Talk:Dunning–Kruger effect

Links

The references are out of control. There are paragraphs with sentences that have up to 5 references each. Each link is used numerous times and they are all out of order. Some of the paragraphs are more redundant than they are informative. This looks like way too much work for now, but over time maybe we can edit and streamline it. Thank you. Tvashtar2919 (talk) 23:56, 4 February 2022 (UTC)

Yes, that is the rubble and debris left from a previous squabble. Constant314 (talk) 00:09, 5 February 2022 (UTC)

I'm not sure why most of the references to the paper "The Dunning-Kruger effect is (mostly) a statistical artifact" by Gilles E. Gignaca and Marcin Zajenkowskib were removed. I restored them for now till we can settle the matter here.

That some of the references are numerically out of order is, I think, not a problem. Usually, the one most relevant to the specific claim should be the first one. After a short look, I only found 3 cases in the whole article where a sentence has 5 or more references and they are all in different paragraphs. One of these cases is a list, where all the sources are needed to cover all the items on the list. If the same reference is accidentally used twice for the same sentence, then one instance should be removed. Phlsph7 (talk) 07:13, 5 February 2022 (UTC)

Yes, there were many (20 or so) instances to the same reference by Gignaca. Many of them were in sections that had nothing to go with their statistical argument. I did leave the references in the "Criticism and alternatives" section where they were relevant. Constant314 (talk) 07:33, 5 February 2022 (UTC)

This paper also includes a general introduction to the Dunning-Kruger effect and talks about how it is measured and analyzed. For example, it contrasts subjective and objective abilities. The decision on where to cite it shouldn't be based on its title alone. We would have to check whether a specific claim in our article is actually backed by the contents found in the source or not. Phlsph7 (talk) 07:47, 5 February 2022 (UTC)

I just had a look at WP:OVERCITE: it says that more than three inline citations for one sentence should generally be avoided, unless there is a good reason. This backs Tvashtar2919's original point. I'll go ahead and try to bring the number of citations down to that number except for the list mentioned earlier. Phlsph7 (talk) 08:07, 5 February 2022 (UTC)

One instance of a particular reference in a short section like the "Measurement and analysis" section is usually enough. One instance in a paragraph is definitely enough. Multiple instances of a long work like a textbook make sense where you direct the reader to different pages. Constant314 (talk) 17:07, 5 February 2022 (UTC)

I don't think that the "one-instance-per-paragraph/section-principle" is generally accepted. See the examples at WP:CITATIONPLACEMENT: reference [2] has two instances used in two subsequent sentences. Phlsph7 (talk) 18:06, 5 February 2022 (UTC)

I generally adhere to WP:REPCITE. Constant314 (talk) 19:39, 5 February 2022 (UTC)

Illustration

The "understanding curve" commonly believed to illustrate Dunning–Kruger effect

Colleagues, while I agree that this image may not have reliable sources and such, I should mention that it is certainly notable as in the public view it epitomizes the idea of Dunning–Kruger effect and is the first thing that comes to mind when the phrase is mentioned (which is the entire reason why I tried to added it to the article, in passing). I am perfectly fine if you disagree with the addition, but as a middle ground I would recommend to either add it with a respective comment, something like : ------>

or at the very least, add respective comment to its Commons page, explaining that; it may save you some effort guarding this article in the future. -- Wesha (talk) 18:28, 8 February 2022 (UTC)

I agree with both sides: an image would be quite helpful but this particular image lacks authenticity and contains a lot of questionable information, such as the curvature and the names of its sections. For better images, have a look at "Dunning, David (1 January 2011). Chapter five - The Dunning–Kruger Effect: On Being Ignorant of One's Own Ignorance. Advances in Experimental Social Psychology. Vol. 44. Academic Press. pp. 247–296. doi:10.1016/B978-0-12-385522-0.00005-6. ISBN 9780123855220" page 263 or "Schlösser, Thomas; Dunning, David; Johnson, Kerri L.; Kruger, Justin (1 December 2013). How unaware are the unskilled? Empirical tests of the signal extraction counterexplanation for the Dunning–Kruger effect in self-evaluation of performance. Journal of Economic Psychology. 39: 85–100. doi:10.1016/j.joep.2013.07.004. ISSN 0167-4870" page 90. I'm not sure about the fair usage policies for such images, but they would be a good addition to the article when paired with a short explanation. Phlsph7 (talk) 18:52, 8 February 2022 (UTC)

I also strongly disagree with the decision to remove the picture. As Wesha mentioned, the vast majority of people who learned about the Dunning-Kruger effects did so as a result of this visual representation, therefore meeting the notability criteria (and a few other Wikipedia languages (e.g., FR, IT, ES) already have a similar chart).

I would like to better understand the rationale behind "needs a reliable verifiable source", as I am a bit confused by what Constant314 means:

• If it means that the image itself needs to be from a verifiable source (as Phlsph7 mentioned), then this would be an "artificial" condition that would be very difficult to believe, as one of the reasons why this chart is so famous is that it is simple to reproduce. The overall shape and terms (e.g., Mt. Stupid) are now fairly well known and standardized, but there is no commonly agreed "original source" of the chart, as far as I know. In addition, securing the copyrights required to publish it on Wikipedia would be extremely difficult, as Wikipedia has a very strict policy on this. A caption indicating that it is a satirical representation of the article, rather than coming from the article itself, would in my opinion be sufficient to make it clear to any reader that the chart is not "original"
• If it means that references showing that this chart is popular are required, then it should not be a problem to satisfy, as it is very simple to find many reliable sources making references to this chart.

Therefore, it could be helpful, Constant314, if you could help clarify which of the two interpretations is the right one regarding your comment.

In either case, while one may argue that the visual representations that were reverted were not perfect, as far as I know, they do not break any Wikipedia policy. Therefore, I believe that removing them is more harmful to the article than keeping them, until a better image is found.

7804j (talk) 14:55, 15 February 2022 (UTC)

Without going in detail about the dispute I must challenge some misconceptions about Wikipedia protocols. First content of an article is not ascertained according to 'notability' (this is a function of deciding whether an article should be created - or removed from the Wikipedia) but according to other protocols such as verifiability WP:V, non-point of view WP:NPOV and due weight WP:DUE. Second just because content - the graph - is present on other language versions of Wikipedia is not a reason for it to be in English Wikipedia. Wikipedia (any language) cannot be used as a source for Wikipedia (any language). Content comes from reliable sources which are independent and verifiable. It doesn't matter whether the content is a graph, photo or text it must be sourced. So your arguments are contrary to Wikipedia protocols and it doesn't matter how strongly you disagree with content removal, if it doesn't accord with Wikipedia protocols then it shouldn't be in the encyclopaedia Robynthehode (talk) 16:54, 15 February 2022 (UTC)

I am not arguing that there are specific Wikipedia policy/protocols that give the right for this image must be added, but I am arguing that there is also no particularly strong argument/protocol that require it to be removed (although I am not an expert on Wikipedia policies). I believe that adding this illustration meets the criteria of verifiability (provided that we add references showing that these "satirical" depictions are common), non-point of view, and due weight (because it represents a viewpoint from a majority).

I also understand that Wikipedia is not a source for Wikipedia, this was more to illustrate the point that other Wikipedia communities seem to have gotten to a different consensus, which can be helpful context in the discussion.

As I mentioned, I think that providing a source to similar graphs is not an issue, if this is what is asked, but providing a source to this exact representation seems technically not possible due to the fact that there is no universally agreed representation.

7804j (talk) 17:40, 15 February 2022 (UTC)

My main point is that the graph contains false or at least questionable information, as I explained in my earlier reply. This graph looks very different from the graphs I referred to. The Dunning-Kruger effect is usually measured concerning the relation between subjective and objective ability, not concerning confidence. When measured this way, the graph looks very different: it is steadily rising, it's not an up-down-up movement. I agree with Robynthehode's critique of 7804j's arguments. Phlsph7 (talk) 17:42, 15 February 2022 (UTC)

If I understand correctly, a key concern is to misrepresent the original findings, or to give too much emphasis on a "subjective interpretation" of its results. On the other hand, my key concern is to not represent a major element of how this concept has been used and interpreted in the real world. So I would like to propose an alternative, of having this image in a new dedicated section, at a lower part of the article, that explains how the DK effect is often interpreted/represented outside of scientific literature, and have the image inserted there rather than at the top of the article (with very visible and clear caveats that it is not directly supported by the original research and therefore should be interpreted with care). I have no strong opinion on the name/wording to be used for this new section. Would this alternative help alleviate these concerns?

7804j (talk) 17:58, 15 February 2022 (UTC)

The problem is that the image may contain false information, so it shouldn't be included anywhere in the article. We would need a reliable source that confirms the contested information. But I agree with you that this article needs a good image, either in the lead or in the "Measurement and analysis"-section. Phlsph7 (talk) 18:10, 15 February 2022 (UTC)

I see two problems with the image. The lessor problem as that there are no quantitative scales on either axis. There is only vague explanation as to what each axis means. Images in Wikipedia should mean something. Is the ratio of the first peak to the first dip accurate? What do the inflection points mean? Is the picture based on real data or is it just a hand drawn squiggle? Here is another version of the chart:

Effetto Dunning-Kruger

.

Which is correct? The second problem is authenticity. The image may be an internet fact. It has been mirrored by so many web sites that it appears to be authentic. Yet, I have not seen it tied to a competent reliable source. In fact, the image itself may be an example of a fact that is believed to be authentic with high confidence by people that have low competence to establish its authenticity. It would be ironic if the DK article itself were an example of a product produced by people on the left end of the chart. Constant314 (talk) 19:37, 15 February 2022 (UTC)

Just a comment about this whole discussion: It's bizarre that some editors would try to include an article that illustrates a popular misconception about a topic, on the basis that the misconception is popular. The purpose of the Wikipedia article should be to explain the topic, not to explain what people wrongly think about the topic. Obviously these images do not belong in the article and are counter-productive. That "a few other Wikipedia languages (e.g., FR, IT, ES) already have a similar chart" or "other Wikipedia communities seem to have gotten to a different consensus" is no argument: Wikipedia can't be used as a source for Wikipedia, and if other versions of Wikipedia are spreading false information then there is no reason that we should. "providing a source to this exact representation seems technically not possible due to the fact that there is no universally agreed representation" seems a pretty decisive argument that we shouldn't be using this representation. MartinPoulter (talk) 10:54, 16 February 2022 (UTC)

Yes. This is obviously true, but apparently it still needs saying. --Hob Gadling (talk) 07:57, 15 March 2022 (UTC)

Opposite effect

I linked the text "opposite effect" to Imposter Syndrome, but it was reverted by User:Constant314, saying it was not the same. A quick Google of "Opposite of dunning-kruger effect" will yield almost exclusively results refering to Imposter Syndrome, so I would like to know why it is not fitting. Sarke (talk) 03:11, 7 March 2022 (UTC)

Underestimating one's skills is a long way from imposter syndrome. Constant314 (talk) 03:35, 7 March 2022 (UTC)

There may be some analogies between the two but, as Constant314 has already pointed out, there are also important differences. We would need a reliable source that says specifically this: that the Dunning-Kruger effect for high performers is identical with the Imposter Syndrome. Google search results are not sufficient. Phlsph7 (talk) 06:07, 7 March 2022 (UTC)

Repeated content

I'm loath to jump in, as I'm new to this much-attended article, but the text does seem baggy and repetitive of main points, for example the explanation of 'dual-burden'. Hope it can be tightened. Onanoff (talk) 19:33, 7 March 2022 (UTC)

The lead is supposed to summarize the contents of the article, see MOS:LEAD. So this form of repetitiveness is explicitly encouraged by the Wikipedia guidelines. For the rest of the article, the term "dual-burden account" is mentioned 3 times: In the definition-section, in the section on the meta-cognitive explanation and in the criticism-section. From what I can tell, each mention is relevant to the issue discussed. Phlsph7 (talk) 06:06, 8 March 2022 (UTC)

Recent publications

In the magazine of the British Psychological Society, there are two recent articles with different perspectives on the D-K effect. "The persistent irony of the Dunning-Kruger Effect" by Robert D. McIntosh and Sergio Della Sala summarises the charge that it's just regression to the mean and "The Dunning-Kruger effect and its discontents" by David Dunning argues that there is much more evidence for the effect than the studies that have attracted the statistical criticism. MartinPoulter (talk) 20:44, 11 March 2022 (UTC)

Thanks for that. Constant314 (talk) 06:49, 15 March 2022 (UTC)

Eleventh reference

what a terrible wiki page... The eleventh reference is to the paper from which the effect is named. It should be in the first paragraph. The paper has some diagrams with empirical datapoints. I suppose it would fair use to nab one of those diagrams. ok, its time for me to get off "mount stupid" now.. — Preceding unsigned comment added by 2A02:1205:34F8:A230:652E:ED7:C4FA:1B6C (talk) 13:00, 7 April 2022 (UTC)

References should be avoided in the lead, see MOS:LEADCITE. It seems to me that you ascribe too much importance to the order in which the sources appear in the article. The important part is that the various claims in the article comply with WP:VERIFY. Phlsph7 (talk) 16:38, 7 April 2022 (UTC)

Neutral Point of View

The lede is asserting that the effect is a cognitive bias, but that assertion is contested by the criticism to a significant level - with enough weight so that we should not make that claim in Wikipedia voice. Other parts of the article take it for granted that the observed effect is due to a cognitive bias, but a neutral point of view should attribute those assertions to the sources making them. How can we improve the article to make it do not assert conclusions without scientific support as absolutes? Diego (talk) 15:59, 19 April 2022 (UTC)

Phlsph7 has pointed me to recent a short archived thread where it was discussed to remove "hypothetical" from the WP:FIRSTSENTENCE. IMO the qualification was needed to make the description not biased towards the view that the effect is caused by a cognitive bias, which doesn't seem to have a strong scientific consensus behind it, given the sourced references attributing it to other causes. Diego (talk) 16:06, 19 April 2022 (UTC)

(edit conflict) P.S.: There are also some of my edits that wouldn't make sense to revert in terms of whether the effect has enough scientific evidence or not, as they are mere rephrasings of previous sentences to remove their MOS:EDITORIAL style. Since Constant314 has agreed that the edit is fine, I'm restoring it while the discussion goes on. Diego (talk) 16:18, 19 April 2022 (UTC)

Hello Diego and thanks for bringing up the issue. For a reliable source, see here:

Dunning-Kruger effect, in psychology, a cognitive bias whereby people with limited knowledge or competence in a given intellectual or social domain greatly overestimate their own knowledge or competence in that domain relative to objective criteria or to the performance of their peers or of people in general.

Most criticisms concern how the Dunning-Kruger effect is to be explained, not how it is defined. It does not seem controversial to me if "cognitive bias" is characterized as "a systematic tendency to engage in erroneous forms of thinking and judging", see the section "Definition". Do you have any sources that explicitly deny this definition of the Dunning-Kruger effect? Phlsph7 (talk) 16:14, 19 April 2022 (UTC)

Hello Diego. I do not disagree with you. However, we have had an edit war over this in the past. It got so bad, that an admin blocked the article from editing by everyone. The article can be changed, but only after a talk page discussion. This is the normal edit-revert-discuss process. You are expected to discuss after you have been reverted. My own position is that DK is a real effect but not necessarily a cognitive bias which "a systematic pattern of deviation from norm or rationality in judgment." I don't see the DK effect as leading to irrational judgements. Constant314 (talk) 16:20, 19 April 2022 (UTC)

@Phlsph7 I have no problem with including that definition that describe the effect as a cognitive bias, since that is the way that it is recognized in popular culture. The problem I have is adopting that definition ourselves in Wikipedia's voice if there is no scientific consensus that this really is the cause, and we have well-sourced criticism positing credible alternative causes for the observed effect - in special the claims that it may be caused by errors in the design or interpretation of the experiments themselves (and therefore not a real psychological effect). We could attribute the "cognitive bias" to the dictionary definition by Britannica (which is a tertiary source, not a psychology journal) while not describing it ourselves as a cognitive bias, to maintain WP:IMPARTIAL tone.

You claim that there is no scientific consensus on its definition. I hold that the lack of consensus mainly concerns its explanation, not its definition. Do you have any sources that explicitly deny defining the Dunning-Kruger effect as a cognitive bias? Phlsph7 (talk) 16:54, 19 April 2022 (UTC)

Sure, they're linked in the Dunning–Kruger_effect#Criticism_and_alternatives section :-P But you're reversing the burden of proof; when we have references making contradictory claims, you'd need a good reference weighting them and describing the scientific consensus as a cognitive bias, explaining why the critics are wrong in their arguments. Otherwise we shouldn't assert in Wikipedia's voice that this cause and no others is the real one; a mere dictionary definition won't do. So far, we have good studies claiming that either statistical errors or other different bias are the cause, and we should present those with the same of due point of view as the view that the effect is a bias on its own merit. The article is in very bad shape in this respect. Diego (talk) 17:09, 19 April 2022 (UTC)

We have at least one source that defines the Dunning–Kruger effect as a cognitive bias. Do we have at least one source that denies this? If there are so many sources in the article that allegedly deny it, then it should be easy to cite which ones you have in mind. Otherwise there is no case for WP:NPOV. Phlsph7 (talk) 17:23, 19 April 2022 (UTC)

I agree that there is no problem with the neutrality of this article as both points of view are fairly represented. The issue is what goes in the lede. Britannica, I believe, is a tertiary source. There were some other sources, but they were blogs. It probably comes down to the meaning of cognitive bias. Constant314 (talk) 17:31, 19 April 2022 (UTC)

@Constant314 I was following a "true WP:BRD" style, including additional edits taking your feedback into account to improve the text, while trying to not contradict what you disliked of my changes. I don't believe in freezing the article state for the sake of it while achieving perfect wording in the talk page, at least not for articles in C-class quality that haven't had a thorough review of each sentence and reference. In my experience, in this cases it's best when all editors engage in good-faith "always-improving" article editing to make the actual words in the article better, rather than reverting "on principle" some good article edits that they nevertheless agree with. If we do reach a position where we radically disagree on how the article should look like, to the point of unproductive edit warring, we can then proceed to a more cautious style. Diego (talk) 16:46, 19 April 2022 (UTC)

I agree with Constant314 on this point: please wait for consensus on the talk page before making any further changes. Phlsph7 (talk) 17:03, 19 April 2022 (UTC)

Ok, I'll if we at least restore the "hypothetical" adjective in the FIRSTSENTENCE. Otherwise, I'll restore the {{npov}} tag - the article is in terrible shape in terms of neutral point of view with respect to the knowledge of the cause of the effect, and we should warn our readers about it in some way or another. Diego (talk) 17:10, 19 April 2022 (UTC)

Diego. Adding "hypothetical" is the edit war trigger. It cannot be added without a consensus. While I am leaning that way, at this point I am not part of a consensus to add that back in. Constant314 (talk) 17:19, 19 April 2022 (UTC)

Ok, then I'm adding the npov tag back, as we do have a formal content dispute now. Both of you, please don't remove the tag until the conditions for removal are met. We'll continue working on the until we agree on the way to best describe the different views around the effect, and the tag may guide readers to the Talk page to see the progress we make. Diego (talk) 17:29, 19 April 2022 (UTC)

There is no consensus for that either. Please be patient. These things take time. You need to wait at least three days so that other editors have a chance to get involved. Some of them are sleeping right now. As I pointed out, there is no problem with the neutrality as both points of view are represented. Constant314 (talk) 17:33, 19 April 2022 (UTC)

I do not agree that both points are fairly represented, that's why the dispute tag - it's meant to signal a lack of consensus. As I explained, I consider it beneficial for readers while we discuss - the longer the dispute until more editors can step it, the more reason to have it. Diego (talk) 17:40, 19 April 2022 (UTC)

The way to address that is to improve the content in the body of the article, especially if you can add reliable sources. In an effort to recruit more editors, I posted a notice in the Wikipedia talk:WikiProject Psychology. I have been aggressively removing the dispute tag, hopefully before the bot posts it on the notice board. That brings out a bunch of editors that have no interest in the subject but have a lot of interest in wiki-lawyering. It becomes a war. Give negotiation a chance. There is nothing here that will cause harm. Constant314 (talk) 17:58, 19 April 2022 (UTC)

As I see it, there is no consensus for adding the tag, so the latest stable version should remain until a consensus is reached (consensus to add the tag, to change the article, or to keep it as it is). The stable version does not contain the tag. Phlsph7 (talk) 05:28, 20 April 2022 (UTC)

I am content to allow minor word smithing on other issues as long as it is not disruptive, which so far it is not. Diego's most recent edit addresses awkward language without changing the meaning. I think this will work if Diego agrees to not make multiple edits without giving us time to improve each edit and if Phisph7 will hold off a little longer before hitting the revert button. Constant314 (talk) 17:17, 19 April 2022 (UTC)

I don't agree with this change since it does more than reformulate the issue: criticism of metacognitive explanation is not the same as denying that a cognitive bias is involved: the better than average effect is a cognitive bias that explains the Dunning-Kruger effect without metacognition. Phlsph7 (talk) 17:27, 19 April 2022 (UTC)

You didn't give me enough time to attempt to fix it. Constant314 (talk) 17:35, 19 April 2022 (UTC)

(edit conflict)@Constant314 - Thanks! I agree that one-off edits with enough time to review each one, and feedback on what is and is not acceptable on each edit, will be beneficial to the article.

@Phlsph7 - The point is that we do have several explanations for the cause of the observed effect (that measurements make some people appear to miscalculate their own performance), some of them are psychological whereas others are not; but we are elevating one of them as THE true reason for the measurements. That is not a position that we can adopt without an overwhelming scientific evidence that cognitive bias is the reason for those observations and all other possible explanations (based on statistical effects or errors in the experimental design) are likely wrong. Diego (talk) 17:36, 19 April 2022 (UTC)

I get the impression that we are partially talking past each other. Diego seems to hold that the main source of disagreement in the academic literature is whether the Dunning-Kruger effect is defined as a cognitive bias or not. If Diego were correct, I would agree that the first sentence shouldn't take a position on this question. However, I hold that the main source of disagreement in the academic literature is whether the metacognitive explanation (i.e. the dual burden account) is correct or not. In this sense, the first sentence (and the rest of the article) is unproblematic since neither position is accepted. For example, from McIntosh, Robert D.; Fowler, Elizabeth A.; Lyu, Tianjiao; Della Sala, Sergio (November 2019). "Wise up: Clarifying the role of metacognition in the Dunning-Kruger effect":

The original account of the DKE involves the idea that metacognitive insight requires the same skills as task performance, so that unskilled people perform poorly and lack insight. However, global measures of self-assessment are prone to statistical and other biases that could explain the same pattern.

This source is exactly in tune with my argument. I keep pushing Diego to cite a source otherwise since it is not a trivial matter and I don't see how we can reason ourselves to a conclusion on this issue without having a look at the sources. Phlsph7 (talk) 05:44, 20 April 2022 (UTC)

I wonder if we could replace "cognitive bias" with "tendency" in the first sentence? Constant314 (talk) 05:48, 20 April 2022 (UTC)

Could we replace "cognitive bias" with "tendency" in the first sentence? Constant314 (talk) 05:49, 20 April 2022 (UTC)

I prefer the term "cognitive bias" but replacing it with "tendency" is not a hard pill to swallow for me and is also supported by the sources. However, my criticism also affects many of the other changes proposed by Diego. Phlsph7 (talk) 05:53, 20 April 2022 (UTC)

What would you say about the expression "The Dunning–Kruger effect is the tendency or cognitive bias whereby..."? This both emphasizes the importance of cognitive bias and leaves open alternative definitions, as Diego contends. Phlsph7 (talk) 10:25, 20 April 2022 (UTC)

I am good with that. Constant314 (talk) 12:10, 20 April 2022 (UTC)

@Phlsph7 Excuse my thickness, but I don't see how assessing the validity of the metacognition explanation is incompatible with the possibility that it is or is not a cognitive bias. If, as your quote says, the same pattern may be explained by statistical biases instead of a shortcoming of low-skilled individuals, the observed effect would not be a cognitive bias at all but an error in the measurement process. That very same article by McIntosh et al [1] themselves don't discard the possibility that it's merely a statistics fluke ("it would be a gross misrepresentation to say that poor insight is the reason for overestimation amongst the unskilled. [...] the pattern can be induced (and greatly inflated) by a host of other factors and biases, some psychologically interesting, and some ‘merely’ statistical", "there has been debate over whether the effect derives from metacognitive differences,[...] general biases of self-estimation or from statistical artefacts [...] showing that each of these factors can contribute to shaping the typical DKE") or a different cognitive bias not correlated to low skill in the given task (i.e., not the cognitive bias that DK posited) - ("people tend to evaluate their relative standing optimistically, a bias known as the better-than-average effect", "They broadly confirmed that increased difficulty reduced the average percentile self-estimate, so that poor performers now seemed better calibrated, reducing and even reversing the asymmetry of the classic DKE"). That doesn't seem at all just a mere disagreement about whether the effect is cognitive or metacognitive.

Prominent studies are taking into account that the effect could be the result of a measurement bias that is unaccounted for in the design of the experiments; or caused by a different bias, very weakly related (Feld [2]) or unrelated (Ames [3]) to low skill in the task; and there doesn't seem to be a consensus on the actual causes of the effect. Therefore, I don't feel confident in definitely attributing it to a cognitive bias related to low skills as the article currently does. The question of whether the effect is not related to low ability but a different cause seems as strong as ever, and researchers recognize it as such.

I can agree with replacing "cognitive bias" in the first sentence, but rather than a "tendency" (which would imply a precise and consistent measurement of its direction and strength that doesn't seem to exist), I would say that it's an "observation" of people with low ability in a task overestimating their ability to perform it. The part that everybody agrees about DKE is that you can make repeatable observations of the effect, performing experiments where measurements of self-assessment will show an observed discrepancy with actual skill, having some perceived correlation with low ability but with disagreements about the reason for that correlation. So, why don't we seek an impartial tone and just describe the effect in those terms, without making ourselves a definite claim about its essence in Wikipedia's voice? For example we could precisely describe what was measured in the original DK paper (aggregated quartile values, and the difference between the self-predicted performance and the mean of the quartile), which now is only vaguely described in the lede with the unspecific "measured by comparing self-assessment with objective performance".

I've also seen several accounts of how people commonly misunderstands the DFE (one of them in the McIntosh paper: "the modern meme that stupid people are too stupid to know they are stupid is a dramatic oversimplification, propounded (perhaps) by those who know sufficiently little of the evidence") either by thinking it's stronger than it is, or even believing that anyone who is confident in their own abilities must be suffering from the Dunning-Krugger effect themselves. At the very least, we should take extra care that the lead section does not give fuel to such misunderstandings, suggesting certainties about the potency and nature of the effect of which we have no certainty. Diego (talk) 21:15, 20 April 2022 (UTC)

Hello Diego and thanks for making the effort of linking your arguments to some sources. The quotes you bring from McIntosh, like "there has been debate over whether the effect derives from metacognitive differences,[...] general biases of self-estimation or from statistical artefacts [...] showing that each of these factors can contribute to shaping the typical DKE" underline my point that the debate is about "metacognitive explanation vs other explanation" and not about "cognitive bias vs no cognitive bias". Feld argues that statistical effects play an important role in the observed pattern but do not present it as a full explanation, they just say that the pattern is weaker when one corrects for statistical mistakes. Ames does not mention the Dunning-Kruger effect (they talk about "the relationship between people’s actual interpersonal sensitivity (such as their ability to identify deception and to infer intentions and emotions) and their perceptions of their own sensitivity.").

As for the changes, I think "tendency" is better than "observation" but we can also use "observation" if that solves the issue. I think the expression "The Dunning–Kruger effect is the observation or cognitive bias whereby..." avoids your initial criticism of POV since it leaves open the possibility of a definition without cognitive bias while also acknowledging that some researchers explicitly define the Dunning-Kruger effect as a cognitive bias. The expression "observation of people with low ability" is problematic since it suggests that it's the people with low ability who observe this themselves. The description of the measurement by comparing self-assessment with objective performance is explained in detail in the section "Measurement and analysis". Your suggestions for additional descriptions in this area would probably fit better there instead of adding all this material to the lead, which should mainly provide a summary of the body of the article. Phlsph7 (talk) 05:22, 21 April 2022 (UTC)