. The world would become better if 20% of software engineers decided they'd rather contribute to their homes and communities.
https://fredblog.stlouisfed.org/2020/03/calculating-the-valu... https://en.wikipedia.org/wiki/Unpaid_work#Gender_and_unpaid_...
> But if you spend your life caring for the people around you, you cannot, your labor still has value, it's simply not compensated. This is a really bad incentive scheme.
That's true, but that goes for any type of activity in the trade economy, whether you're raising kids, doing the shopping for someone, building their shed, teaching them to drive, troubleshooting their devices, fixing their flat tire, etc. That incentive scheme is evidence of a mercantilist attitude in society: social capital is just not valued in monetary terms.
Calling it sexism is reductive, because there are a lot of things men do for free, too. That's also my problem with your first link, which didn't examine men's unpaid labour at all as far as I could tell. I'm unable to tell what the picture looks like on balance.
> We want smart, capable people to become social workers and teachers and pediatricians without sabotaging their finances.
Public sector jobs will always be subject to government budget whims, "cost saving" initiatives and the like. What's needed there isn't feminism but collective bargaining and lobbying. It's also worth remembering that the rest of the West isn't like the US: in Europe, teachers can make median income or above.
As to pediatricians: I see no evidence of them being paid badly across the West. In fact, here in Belgium gynaecologists make more than ER doctors, and pediatricians make more than neurologists and oncologists, but less than ER doctors.
Sadly, the recurring theme everywhere seems to be nursing, which is just undervalued, micro-managed and thankless in general. Nurses can generally get somewhat better wages when they get some extra certificates. Of all examples I know, this one is the most suspiciously low across the board. Then again, I don't think it would be considered less menial or low-status if men did most of it.
> We want parents to spend time with their children, not just because children with present parents perform better, but because of course we do.
When a woman stays at home taking care of the kids, does she not enjoy the same lifestyle as the man? How is that not being compensated? When they get divorced, is she not entitled to half the money and how is that not getting paid? Is she not entitled to child support? Women aren't being disadvantaged simply because they don't get a payslip every month. Now that definitely isn't true everywhere else yet, so globally there's an argument for change towards our current status quo. That also seems to be what your second link is referring to: there is more yet to be done in the Middle East and beyond.
If you think parenting for free (as a couple) is unfair, what would be fairer? Parents already get tax cuts and benefits by virtue of having kids, so there's already a wealth transfer going on from childless people to parents. What would you do instead/additionally? In concrete terms: who should be paying what? I'm not asking for an exact solution, but would like to know what principles or policy levers you're thinking of.
> To put another way, the world would be made worse, if 20% of the people who are currently working in their homes, decided to become software engineers instead.
I agree on that too, but that is also not evidence of sexism but of a machinistic, homo economicus kind of collective philosophy. It's admirable to dream of and take action in order to create a better world, but I'm not sure how we'd get there, in terms of re-evaluating what's fundamentally important. Perhaps restructuring money creation itself (pivoting from loan-based to UBI based) would get us there, but I don't think that movement is anywhere near critical mass.
> The world would become better if 20% of software engineers decided they'd rather contribute to their homes and communities.
The assumption being that they don't, right? There's more to it than raising kids. Let's also please not pretend like men do none of that, or that being a stay at home parent is a full-time job at all kids' ages. I mean: do professional women not contribute to their homes and communities?
If your objective performance falls far outside the norm, even if you know it and are confident this is true, people may not accept that fact and trying to insist on getting credit for it won't get you credit. It will get push-back. It will get denial. It may even result in harassment.
If your objective performance is poor, you may be subconsciously hoping to influence the outcome positively. Most forms of assessment aren't entirely objective. To whatever degree you can protect yourself from negative social consequences by suggesting to other people that your performance was within some acceptable and expected norm, it is in your best interest to do so.
People who habitually do this kind of thing at work or in other socially meaningful settings may be unable to turn that habit off for purposes of a psychological study. They may not even be consciously aware of it. It's just a thing they do without really planning it. It's what they have always done.
For easy tasks, it may have the most impact to fudge a little and say "My performance is sort of normal-ish. Nothing to see here. Move along." whether you perform abnormally well or abysmally. Those are tasks most likely to not matter so much, not have high stakes, be easily done by just about anyone and so on.
In other words, if you suck at taking out the trash, probably anyone in the office can gather up the trash and take it out. It's not likely to be a firing offense. They will just have someone else do it who doesn't drop everything everywhere all the time and never bother you about your failure to be good at it.
But if they hired you for your PhD in mathematics, your understanding of complex math actually matters. You aren't readily replaceable. Trying to fudge about how well you understand it and claim greater ability than you have is something people will take great offense at. Being a little modest about it can help you keep your job while reducing expectations.
There can be a lot of benefits to managing the expectations of other people who know you can do something that's hard but don't understand the conditions under which you can pull that off, how much prep time you need, etc. It is actively problematic for everyone to think you can do (amazing thing) anytime and all the time with no lead time and no recovery time. That's a great recipe for burn out and other disasters.
Haha, I don't know why, but as soon as I read that sentence, I thought "Is this Doreen Michele?" and checked your username and it was! Something unique about your writing.
I used to have a lot of these kinds of concerns about how I'm perceived and in hindsight it held me back and I regret things I didn't do because of it. Now I have achieved things that are clearly respectable and I don't feel any need to prove myself. Instead, I'm happy to look like a fool. I also have modest enough ambitions that I don't need to worry about the consequences of people's perceptions for things like promotions or other opportunities. I create my own opportunities.
> People may self-report opinions of their own performance that they feel are socially acceptable or socially protective of themselves. So it reverts to the mean because it remains within The Overton Window so to speak.
> If your objective performance is poor [...]
The DK paper didn’t study objective performance, it studied perceived performance. They didn’t exactly have participants opine on their own performance at all, they were asked to rank themselves within the group. That is a relative opinion, and it means the participant is evaluating the other participants’ abilities too. It also means everyone in the study was just guessing, which is the biggest reason why I think Tal is right, that this is nothing more than noise leading to regression to the mean.
> But if they hired you for your PhD in mathematics [...]
DK did not study any difficult cognitive tasks akin to research or complex math either. The paper only had 4 tasks: ability to get a joke, some standardized test logic questions, a little English grammar, and grading other people’s tests.
I’m not sure if the things I’m pointing out support your theory or not. Mostly I suspect DK is specious and contagious because it lets people justify their own negative judgements of others, so in that sense I think using DK as an explanation of someone’s behavior is a way to feel socially acceptable and be socially protective of one’s self. It’s just unfortunate that the common judgements and conclusions with this study are almost always unjustified because the paper is so widely misunderstood.
It's not a theory. It's a comment on a public forum, not a PhD thesis.
I've had some pertinent classes, like Intro to Psychology, Social Psychology, and Negotiation and Conflict Management. In my youth, I spent about 3.5 years in therapy to sort out baggage from childhood trauma. I've journaled and blogged for a lot of years, so introspection and trying to figure out what makes me tick is a longstanding habit.
I raised two special-needs sons who don't intuitively understand social stuff, so trying to explain things to them really helped hone my understanding of some things. I had to up my game by quite a lot to help them get a meaningful and useful understanding of social phenomenon.
Social phenomenon are inherently hard to find good information about. Studies tend to be really bad. People who know they are participating in a study almost by definition don't behave like the do "normally."
I continue to talk with my sons about social phenomenon because it remains an ongoing area of interest. My sons and I talk fairly often about what constitutes a good source of social insight and what doesn't. I comment about it sometimes on HN because I hang here and sometimes people find my comments of value.
It's my opinion. My opinions on such topics seem to generally be on more solid footing than average. They usually aren't, per se, backed up by some kind of scientific study, though they are generally informed by gathering solid tidbits of useful info over the years.
If you don't agree with it, it's totes fine.
I think I’m agreeing with you, and I think it’s an interesting explanation and a valuable comment. I have no doubt that you’re right that people are socially protective, and some people pull back publicly to fit in. It’s just that as a response to DK or to Tal, I feel like it might be mostly orthogonal and independent. I’m pretty sure that the DK study itself didn’t experience or measure this kind of social self-protection, because of the methods described in the paper.
What I disagree with is the DK paper. They make broad generalizations and speculations about behavior throughout that their own data do not support. And the paper seems to be almost completely misinterpreted. Despite the data and the statements in the paper, everyone seems to think that the paper shows confidence being an indicator of lack of skill. It doesn’t help that the paper’s title misleadingly walks you right into that idea.
> Studies tend to be really bad.
You can say that again.
https://podcasts.apple.com/us/podcast/david-dunning-on-the-d...
https://podcasts.google.com/feed/aHR0cDovL2ZlZWRzLmJsb29tYmV...
The results of Dunning-Kruger-adjacent experiments always present their results as measuring "How people think they perform" against "How they actually perform", when they seem to actually be measuring "How people tell other people they perform" against "How they actually perform".
Which seems to me like it's a slightly different thing? It seems like the pop-sci "those people are too stupid to know how stupid they are" interpretation isn't the only plausible interpretation, anyway.
Let's say a 50% score at a test puts you in the 1. percentile and 70% in the 99th. I would actually expect people to estimate their relative position roughly in line with their results.
@dang?
How meta.
https://timharford.com/2021/03/cautionary-tales-the-dunning-...
I've never seen someone cast themselves as the one that might be overestimating their ability.
It’s really too bad Dunning & Kruger’s data didn’t show anyone being too dumb to know it, nor did it even show anyone being dumb at all. So maybe the meaning isn’t as precise as you think? All the participant scores were relative to their peer group (no way to know if they were dumb or smart), and all the participants were Ivy League college kids. That doesn’t mean there were no dummies in the group, but it is certainly not a random sample of the population.
> I’ve never seen someone cast themselves as the one that might be overestimating their ability.
The DK paper shows a positive correlation between confidence and skill. While I have seen people suggest they might be overestimating (in social situations where doing that might add credibility), if you believe in the DK effect and trust their data, then saying out loud that you’re overestimating your ability is the same as saying you think you might be dumber than you are, right?
Why bother going to the effort trying to correct someone without bothering to take 30 seconds to even read their comment?
If that makes sense.
The key is to be aware that there are things you don't yet know, and not let your ignorance get the best of you.
There’s a very important difference there, because self-ranking requires knowledge of other people’s skills.
Do note that the DK paper found a positive correlation between confidence and skill. Most people here and elsewhere seem to believe the DK effect shows a negative correlation.
Also note that the skills people were self-ranking were all easy. Ability to get a joke was one of the 4 tasks. How would you rank your own ability to get a joke against those around you? The question barely makes sense.
I think it’s a big mistake to take any lesson from the DK paper other than sometimes people see what they want to.
For confidence, the positive correlation doesn't help with the problem where someone with low end confidence estimates similarly to someone with higher confidence but is much more wrong.
You are right, they did mention self-evaluation of score. "Lest one think these results reflect erroneous peer assessment rather then erroneous self-assessment, participants in the bottom quartile also overestimated the number of test items they had gotten right by nearly 50%."
But the results and conclusions of all the studies are focusing on the rankings and not the scores. It's difficult to even understand what the absolute range of scores is, and of course we don't see the actual questions they asked, so we don't know what the scores mean.
Figures 2 and 3 say to me that zero people in the study knew how well they ranked or what their score was, the survey data is basically flat with noise. This could be explained by questions that are hard to understand, or by participants guessing at the answers. They didn't propose those explanations though, they wax on and on about how people who underperform overestimate their abilities.
How do we know the participants even tried to answer correctly? They were all undergrads getting extra credit (which suggests they needed extra credit and were within the range of grades where it might help to get extra credit... consider what that means relative to skill: it's a narrow range of students not getting an F nor getting an A.) And they were getting extra credit just for being there, not for answering correctly. What if they were lazy? What if they were told their answers didn't matter? Why should we assume they didn't answer randomly? Even if they did try, the participants were all Ivy League undergrads who have been told they're among the elite and made it through a highly selective process of admission. Of course they rate & rank themselves highly. 100% of the participants were in a single college psych class, there were no uneducated people, no morons, and nobody who actually had meaningful "deficits in metacognitive skill" relative to the general population. Why is the set of people used in this study not considered an invalidating confounding factor for the results?
But yeah, since it's a broadly applicable principle, it should apply to itself. In that people who don't understand the Dunning-Kruger principle believe they understand it better than they do.
Not the greatest for mental health, though. Now I am glad that I am working with very competent but also nice and supportive colleagues. It really is a blast to be able to have some bad days without feeling you should quit and buy a farm in the Larzac instead.
Seems like this and bloom filters. Constantly on HN. Both mostly useless
The Voight-Kampff effect.
https://www.youtube.com/watch?v=vRJ5cCP0ZPE&ab_channel=Movie...
The more you drive, the less intelligent you are.
* Dunning-Kruger
* That xkcd 10,000 People comic[0] (Getting meta in my own comment!)
* "HR is not your friend"
* "Corporations only goal is to increase share price"
* "You're the product"
I'm sure I could name another dozen or so if I had the inclination.
At the risk of replecating myself and getting a crisis, everybody with a brain like mine could have seen that coming.
Sigh...
:-)
I don't believe that the Dunning-Kruger effect is very significant these days because our modern society is full of:
1. Extremely talented, hard working people who have achieved nothing because they never got very lucky.
2. Lazy idiots who have achieved a lot because they got extremely lucky.
In contradiction with the Dunning-Kruger effect, people in group #1 are less likely to feel impostor syndrome because they're been working hard for a long time and they've made an impression and received excellent, honest feedback wherever they've worked over many years and delivered excellent results by all accounts. Their past employers are praising them, ex-colleagues want to work with them, etc... this gives them confidence in their ability in spite of a lack of financial success. In our modern age, skill does not guarantee financial success because skill has limited utility value in our dysfunctional economy built on unsound money printing and asymmetric playing fields... Everyone wants to work with a genius, but genius has limited financial value. Reckless foolishness can often yield higher returns if the game is played enough times.
Also in contradiction with the Dunning-Kruger effect, people in group #2 are bound to feel like impostors because they are impostors. All their confidence is derived from their financial success. They've never delivered any lasting, meaningful results which can be attributed to their work... and they've never received any praise from any intelligent person until they became rich (then suddenly, the smart bastards all line up to give praise)... The incentives in this case aren't conducive to honest feedback. All the ingredients to feeling like an impostor are there and even with their limited intellect and underdeveloped primal ego; deep down, they can sense that something is not quite right.
Really most of what people think is D-K is just statistics analogous to reversion to the mean: people assume they are typical, so less-skilled people naturally guess they are better, and higher-skilled guess they are less.
If D-K means anything, it has to be a variance on top of this effect. Such an effect does seem to appear, where the very least skilled imagine there is no deep skill to learn.
If it were just that, people would rate themselves closer to the median than they actually are, but D-K found the people rated themselves closer to the 70th percentile than they are.