Wikipedia:Featured article candidates/Confirmation bias/archive1

The following is an archived discussion of a featured article nomination. Please do not modify it. Subsequent comments should be made on the article's talk page or in Wikipedia talk:Featured article candidates. No further edits should be made to this page.

The article was promoted by Karanacs 21:36, 6 July 2010 [1].

---

Confirmation bias

Confirmation bias (edit | talk | history | protect | delete | links | watch | logs | views)

• Featured article candidates/Confirmation bias/archive1

Nominator(s): MartinPoulter (talk) 16:10, 25 May 2010 (UTC)

This is my first FAC. Wikipedia has a very little Featured Content on academic psychology, so I have substantially rewritten this article and taken it to GA. It has just had a very favourable Peer Review. If it is indeed worthy of an FA, I'd like to document the process of improving it, and apply this process to other psychology articles. Thanks in advance for all comments, MartinPoulter (talk) 16:10, 25 May 2010 (UTC)

• Comment—no links to dab pages; no dead external links. Ucucha 16:17, 25 May 2010 (UTC)

• Comments: This is definitely an impressive task to take on for a first FAC, so kudos to you. :) I have a few concerns/suggestions after a quick glance:

• It is common practice not to include numerous citations in the lead; most readers never get beyond the first few paragraphs, and[1] these things[2] tend to slow[3][4] them down. If the intro correctly follows WP:LEAD, all information should be cited further in the article anyway. Because the lead currently has 15+ citations, I suggest perhaps removing most to improve flow.

Most of these have been cut out, and the ref that appeared immediately after "Myside bias" has been moved to the end of the sentence so as to interfere with readability less. MartinPoulter (talk) 13:44, 29 May 2010 (UTC)

• There is a lot of italicization in the lead; please make sure these uses follow what's outlined in WP:ITALICS.

Yes, I've been overzealous. Now pared down to where a technical term is being introduced and defined (and used rather than mentioned), although there are several cases of this in the lead. MartinPoulter (talk) 19:24, 27 May 2010 (UTC)

• The "See also" section is quite long. Are all of these links vital?

Now pared down to four. MartinPoulter (talk) 19:24, 27 May 2010 (UTC)

• My main concern lies with the "Notes" section, although it's formatted beautifully and includes all pertinent info. Why are some of these works, especially those with numerous citations, not listed under "References"? Also, some of the page ranges seem incredibly broad: "Baron 2000, pp. 162–164" is fine, as it's pointing to only three pages, but "Oswald & Grosjean 2004, pp. 79–96" is potentially problematic. That one citation (used 14 times) covers a range of seventeen pages! Could you break these individual citations apart and state specific page numbers instead? That would be so much more helpful to researchers. The more specific the better. There may be other potential examples throughout, although this seems to be the largest range cited.

Those are my initial reactions, hope they help! Best of luck, María (habla conmigo) 16:51, 25 May 2010 (UTC)

I wasn't sure whether this was required, so thanks for clarifying. I will start work on the Oswald & Grosjean and the Nickerson refs. Lewicka as well perhaps. MartinPoulter (talk) 19:24, 27 May 2010 (UTC)

These three refs now done. MartinPoulter (talk) 16:04, 28 May 2010 (UTC)

I've struck my comments, as all concerns have been addressed. Although I don't feel I know enough about this subject matter to lend my full support, from where I stand everything looks much better. Great work. :) María (habla conmigo) 12:15, 24 June 2010 (UTC)

• Comment Some books (eg footnote 57 59 adn 60) have no page. Can a page be given so that it can be looked up easily? YellowMonkey (vote in the Southern Stars and White Ferns supermodel photo poll) 06:54, 26 May 2010 (UTC)

The first two book now has the page ref. The second one I've deleted as redundant. The third one I can't get hold of, but there are more up-to-date sources which make the same point, so I will substitute them tomorrow. MartinPoulter (talk) 19:24, 27 May 2010 (UTC)

Oooops, didn't get round to it today (the Aaron Beck ref which was number 60 is now number 74). Hopefully I will get time to do this tomorrow. MartinPoulter (talk) 16:04, 28 May 2010 (UTC)

The Beck paragraph has been totally rewritten, so all three of these refs are now fixed. MartinPoulter (talk) 13:27, 29 May 2010 (UTC)

Sources issues: The sources themselves look good, but there are a few presentation and formatting issues:-

• Why are some books, but not others, listed as "References"? There are many unlisted books among the citations (Ross and Anderson, Sutherland, etc.). As it stands, the partial list of books is useless as a basis of assessing the range of references used in the article.

All the refs are inline. The References section is there just so that citations can be made of multiple different pages within one source. What do you suggest should change? MartinPoulter (talk) 19:24, 27 May 2010 (UTC)

Perhaps the heading "References" was misleading. I've changed it to "Cited works" (maybe "Sources" would be better). Is that the right approach? MartinPoulter (talk) 15:42, 28 May 2010 (UTC)

I'd still say the title is misleading, because this is a list of only some of the cited sources. I'm afraid that the MOS advice in this area is very unclear, leading to a lot of confusion among terms such as "references", "sources", "footnotes", etc; editors use them to mean different things. My personal advice would be to treat all books in the same way, and list them all, under a section that could then be called "Bibliography". This would mean some tiresome work, but would have some logical consistency. Brianboulton (talk) 18:30, 11 June 2010 (UTC)

No problem moving the books into the list, but before I do so I want to be sure it will actually achieve consistency. Two of the sources (Friedrich and Nickerson) are journal articles rather than books, so it seems they wouldn't belong in a bibliography. Looking at recent FAs, Karl Aloys zu Fürstenberg has Footnotes and Sources, with some of the articles sources in the former. Keith Miller with the Australian cricket team in England in 1948 has a similar scheme, but named differently. Lycoperdon_echinatum has References and Cited text where some books are included in References. If I can't do that in this article, then fair enough, but I'd like to be clear why. MartinPoulter (talk) 09:48, 12 June 2010 (UTC)

• Ref. 44: What is meant by "Adapted from..." What has been "adapted", and by whom?

From the sources, it's clear that the arthritis experiment is a pivotal study historically, but another textbook source gave a fictional numerical example which I think makes a useful illustration. This numerical example was about good mood, not arthritis. Rather than introduce a confusing extra example, I thought I would use the textbook numbers, but label them as an example of how someone might think about arthritis. I also changed the smallest number from 3 to 2, to make the correlation negative rather than zero. This is a fictional, purely illustrative set of numbers, and is stated as such in the article. Is this okay to do, and how do you recommend I label it? MartinPoulter (talk) 19:24, 27 May 2010 (UTC)

Sorry, I can't advise here as I don't understand any of the above. I am happy to accept what you've done. Brianboulton (talk) 18:30, 11 June 2010 (UTC)

• Ref. 46: The formatting here is incomplete. As you have otherwise used citation templates, why not in this case?

Citation template now applied. MartinPoulter (talk) 19:24, 27 May 2010 (UTC)

• Page numbers missing from several books: refs 57, 59, 60 (I see this has already been noted)

Fixed (see above). MartinPoulter (talk) 13:27, 29 May 2010 (UTC)

Brianboulton (talk) 08:58, 26 May 2010 (UTC)

Please see WP:LAYOUT in relation to the "See also" section. Featured articles should not have an extensive See also section-- they should be comprehensive such that most links will have already been covered in the article. See also needs to be pruned to only items not already mentioned in the article, and anything that isn't mentioned already raises comprehensive concerns. I'll review this article as I find time. Also, use of WP:ITALICS needs review, and author names in citations are inconsistent; some have last name, first name; others first name, last name. SandyGeorgia (Talk) 13:22, 26 May 2010 (UTC)

"See also" has been pruned, italics have been cut down and author, coauthor, editor and translator names are now all formatted the same way. MartinPoulter (talk) 10:27, 12 June 2010 (UTC)

Sandy, where do you see that the author names are formatted inconsistently? From what I can tell, they follow standard citation styles outlined in MLA and TUR: the first author of a work is listed as "surname, first name" while any other additional authors are "first name surname". APA is different I believe, but as long as the styles aren't mixed it should be okay. María (habla conmigo) 14:02, 26 May 2010 (UTC)

I found a couple of places where co-author or editor names were lastname firstname. These have been changed for consistency. MartinPoulter (talk) 21:35, 2 June 2010 (UTC)

That is correct; I thought to raise the same point myself, but it's how this template formats co-authors. There's a way round this - you can juggle with the template to get:

• Hastie, Reid; Park, Bernadette (2005), "The Relationship Between Memory and Judgment Depends on Whether the Judgment Task is Memory-Based or On-Line", in David L. Hamilton (ed.), Social cognition: key readings, New York: Psychology Press, p. 394, ISBN 0863775918{{citation}}: CS1 maint: multiple names: authors list (link)

It's what I tend to do in my own articles, don't know if it's really any better. Brianboulton (talk) 16:03, 26 May 2010 (UTC)

Response: Thanks for the comments so far. I've been away yesterday and am occupied today, but will get serious time to work on page numbers and other issues tomorrow. I agree with María on the author name order: I have used the citation templates in an consistent and acceptable way (but thanks to Brian and YellowMonkey for pointing out the few exceptions) and in a way that distinguishes authors from co-authors. As to why some books are in References: this is so that multiple footnotes can be made referring to different pages of that source. Should I move all books to the references section, including when only one page of the book is cited? MartinPoulter (talk) 12:43, 27 May 2010 (UTC)

I think that is standard to do yes. Personally I believe it looks neatest that way.Esuzu (talk • contribs) 14:33, 28 May 2010 (UTC)

I question whether it is necessary for FA though. Here are some recent FAs that have separate notes and bibliography sections (however named), but include at least some book details in the notes: 1937 Social Credit backbenchers' revolt, Neville Chamberlain, George V of the United Kingdom, Karl Aloys zu Fürstenberg, Margaret Fuller. There are many more where the notes and refs are all in one section without a separate book list, see e.g. Bird.

There's also the usability question of whether users should have to make two clicks to see the citation details, or just one when the ref is only used once.

I'll move all the books to the bottom if there's consensus, but I'd like to be sure beforehand if it's necessary. MartinPoulter (talk) 15:40, 28 May 2010 (UTC)

Sorry for not answering directly, I forgot to watch this page. You have good points, there is no need for it at all. I just personally like it because I can see all the books used easily. But it doesn't really matter. Esuzu (talk • contribs) 18:45, 2 June 2010 (UTC)

It seems that I have answered the objections raised above this comment, although there's scope for clarification of some of them. Would the editors who posted above like to comment further? MartinPoulter (talk) 15:03, 11 June 2010 (UTC)

• Support—I'm satisfied that the article meets the FA criteria. (The lead image issue is more of a style issue than an FA concern.) Thank you for putting this interesting article together. I hope we see more like it in the future.—RJH (talk)
  • Comment—It looks pretty good, but a few concerns lead me to hold of on supporting the article just yet.
  • The lead does not appear to be a summary of the article. (See WP:LEAD.) For example, it does not summarize the history.

    Is it going in the right direction now? I'm still concerned about the long sentence with the four different technical terms, but still working out how to simplify it and get a clearly structured result. MartinPoulter (talk) 21:11, 4 June 2010 (UTC)

    It looks good now. Thank you.—RJH (talk)

  • I'm not sure I'm fully happy with the starting paragraph. The first two sentences don't appear to be communicating the same concept, at least to me. The first relates a preference to types of information; the second regards how information is processed.

    I've made the first sentence about preferential treatment, which is clearer than "preference". MartinPoulter (talk) 13:48, 4 June 2010 (UTC)

    Would "favor" be better? "tendency for people to favor information that confirms their preconceptions or hypotheses..." MartinPoulter (talk) 14:09, 4 June 2010 (UTC)

    I've implemented the above after it was discussed in article talk. MartinPoulter (talk) 16:44, 4 June 2010 (UTC)

    • This Science Daily article defines it as "a tendency to search for or interpret information in a way that confirms one's preconceptions". The current first sentence lacks that interpretation aspect. The second sentence reads as a more detailed expansion on the first, so I don't see it as resolving the issue.—RJH (talk)

"give preferential treatment to" and "treat favorably" are more strictly correct, but make the sentence unwieldy. I was impressed by a remark in the Oswald & Grosjean ref that we are dealing with a phenomenon which is basically a preference for one kind of information, that manifests in three different ways. Hence the first two sentences. It's only that one ref that puts it that way, and I'm happy to do something more like your link (which, by the way, is quoted from an earlier version of this article). How about "...tendency for people to seek, interpret or remember information so as to confirm their preconceptions..."? It might make the second sentence redundant, but then I could move the second sentence to the top of the Types section. MartinPoulter (talk) 21:11, 4 June 2010 (UTC)

That might work. Or possibly you could clarify the connection between the first two sentences by starting the second with something like, "This inclination causes people to reinforce their..."—RJH (talk) 21:25, 4 June 2010 (UTC)

• • To me some parts of the article have far too much parenthetical text, which can tend to break up the flow.

    I've removed a lot of this. There's still some in the lead. I'm bending towards the idea that I need to eliminate some technical terms (like attitude polarization) from the lead. It might make the whole thing more readable. MartinPoulter (talk) 13:48, 4 June 2010 (UTC)

    • It looks good now. Thank you.—RJH (talk)

  • An illustration of confirmation bias in the lead would be nice. Would Clever Hans be a good example, or is that a distinct phenomenon? If not, then perhaps something to demonstrate confirmation bias in the global warming debate, or in the belief that accidents increase during a full moon. Or whatever. It just looks a little bare up there.

    I've used gun control as an example, since that topic's explored quite thoroughly in the Tabor and Lodge paper. MartinPoulter (talk) 14:52, 4 June 2010 (UTC)

    When I said illustration, I probably should have made clear that I meant a graphical presentation. Something that will grab the attention of the casual browser, gain their interest, and immediately make clear the concept. I know it's not a requirement, but I just believe that, since this is an encyclopedia, a certain amount of eye-candy is needed to draw casual readers in long enough to engage them with the material.

    As an example, for the gun control topic, you could use some type of iconic illustration. E.g. File:Colddead-fp.jpg or File:Virginia Tech massacre memorial flowers.jpg. Or perhaps a two panel illustration that would symbolically present the arguments in favor of gun ownership vs. those in favor of gun control.—RJH (talk)

    Doh! Well, I'm glad I misunderstood you, because the lead reads better with an example. This article has been difficult to find images for. A picture of the brain was removed in the GA review because it wasn't directly related to the article content. A Venn diagram at the start will scare some readers off. The relevance of a specific event like Pearl Harbour won't be apparent until the end of the article. I like your gun control suggestion and will give it some consideration when I come back to this tomorrow. MartinPoulter (talk) 21:11, 4 June 2010 (UTC)

    Hmm, here's a thought: the three wise monkeys (E.g. File:Hear speak see no evil Toshogu.jpg).—RJH (talk)

    Having thought more about the "Cold, dead hand"s suggestion, I like the idea of someone holding a firearm to signify the gun control issue, but not an identifiable individual. A fan of Charlton Heston might object to his face at the top of the Confirmation bias article. There isn't a similarly internationally recognisable pro-gun-control figure who can balance, and the relevance of the image of flowers won't be apparent when it's a thumbnail. Just the image of a firearm being brandished by a civilian would do the work of suggesting the issue. Maybe File:Matt - pistol.jpg? MartinPoulter (talk) 12:35, 6 June 2010 (UTC)

    That would probably work, although the ear protection suggests a responsible gun owner engaged in target practice. Here's another interesting image: File:Evstafiev-bosnia-serbs-boy-gun-to-head.jpg. Thanks.—RJH (talk)

Sorry I've been away from this, and will be away from WP for the next two days due to work. When I come back, I plan to add a gun picture, and Commons has a picture of an MRI machine that would be relevant to the section on biased assimilation. I don't see a problem with illustrating the issue of gun ownership with a responsible gun owner. I just want the reader to think of "ownership of guns by civilians" in a colourful way, not to suggest either a positive nor a negative connotation. Thanks for digging up interesting suggestions. MartinPoulter (talk) 20:49, 7 June 2010 (UTC)

That works for me. Thank you.—RJH (talk)

• • "Another study of biased interpretation..." The word 'another' is additive here, and thus unnecessary.

    Fixed. MartinPoulter (talk) 13:48, 4 June 2010 (UTC)

  • The 'Explanations' section should probably include some discussion as to why this type of thinking would have an evolutionary advantage (or not). I.e. why it was not selected against.

    I've put more content in the Explanations section about why the biases may be adaptive. The evolutionary connection is there, but subtle. I hope this is enough: tell me if not. MartinPoulter (talk) 13:48, 4 June 2010 (UTC)

    • That's perfect. Thank you.—RJH (talk)

  I'll check back later. Thank you.—RJH (talk) 17:02, 2 June 2010 (UTC)

  • Thanks very much. A couple of these are minor and will be fixed straight away. Others will take a bit longer: I'll need to go back to the paper sources. I'll have to choose an illustration carefully: Clever Hans is mentioned in the bias literature, but not in conjunction with confirmation bias. I take your point about the need for an example though. MartinPoulter (talk) 17:33, 2 June 2010 (UTC)

    I've done some work this evening, and found a useful source for the evolutionary connection that I need to read more thoroughly. The lead needs major surgery, but I'll come back to that later in the week. MartinPoulter (talk) 21:35, 2 June 2010 (UTC)

    Thank you.—RJH (talk) 16:51, 3 June 2010 (UTC)

• Comments - Interesting article. A few comments....
  • Tom Hanks is not politically neutral, as the article asserts. He's a frequent donor to Democratic candidates, and supported President Obama's candidacy in 2008 (not that there's anything wrong with that).

He's described as a neutral figure in the experimental report. I could get the report again (haven't got access to it right now) and find another example of a person they used.

It turns out the report is freely available online. The three examples of neutral figures mentioned are "Tom Hanks, Hank Aaron, William Styron". Hanks is the only one of these with an international reputation. I've no idea who the other two are without looking them up. Was Hanks a supporter of the Democrats in 2004? His support for Obama in 2008 does not in itself undermine his being a politically neutral figure in 2004. MartinPoulter (talk) 09:38, 13 June 2010 (UTC)

Yes, Hanks supported Kerry in 2004, and Gore in 2000.[2]. Maybe you could just replace "Tom Hanks" with "someone".Anythingyouwant (talk) 17:20, 13 June 2010 (UTC)

That doesn't work grammatically, so I just removed mention of Tom Hanks altogether. MartinPoulter (talk) 20:12, 13 June 2010 (UTC)

• • Why put a picture of a gun-owner at the top of an article about biased thinking? We ought to have a neutral picture at the top, like the MRI machine, or just a picture of a gun. At least balance it out with a picture of an anti-gun protester.

See the discussion with RJH above. The idea is to connote the topic of gun ownership without suggesting either a positive or negative reaction. I'd rather have a single image of a gun, if that would be an improvement, rather than a more confusing double image.

Would any of these be acceptable: File:Handgun collection.JPG File:Smith&Wesson_Model_19.jpg File:P14-45 handgun .jpg File:Mountaingun2006.jpg? MartinPoulter (talk) 09:31, 13 June 2010 (UTC)

I've decided I personally prefer the first one, so I've taken the liberty of putting that in. MartinPoulter (talk) 09:44, 13 June 2010 (UTC)

Good change, thanks.Anythingyouwant (talk) 17:20, 13 June 2010 (UTC)

• • The lede says that a synonym is "myside bias". But isn't there a much more common synonym, namely "confirmatory bias"? In my opinion, this last term should not just be mentioned in the lead, but should also be the new title for this article. The word "confirmation" suggests either a religious ceremony, or the political approval of nominees. I think "confirmatory bias" would be a clearer title, but in any event it needs to be mentioned as a synonym with myside bias.

The sources do in fact mention confirmatory bias, but they overwhelmingly use the term "confirmation bias", so there's no dispute that the article is correctly named: just look at the titles of the sources or do a Google Scholar search. I have put confirmatory bias as a synonym. Yes, words like "confirmation" have multiple meanings: that isn't a good reason to go against the language used by the sources. MartinPoulter (talk) 09:57, 13 June 2010 (UTC)

Thanks for adding this in the lede. Please consider adding to the title "(psychology)". I came here thinking the article was about confirmation of nominees.Anythingyouwant (talk) 17:20, 13 June 2010 (UTC)

Considered but rejected: "Confirmation" needs disambiguation but "Confirmation bias" doesn't. MartinPoulter (talk) 20:12, 13 June 2010 (UTC)

• • Isn't a major symptom of confirmation bias a preference to hire people who will always say "yes"? See Assistant#Yes_man. Maybe this could use a mention in the article (please tell me I'm right!).Anythingyouwant (talk) 04:27, 13 June 2010 (UTC)

I haven't seen it mentioned in the sources. It could be an aspect of biased search for information, but you'd have to find a reliable, hopefully academic, source making the link. MartinPoulter (talk) 09:25, 13 June 2010 (UTC)

The connection between "confirmation bias" and the "yes man" has been mentioned by reliable sources. See here. Since many lay people are familiar with the concept of a "yes man", this might be an excellent concept to make this article a bit more persuasive and relevant to people (you might even include a pic of the most famous yes man in literature: Uriah Heep). For millennia, yes men have been well aware of people's desire to have their biasses confirmed. It's not like the idea of confirmation bias was suddenly discovered by modern psychologists. And reliable sources amusingly characterize confirmation bias as having a little "yes man" in your head. And the little yes man in my head is telling me I'm right about this!!!  :-). Incidentally, it's fine for you to cite scholarly literature, but don't be afraid to cite other reliable sources too: further perspectives can make an encyclopedia article more interesting. I particularly like this Wall Street Journal article. Anythingyouwant (talk) 17:34, 13 June 2010 (UTC)

I've been through a lot of the sources in that Google search, and I couldn't find any that support the point you want to make (that "a major symptom of confirmation bias [is] a preference to hire people who will always say "yes" "). I don't see anything that suggests the article content is not complete enough for Featured Article.

The WSJ journal uses "yes man" as a metaphor to explain what confirmation bias is; something which the article doesn't need. All the other examples I could find were quotes or paraphrases of the WSJ article. Can you find individual sources that make your initial point? Or do you just want a picture caption saying "The Wall Street Journal describes confirmation bias as like a "yes man" in your head"?

"It's not like the idea of confirmation bias was suddenly discovered by modern psychologists." The article does not make that claim; in fact it says the opposite: see the History section. MartinPoulter (talk) 20:22, 13 June 2010 (UTC)

If you don't find anything in the WSJ article worth using in this Wikipedia article, then that's your decision, but I disagree. When a reliable source popularizes an arcane technical subject, that's pretty much a red flag for me that the Wikipedia article should take notice.Anythingyouwant (talk) 20:24, 13 June 2010 (UTC)

I'm open to putting in content from the WSJ article, but what specifically? There must be hundreds or thousands of reliable sources mentioning confirmation bias: they can't *all* be used in this article. MartinPoulter (talk) 20:32, 13 June 2010 (UTC)

Please see this book: "the more that confirmation bias influences decisions....the stronger the 'yes man' tug is." As far as the WSJ article is concerned, I thought the analogy of a little "yes man" in your head was kind of fun; the point is to jazz up this article, so it's not just a dry summary of technical literature, although I must say that you do a fine job of summarizing the technical literature.Anythingyouwant (talk) 20:37, 13 June 2010 (UTC)

Okay, so a picture of Uriah Heep, with a caption "Like Dickens' Uriah Heep, confirmation bias acts like an internal "yes man" reinforcing a person's beliefs", sourced to the WSJ article: would that be okay? MartinPoulter (talk) 20:44, 13 June 2010 (UTC)

Sure. I don't mean to be pushy about this. Secondary sources are actually preferred at Wikipedia, compared to primary sources, so really there's no need to go hunting down work by Scott Lilienfeld. The WSJ piece also mentions psychologist Gary Klein, and there would be no need to hunt down Klein's data either. One thing the WSJ article does is suggest a way for people to combat their confirmation bias; is that presently covered by this Wikipedia article?Anythingyouwant (talk) 20:46, 13 June 2010 (UTC)

Scientific papers are secondary, not primary, literature. That's what I was proposing, so we're talking cross-purposes if you think I was talking about scientific data. I'll think about adding some more about mitigating the bias.

I hope you don't mind if I set something down that you may well already know, just to prevent a further misunderstanding. The coverage of confirmation bias in WSJ (as well as many, many other reliable sources over the years) establishes that the topic is notable. If someone were suggesting that the article be deleted, then we could show the WSJ article and other links (in fact, all of the sources that are cited by the present WP article). Now, however, there's no debate over whether the article should exist. It's been written, gone through Good Article review and Peer Review and the purpose of this discussion is to decide whether it's FA quality. Since it's an article on a scientific subject, it makes sense to write it as far as possible from the best quality sources: review papers in journals, textbooks, popularisations written by scientists and so on. The existence of other sources doesn't itself count against the completeness of the article. Hope this clarifies. MartinPoulter (talk) 21:11, 13 June 2010 (UTC)

Uriah Heep image and "yes man" caption now added to the article. Thanks for suggesting an image: this has been a hard article to illustrate. MartinPoulter (talk) 21:31, 13 June 2010 (UTC)

I'll add a little section under Applications called "In Finance", mentioning that confirmation bias can lead to bad investments through overconfidence and mentioning the WSJ's advice for mitigation. How about that? MartinPoulter (talk) 21:45, 13 June 2010 (UTC)

Thanks. :-)

(Delayed by edit conflict) At the moment the article links to a blog post about confirmation bias in finance. I could replace that link with a link to the WSJ article. The WSJ article refers to, but doesn't cite, some research by a psychologist called Scott Lilienfeld. That research would seem to be the reliable source to track down and summarise in the article. MartinPoulter (talk) 20:41, 13 June 2010 (UTC)

See reply above about Lilienfeld and secondary sources.Anythingyouwant (talk) 21:00, 13 June 2010 (UTC)

Just to be clear, I meant research *papers* (secondary) rather than research *data* (primary). MartinPoulter (talk) 21:18, 13 June 2010 (UTC)

• Further comment - I notice that a lot of the sources are not linked. For example, this one is at Google Books. If a source is online, it would be good to link to it.Anythingyouwant (talk) 21:00, 13 June 2010 (UTC)

See the discussion at Wikipedia:Reliable_sources/Noticeboard/Archive_17#Using_Google_Books_as_an_appropriate_reference, particularly the final comment by Vassyana. MartinPoulter (talk) 21:17, 13 June 2010 (UTC)

Well, I guess that boils down to a technical issue. If Google Book links tend to become dead, then I agree with Vassyana. On the other hand, I haven't noticed them becoming dead, so I'll have to look into it some more. If they don't become dead, then they're extremely useful, IMO. Even if a Google Book link no longer links to the precise page you want, my experience has been that it still links to the book, and the reader can then go to the desired spot in the book.Anythingyouwant (talk) 21:54, 13 June 2010 (UTC)

Here are the Featured Article criteria. As well as having the problems documented above, Google Books links aren't required by the criteria. MartinPoulter (talk) 21:34, 13 June 2010 (UTC)

• Support - Well-written, and Martin has made a number of beneficial improvements during this FAC process. I still think that links to Google Books would be desirable, because they usually don't become dead (instead they sometimes link to page one instead of the desired page, but a person can then go to the desired page). If you do cave in and link to Google Books, here's an easy tool to use.Anythingyouwant (talk) 22:31, 13 June 2010 (UTC)

Looking back over what I wrote last night, I realise I didn't thank you much for your contributions, so thanks for all the time you've spent on this and for the resulting improvements to the article. The next step for me is to put in a section about investment, which will use academic sources in addition to the WSJ. I expect I'll do that later today. Cheers, MartinPoulter (talk) 12:26, 14 June 2010 (UTC)

• Discussion of FAC archives moved to Wikipedia talk:Featured article candidates/Confirmation bias/archive1; this has been raised before at WT:FAC, and isn't likely to change. It makes botifying of closed FACs much easier. SandyGeorgia (Talk) 17:24, 14 June 2010 (UTC)

• I had intended (for weeks) to weigh in on this FAC, but just haven't found the time-- my apologies! The article looks sound, but I have some prose concerns that I hope others will address. Speaking very generally, everything with wording like "people tend to" (or similar) just sounds too "weasly" ... I'd like to see those phrases somehow reworded, but am too busy now to suggest alternatives. "People" with a statistical background may view hypotheses more rigorously, so the generality in wording to "people" troubles me. I apologize for not having time to be more specific, and have no problem if my concern is ignored in closing. SandyGeorgia (Talk) 13:42, 14 June 2010 (UTC)

Grateful for your input, Sandy, and I'm happy to discuss improving the prose. I appreciate that you're enormously busy.

I agree that a generalisation would be wrong: people show these biases to different extents and on different issues. So we can't justify an unqualified statement such as "people interpret information to confirm their existing attitudes". Hence the phrasing in terms of a tendency or a disposition. Stating it explicitly as "people" makes clear that we're not talking about information processing in human beings rather than computers or statistical methods, which is a potential confusion. If there were a particular type of person who showed confirmation bias, or if it happened in a particular area, then we would have to focus on that, but the sources don't suggest this. Even people who have been explicitly trained to be even-handed show the bias. So the phrasing is a compromise that says these effects are ubiquitous in how human beings process information, but are not inevitable, and the scientific sources don't draw a hard line between when confirmation bias happens and when it doesn't.

Would it help if I put in more qualifiers to make statements sound less general, or would that make the prose sound even more weasly? MartinPoulter (talk) 16:10, 14 June 2010 (UTC)

Sandy, to directly address your concern about whether the language of the article is general, here are some quotes from some of the article's key sources. You can see they do phrase it in terms of a general tendency of people:

• Oswald & Grosjean, page 93: "People do indeed search for results that would confirm their hypothesis if the corresponding results could be found."
• Sutherland, page 102: "There are many other experiments showing that people do not try to disprove a hypothesis they are entertaining."
• Gale & Ball, page 341: "They invoke Klayman & Ha's (1987) proposal that a central feature of hypothesis testing behaviour is a tendency for individuals to adopt a positive test strategy..."
• Kunda, page 118: "Nevertheless, we also rely on the positive-test strategy when seeking evidence needed to test our hypotheses, and this can bias our judgements." and page 119: "Therefore people asked whether they are assertive will likely view themselves as more assertive than people asked whether they are unassertive."
• Fine, page 61: "Our first problem is that we are poor scientists. All sorts of biases can slip in unnoticed as we form and test beliefs, and these tendencies can lead us astray to a surprising degree."

MartinPoulter (talk) 13:32, 19 June 2010 (UTC)

That relieves my concern about the prose (but I'm still hoping we can do better-- not a big concern). SandyGeorgia (Talk) 18:21, 19 June 2010 (UTC)

• Comments about the lede in particular - Since I've gotten involved here, I may as well offer some more particular comments (bracketed in ALLCAPS), at least about the lede, which is a critically important part of any Wikipedia article. I'm not withdrawing my "support" but just indicating how my support could be made stronger....

Confirmation bias (also called confirmatory bias or myside bias) is a tendency for people to favor information that confirms their preconceptions or hypotheses, independently of whether they are true. This results in people selectively collecting new evidence, interpreting evidence in a biased way, or selectively recalling information from memory.

People tend to test hypotheses [PRECONCEPTIONS TOO? FIRST PARAGRAPH MENTIONED BOTH.] in a one-sided way, focusing on one possibility and neglecting alternatives. This strategy is not necessarily a bias, but combined with other effects it can reinforce existing beliefs [SO, DOES IT BECOME A BIAS WHEN COMBINED WITH OTHER EFFECTS?]. The biases appear in particular for emotionally significant issues and for established beliefs [SO, DO YOU NEED EMOTIONALLY SIGNIFICANT ISSUES OR ESTABLISHED BELIEFS FOR THE STRATEGY TO BECOME A BIAS?]. For example, in reading about gun control, people [“USUALLY”] prefer sources that affirm their existing attitudes and [“THEY”] tend to interpret ambiguous evidence as supporting their existing position. Biased search, interpretation and/or recall have been invoked to explain attitude polarization (when a disagreement becomes more extreme even though the different parties are exposed to the same evidence), belief perseverance (when beliefs persist after the evidence for them is shown to be false), the irrational primacy effect [JUST SAY “IRRATIONAL PRIMACY” INSTEAD OF “THE IRRATIONAL PRIMACY EFFECT” TO CONFORM WITH OTHER ITEMS ON THIS LIST] (a stronger weighting for data encountered early in an arbitrary series) and illusory correlation [USE PARENTHETICAL HERE AS IN THE PRECEDING ITEMS ON THE LIST] in which people falsely perceive an association between two events.

A series of experiments in the 1960s appeared to show [“SUGGESTED” INSTEAD OF “APPEARED TO SHOW” TO BE CONCISE] that people are biased towards confirming their existing beliefs. Later work has challenged the interpretation of these [“THESE” TO “THOSE”] results, but has found biased treatment of information in other contexts [SUCH AS? DOES "OTHER CONTEXTS" MEAN OTHER THAN EXISTING BELIEFS?]. Theories offered to explain the observed [“EXPERIMENTAL” TO TIE IN WITH THE 1960s EXPERIMENTS] effects include wishful thinking and information processing limitations [LIMITATIONS LIKE STUPIDITY OR WHAT?]. It has also been proposed that people show these biases [YOU MEAN CONFIRMATION BIAS GENERALLY, OR JUST WISHFUL THINKING AND/OR INFORMATION PROCESSING LIMITATIONS?] because they are pragmatically assessing the costs of being wrong, rather than investigating in a neutral, scientific way.

Confirmation biases contribute to overconfidence in personal beliefs and can maintain or strengthen beliefs in the face of contrary evidence. Hence they can lead to disastrous decisions, especially in organizational, military and political contexts.

Anythingyouwant (talk) 18:06, 14 June 2010 (UTC)

I've made some changes: I'll consider others. If the lede leaves some questions unanswered, that's a good thing because it has to draw readers into the article. Grammar that's ambiguous or misleading should be corrected though, so thanks. I know the phrase "the irrational primacy effect" seems cumbersome, but when I see irrational primacy in the sources that discuss confirmation bias, it's always in the full phrase "irrational primacy effect". "Irrational primacy" can have other meanings, as you see when you do a Google search. MartinPoulter (talk) 18:38, 14 June 2010 (UTC)

Thanks. I think maybe you ought to consider removing these sentences to make the lede more concise: "People tend to test hypotheses in a one-sided way, focusing on one possibility and neglecting alternatives. This strategy is not necessarily a bias, but combined with other effects it can reinforce existing beliefs.". My concern here is that you may lose readers at this point, and it would be better to get right into the discussion of how emotional issues like gun control can trigger confirmation bias (no pun intended). People will already be scratching their heads wondering why there's a picture of guns at the top of the article (even with the caption), and the two sentences that I'm putting up for deletion are kind of vague, ambiguous, and merely refer to preconditions that can sometimes make confirmation bias more likely, rather than addressing confirmation bias itself.Anythingyouwant (talk) 04:54, 15 June 2010 (UTC)

I'm reluctant to remove those sentences because they summarise one of the main points of the article, in the History and the Biased search for information section, that a lot of what *looks* like confirmation bias actually isn't. That bit of the lede coule be made less intimidating though. Maybe replace "hypotheses" with "ideas" which will be less technical and more general (hence more correct!) I'll have a go. MartinPoulter (talk) 23:59, 17 June 2010 (UTC)

That's better, I think. But I still see a problem in those sentences. I could be mistaken, of course, but let me try to pinpoint. Many people have a low tolerance for psychobabble gobbledygook, so it's important to not let people get a hint of a whiff of that in the lead. Right now, the lead tells us that people often test ideas in a one-sided way, while ignoring alternatives, and this can reinforce existing beliefs. How the heck is that not a bias??? The lead is saying that it's not necessarily a bias, but not giving any clue why. The implication is that the author of this Wikipedia article is using some hypertechnical definition of the word "bias" that we may or may not be made privy to later in the article. See?  :-)Anythingyouwant (talk) 01:38, 18 June 2010 (UTC)

How do you suggest the relevant bit of the article be summarised? MartinPoulter (talk) 08:15, 18 June 2010 (UTC)

I do not have the cited source Oswald & Grosjean (pages 82-83). However, I suspect what they mean can be crystallized by adding the capitalized words to the sentence in the lead: "Combined with other effects, this can reinfoce existing beliefs, RESULTING IN BIAS." If this is not what Oswald & Grosjean mean, then what they do mean should be explained better in both the lead and the body of this article, IMO. Incidentally, it seems there are now two separate lists of footnotes, which seems odd.Anythingyouwant (talk) 16:06, 18 June 2010 (UTC)

(unindent) Two problems with this proposal: reinforcement of existing beliefs is bias, not the other way round. Also, it doesn't address what you said above: "The implication is that the author of this Wikipedia article is using some hypertechnical definition of the word "bias" that we may or may not be made privy to later in the article." I would have thought that someone testing new ideas in a way that reinforces their existing beliefs is a bias.

Separate lists for informative footnotes and citation footnotes is a practice adopted by some existing Featured Articles. See for example Helmut_Lent. MartinPoulter (talk) 16:25, 18 June 2010 (UTC)

Okay, you're saying here that reinforcement of existing beliefs is bias. So why not say so in the sentence at issue? "Combined with other effects, this can reinforce existing beliefs, WHICH IS BIASSED.". The way it's written now gives the impression that it is biassed only in the particular situations described by the next sentence (emotional issues and established beliefs).Anythingyouwant (talk) 16:48, 18 June 2010 (UTC)

I guess I'll make this edit if there is no objection.Anythingyouwant (talk) 06:22, 20 June 2010 (UTC)

Sorry I didn't feed back. I agree that the lead would have more flow if it went into "The biases appear in particular..." after the second sentence because "the biases" refer to the three effects listed in that second sentence. However, we still need to summarise what the article says about positive test strategies, because that's one of the key recurring points of the whole literature on this topic. Notice that I've asked for help in summarising that part of the article, although you've taken it above as a request to summarise the sources.

I accept the argument you give above that it's redundant to describe biased behaviour and then say it's biased.

I've reworded and reordered the lead in a way which hopefully addresses both Anythingyouwant's and SandyGeorgia's concerns. MartinPoulter (talk) 09:58, 20 June 2010 (UTC)

That edit looks fine to me. Incidentally, I don't think I said that it's redundant to describe biased behaviour and then say it's "biased". On the contrary, when biased behavior is described, then it's very useful to say that it's "biased", because doing so ties into the main theme of the article, assures the reader that everyone is using the same vocabulary, and removes confusion about why biased behavior is described as "bias" in the following sentences. Anyway, this appears to be moot now in view of your recent edits, which look fine. Thanks.Anythingyouwant (talk) 16:22, 20 June 2010 (UTC)

Comment: A fascinating, well thought-out article. I look forward to supporting soon. Please forgive me for entering comments in sporadic fashion—that's all I can do at the moment. I promise I will work through the entire article in the next few days. *The main text needs to be internally coherent, distinct from the lede. Thus, this prefatory sentence in the Types section is inadequate: "Some psychologists use 'confirmation bias' for any of three different cognitive biases, while others restrict the term to selective collection of evidence, using 'assimilation bias' for biased interpretation." (That is, I assume that the "three different cognitive biases" are "selectively collecting new evidence, interpreting evidence in a biased way, or selectively recalling information from memory," as presented in the lede. But you cannot assume that the reader has actually read the lede or is sharp enough to instantly make that connection if they have.) You need to clearly state what the "three different cognitive biases" are—probably within em-dashes such as these—to make this sentence in Types work.—DCGeist (talk) 05:04, 24 June 2010 (UTC)

Thanks for this and I look forward to your other comments.

That was a weak part of the article and I've had a go at rephrasing it. This has meant removing the wikilink to cognitive bias but we already have List of cognitive biases in the See also list, and frankly "cognitive bias" is such a nebulous term it might not be essential. Since there are already two footnotes about terminology I've moved more information about terminology into footnotes so that people are taken more quickly to the "meat". Maybe I should do this with more of the content you quote? MartinPoulter (talk) 16:59, 24 June 2010 (UTC)

Your solution here works very well. I'll keep the possibility of footnoting information about terminology in mind as I read on.—DCGeist (talk) 18:26, 24 June 2010 (UTC)

Support. Full comments on the article talk page. --Moni3 (talk) 19:35, 24 June 2010 (UTC)

• Comment Lead seems weak on wikilinks (WP:BTW and so on - words like bias, belief, and such should be linked) . --_{Piotr Konieczny aka Prokonsul Piotrus| talk} 12:59, 25 June 2010 (UTC)

Fair point. I'll address this in a few hours time, after work. MartinPoulter (talk) 13:52, 25 June 2010 (UTC)

A bunch of terms are now wikilinked. MartinPoulter (talk) 16:17, 25 June 2010 (UTC)

Queries on Biased search for information subsection: *"They look for the evidence that they would expect to see if their hypothesis were true, neglecting what would happen if it were false."

• "What would happen" doesn't work as a parallel for "the evidence". Please rephrase, perhaps along these lines: "They look for the evidence that they would expect to see if their hypothesis were true, rather than information that might demonstrate it is false."

Reworded

*"This pattern, of a main preference for diagnostic tests and a weaker secondary preference for positive tests, has been replicated in other studies."

• Is "weaker secondary" not redundant in this context?

Good catch: removed "secondary"

*"One particularly complex rule-discovery task used a computer simulation of a dynamic system."

• This introduces a new paragraph and—I think—a new test. You can't speak of a task within a test without establishing the existence of the test itself.

reworded and simplified

*"They tended to stick to hypotheses even after they had been falsified by the evidence."

• This is an example of belief perseverance rather than biased search for information, is it not? It is fine to consolidate discussion of the test in a single paragraph within a single subsection, but it has to be clarified that this element of the test's results actually exemplifies a different aspect or effect of confirmation bias.—DCGeist (talk) 17:32, 25 June 2010 (UTC)

reworded

Queries on Biased interpretation subsection: *"Charles Lord, Lee Ross, and Mark Lepper ran an experiment with subjects who felt strongly about capital punishment, with half in favor and half against."

• I had found it a little odd that the three specific experiments described in the preceding subsection had been introduced without information about who conducted them or where or when they were conducted, but at least the style was consistent. Here, suddenly, the style changes for no apparent reason and we are given the names of the three experimenters. You need to revisit how you are presenting all of these experiments and ensure that the article introduces them in a fashion that is more consistent or one in which the rationale for variation in introduction styles is apparent.

  Having now gotten to the Polarization of opinion subsection I see the logic of providing the experimenters' names in this case, because the study is referred to again later and you can use an effective shorthand. The earlier impression of inconsistency means that this is still not ideal, but as long as there is a reason for it, it's satisfactory.—DCGeist (talk) 06:20, 26 June 2010 (UTC)

It was agreed in the GA review that where the method and results of an experiment were being discussed, the names would be left out, but where the scientists' analysis, opinions or interpretations were being discussed, they would be attributed to a named person. It would be nice to establish this or something like it as a guideline for WP social science articles. Looking again, it seems the paragraph about the Lord, Ross and Lepper experiment just discussed the method and outcomes, not what their opinions were, so this use violates that principle strictly, but as you say it enables a shorthand. I'm open to removing their names. MartinPoulter (talk) 16:08, 26 June 2010 (UTC)

Third party opinion here. If an experiment is especially revolutionary, historic, or groundbreaking, then name the lead researcher. Otherwise don't. Instead of saying the Smith and Jones study, you can have a lot of fun making up descriptive names of each study (e.g. the hypocritical-candidate-study) which would make the article more fun to read (not that it isn't huge fun already!). Descriptive names would be more meaningful and memorable for readers, and there would no longer be the apparent inconsistency of naming researchers for many studies but not for many other studies.Anythingyouwant (talk) 16:58, 26 June 2010 (UTC)

I see. If I can conjure up a rewording for the passage in Polarization of opinion that constitutes a material improvement there and eliminates the value of including the names in Biased interpretation, I'll suggest it. Again, it's satisfactory as is.

(ec) AYW's suggestion is spot on—I agree 100%.—DCGeist (talk) 17:23, 26 June 2010 (UTC)

By that criterion, the Lord, Ross, Lepper study needs names, because along with the Wason experiments, it's a central piece of research that is mentioned by almost all the sources and which influenced the direction of subsequent research. I question whether we should restrict it to the lead author though. Where there's a large list of authors, then it's sensible to say "A team led by Smith..." or "Smith and colleagues..." but when there are two or three it seems unfair to do this, especially as there are cases where the order is decided arbitrarily.

Stuart Sutherland's book Irrationality I consider the gold standard of popularisation in this area. He mostly avoids naming the authors in his description of an experiment, except for some pivotal studies, or using an author to connect a strand of research such as Wason's. MartinPoulter (talk) 11:24, 27 June 2010 (UTC)

Here's a proposal: Lord, Ross and Lepper were all at Stanford. I could replace their names with "a team from Stanford University" and refer back to the experiment as "the biased interpretation experiment". That would achieve consistency around the principle that we mention names only where the scientist's arguments or conclusions are being discussed. MartinPoulter (talk) 08:14, 3 July 2010 (UTC)

That seems good to me, though I think you'd want to refer back to "the Stanford biased interpretation experiment" for maximum clarity (though perhaps that's just what you had in mind).—DCGeist (talk) 07:50, 4 July 2010 (UTC)

Now done. MartinPoulter (talk) 18:37, 4 July 2010 (UTC)

*"They then read a much more detailed account of the study's procedure and had to rate how well-conducted and convincing that research was. In fact, the studies were fictional. Half the subjects were told that one kind of study supported the death penalty and the other undermined it, while for other subjects the conclusions were swapped."

• This passage evidently needs to be copyedited. I believe the following is correct: "They then read a much more detailed account of each study's procedure and had to rate how well-conducted and convincing the research was. In fact, the studies were fictional. Half the subjects were told that one study supported the death penalty and the other undermined it, while for the other subjects the conclusions were swapped." Please verify and apply, or otherwise copyedit. The next paragraph suggests that what is currently (and obviously incorrectly) "They then read a much more detailed account of the study's procedure", which looks like it should be ""They then read a much more detailed account of each study's procedure", should actually be "They then read a much more detailed account of one study's procedure".

• "Once they read the more detailed study, they almost all returned to their original belief regardless of the evidence provided, pointing to details that supported their viewpoint and disregarding anything contrary."

  Confusing. Was one study actually more detailed than the other study? Or did the subjects simply read a more detailed account of one of the studies? Please edit for accuracy and clarity.

• "Subjects described studies supporting their pre-existing view as superior to those that contradicted it, in detailed and specific ways."

  Confusing. Now it appears that subjects were exposed to multiple studies that were said to support or contradict their pre-existing view. Were they? If so, then the preceding discussion needs to be edited. If not, then this sentence needs to be edited, along these lines: "Subjects described the study that was said to support their pre-existing view as superior to the one that was said to contradict it, in detailed and specific ways."

reworded this section to meet the above three points. Does it need further change?

*"There were strong differences in these evaluations, with subjects much more likely to interpret their opposing candidate as contradictory."

• Poor grammar. You mean "with subjects much more likely to perceive the candidate they opposed as contradictory" or "with subjects much more likely to interpret the statements by the candidate they opposed as contradictory", right?

reworded

*"Instead, the subjects were actively reducing the cognitive dissonance of reading about their favored candidate's irrational or hypocritical behavior."

• I don't believe it's idiomatic to speak of the "cognitive dissonance of reading about X". I think a verb is necessary. How about something like "cognitive dissonance created by reading"?—DCGeist (talk) 18:13, 25 June 2010 (UTC)

reworded

Queries on Biased memory subsection: *"Existing psychological theories make conflicting predictions about selective recall."

• No need for this "existing" or anything like it, unless a contrast is being drawn with outdated psychological theories. No such contrast is present.

reworded

*"In one study, subjects read a description of a woman, including both introverted and extraverted behaviors."

• Awkward. A "description of person X" doesn't automatically "include" their behavior. Perhaps "detailing both introverted and extraverted behaviors" or something similar.—DCGeist (talk) 06:00, 26 June 2010 (UTC)

reworded

Query on Polarization of opinion subsection: *"Subjects were told that the basket either contained 60% black and 40% red balls or 40% black and 60% red: their task was to decide which. When one of each color were drawn in succession, subjects usually became more confident in their hypotheses, even though those two observations give no evidence either way. This only happened when the subjects had to commit to their hypotheses, by stating them out loud after each draw."

• This is quite confusing. The second sentence suggests that subjects were asked to state a hypothesis before any balls were drawn—which, if so, should be stated. The third sentence compounds the confusion: "This only happened"—what only happened? That "subjects usually became more confident in their hypotheses"? OK, but then why is "only" part of this phrase? Was there an undiscussed alternative process in which subjects were not obliged to state their hypotheses out loud after each draw? This whole passage requires a careful edit; it may need to be pulled apart and reconstructed from the ground up.—DCGeist (talk) 06:20, 26 June 2010 (UTC)

reworded MartinPoulter (talk) 16:08, 26 June 2010 (UTC)

Though improved, I still found the passage confusing—one aspect of the experiment, for instance, that remained unclear is if the subjects were presented with one or two baskets. I sought out the book and the relevant passage via Google Books and have edited the passage given my now fuller understanding of the experiment. Please read through what I did and see if I've introduced anything that strikes you as problematic.

I also changed the paragraph's (and subsection's) lede sentence: from "When people with strongly opposing views interpret new information in a biased way, their views can move even further apart", to "When people with opposing views interpret new information in a biased way, their views can move even further apart." While the "strongly" is surely an accurate qualification, it reads as misplaced because you immediately move to a description of an experiment (the bingo balls) where we can hardly say the subjects' views are "strongly" opposed—they're simply opposed, period (one might argue that they become strongly opposed as a result of attitude polarization, but that's an interpretive leap that the source doesn't make).—DCGeist (talk) 17:23, 26 June 2010 (UTC)

Good catch. I applaud what you've done with the description of the slide show: very elegant prose indeed. The description of the bingo basket experiment seems fine to me. MartinPoulter (talk) 11:28, 27 June 2010 (UTC)

Comments: Aside from phrase flagged for rewording by Casliber below, Preference for early information subsection looks good. Illusory association between events subsection looks good.—DCGeist (talk) 04:29, 27 June 2010 (UTC) Informal observation and Wason's research on hypothesis-testing subsections look good.—DCGeist (talk) 05:13, 27 June 2010 (UTC)

Queries on Klayman and Ha's critique subsection: *"A 1987 paper by Joshua Klayman and Young-Won Ha showed that the Wason experiments had not actually demonstrated a bias towards confirmation."

• Is it actually widely accepted that Klayman and Ha definitively "showed" this (a factive statement)? If so, rephrase to clarify that this is the consensus view, with sourcing. If it is more accurate to saw that they argued "that the Wason experiments had not actually demonstrated a bias towards confirmation", please rephrase thus. Rewording good.

It's deliberately factive but, as you point out, it needs better sourcing.

*"Klayman and Ha argued that in most real situations, targets are specific and have a small initial probability."

• I can surmise what is meant by a "target" in a "real situation"—the accurate explanation of how a given set of facts came to be. Whether my conjecture is right or wrong, this sentence is unnecessarily obscure to a layman. It should be clarified, and if that takes a whole additional sentence, that's fine. Rewording good.

*"This interpretation was supported by a similar experiment that used the labels 'DAX' and 'MED' in place of 'fits the rule' and 'doesn't fit the rule'. Subjects in this version of the experiment were much more successful at finding the correct rule."

• Two issues here: (1) "This interpretation" is unclear, as no "interpretation" that fits has been mentioned in the preceding couple of sentences. Is the reference to the overall interpretation described in the paragraph's second sentence? If so, then: "Klayman and Ha's interpretation of Wason's experiments was supported by..." (2) The arrangement of the passage suggests that it should be clear why replacing "fits the rule" and "doesn't fit the rule" with "DAX" and "MED" would lead to subjects having greater success in finding the correct rule, but it's not. Can this be clarified?—DCGeist (talk) 14:48, 27 June 2010 (UTC)

I've had a go at rewording. An improvement? MartinPoulter (talk) 18:52, 3 July 2010 (UTC)

Yup.—DCGeist (talk) 07:50, 4 July 2010 (UTC)

Queries on Explanations section: *"One-sided questions such as, 'Which situations make you feel awkward?' come across as more empathic than more neutral questions such as, 'Do you feel awkward in social situations?' This suggests that a preference for one-sided questions is a sign of social skills. This prediction was confirmed in an experiment where college students asked more of these questions when getting to know a high-status staff member than when interviewing a fellow student."

• The conclusion here struck me as dubious, so I sought out the abstract. I'm concerned about the accuracy of our description of the experiment, though as I can't access the full article, I must rely entirely on the abstract. (1) The abstract does not refer to a preference for one-sided over neutral questions, but for hypothesis-matching questions over (implicitly) non-hypothesis-matching questions. If "one-sided" is intended to mean "hypothesis-matching", that's not at all clear—in the present context, "one-sided" appears to mean something like "relatively presumptive". (2) The abstract indicates that it is specifically and significantly highly self-monitoring students who "asked more of these questions when getting to know a high-status staff member than when interviewing a fellow student." We might want to come up with a more accessible substitute for "highly self-monitoring". (3) The abstract says, "Results are discussed in light of a pragmatic perspective that points out the adaptive and socially useful value of what look like errors and biases from a strictly rationalist perspective." It does not mention empathy. Does the article, in fact, support the description of "one-sided"/matching questions as seemingly "more empathic"?—DCGeist (talk) 15:45, 27 June 2010 (UTC)

Yes, the paper does specifically say that matching questions come across as more empathic. That's how they cash out the "socially useful value of what look like errors". I only described one condition of the experiment, as bringing in the self-monitoring issue struck me as involving a relatively large payload of psychological terminology to get a small point across. "Hypothesis-matching questions" is more technically correct, but "one-sided" has been introduced in this article's review processes as more accessible. Are you suggesting an extra sentence to spell out "one-sided" in this context? MartinPoulter (talk) 08:14, 3 July 2010 (UTC)

Yes. The meaning of one-sided in this context is not at all self-evident. I would prefer hypothesis-matching as straightforward and more clear, even if ungainly. As the review process produced one-sided, that's fine—but it does need to be explained.

I appreciate your hesitation about bringing in the self-monitoring condition for the reason you describe. Nonetheless, I believe it's necessary. The description of the experiment's results rang false to me, and I think that the absence of that condition is the primary cause.—DCGeist (talk) 07:50, 4 July 2010 (UTC)

Having another read of the paper, I see it was a mistake to focus on presumptive/non-presumptive rather than matching/non-matching. I've rewritten the paragraph at greater length. MartinPoulter (talk) 19:57, 4 July 2010 (UTC)

Well done.—DCGeist (talk) 20:29, 4 July 2010 (UTC)

Query on Biased interpretation subsection: *"A study of biased interpretation took place during the 2004 US presidential election, and involved subjects who described themselves as having strong emotions about the candidates."

• Did the subjects actually describe themselves as "having strong emotions about the candidates" or strong opinions or feelings/sentiments?—DCGeist (talk) 22:33, 27 June 2010 (UTC)

I will check this when I'm next in the office, some time this weekend. I edit from home and from work: most of the sources are only accessible from the office. MartinPoulter (talk) 08:14, 3 July 2010 (UTC)

Quoted from the source: "Using NES item wording, we asked about nature and strength of party affiliation; obtained ratings on their feelings toward George W. Bush, John Kerry, Bill Clinton, Dick Cheney, the Democratic Party, and the Republican party using a 0–100 ‘‘feeling thermometer’’ (from cold to warm); obtained 4-point ratings of how often Bush and Kerry made them feel angry, hopeful, afraid, proud, and disgusted; and obtained 4-point ratings of the extent to which they saw the two candidates as moral, intelligent, dishonest, and out of touch with ordinary people. To be included subjects had to rate themselves as a strong Democrat or Republican and to endorse a difference between the two parties or the two candidates > 30 points on the feeling thermometer."

Maybe "feelings" is better then. Changed MartinPoulter (talk) 18:27, 3 July 2010 (UTC)

Comment: I made several copyedits to the final section, Consequences. Nothing I'd call major, but do look them over to make sure I haven't troubled the substance. That done, the section looks good.—DCGeist (talk) 20:41, 27 June 2010 (UTC)

Nice work: very elegant prose indeed. MartinPoulter (talk) 08:27, 3 July 2010 (UTC)

Comment: The article was overlinked. I eliminated links to such words as medicine, war, conflict, arthritis, murder, and homosexual—all (a) common dictionary words (b) used in utterly ordinary ways that (c) obviously have no special relevance to the topic of confirmation bias. Something to keep in mind for your future FA efforts, which I look forward to.—DCGeist (talk) 22:33, 27 June 2010 (UTC)

Thanks. MartinPoulter (talk) 08:14, 3 July 2010 (UTC)

• Support Comments I have read through once already to get my head around it. I am still ruminating. Writing these sort of articles (as with many) is a balancing act between accuracy and accessibility. I think I need to read through again to see if there are any bits which can still be clarified without losing meaning. Looks good. I will add my voice to being a little concerned about the lead image - a little "loaded" - especially as the subject only gets a brief mention, but not a deal-breaker as such and these topics can be insanely hard to find free interesting and relevant images for. I'll jot queries below. no deal-breakers or obvious improvements remain. Casliber (talk · contribs) 22:35, 26 June 2010 (UTC)

Thanks for your comments and very helpful copy editing, Casliber. The lead image was introduced after discussion earlier in this FAC. It's intended merely to connote a topic. I'm not sure which way it could be said to be loaded: I can imagine complaints from different perspectives. That, in a way, is the point. MartinPoulter (talk) 11:12, 27 June 2010 (UTC)

• Sorting out italics vs quotes - I thought when we are discussing word-as-word then italics was what one used. See WP:ITALICS bottom of emphasis, and words as words bits. Hence in this sentence:

That bit of the style guide seems to give multiple contradictory advice. I was working on the principle of using italics when a word or phrase is being used and quote marks when it is mentioned. Hence "Psychologists use the debiasing paradigm..." but "Psychologists use a technique called the "debiasing paradigm"... " This explains the examples that you give below. The wording of WP:ITALICS doesn't seem to rule this out, and it does cite the importance of the use/mention distinction. I'm happy to change your examples to italics if you still think it necessary. MartinPoulter (talk) 11:12, 27 June 2010 (UTC)

I agree some of the items could be construed in different ways - see I'd see the mention as word-as-word as described in the italics bit. I will ask another style editor as well. Casliber (talk · contribs) 20:03, 5 July 2010 (UTC)

WP:ITALICS says that "This category may also use quotation marks to distinguish words as words." and at the end of the Emphasis section we have "A term being introduced [...] should be italicised or quoted, but not both." So it seems the guideline gives us the option to use italics rather than mandating them. Most of the phrases in the list below were italicised when the article was submitted to FAC, and have gone to quotes after complaints that italics were over-used. MartinPoulter (talk) 12:35, 6 July 2010 (UTC)

Some psychologists use "confirmation bias" for any way - it should be italics not quotes (?)

and People prefer this sort of question, called a "positive test" (?)

and This effect is called "selective recall", "confirmatory memory" or "access-biased memory" (?)

This is called "attitude polarization" (?)

This belief perseverance effect has been shown by a series of experiments using what is called the "debriefing paradigm" (?)

Just to reiterate in case there's any confusion, it's my interpretation that the above phrases are being mentioned rather than used, and hence that italics are not appropriate. I've taken another mentioned phrase out of italics for consistency. MartinPoulter (talk) 18:45, 3 July 2010 (UTC)

• ...even when the order is evidentially unimportant - should be easy to reword plainly.

I've removed "evidentially"

• back again - I was thinking that the Consequences section was an important section to do well, to illustrate to the lay-reader how this can impact on real-world function:

Hence. "In studies of political stock markets, investors who resisted bias and treated news in a relatively neutral way made more profit" - can this be elaborated on at all as to how news can be treated more neutrally, or is that all explained in the next sentence?

I've expanded it with an example from the source.

Tetlock divided experts into "foxes" who maintained multiple hypotheses, and "hedgehogs" who were more dogmatic. - I think a sentence is missing here on how much more successful the foxes were than the hedgehogs (??)

Good catch. I've reworded.

should the In self-image section be integrated into the In physical and mental health - which you could rename in medicine and psychology maybe?

I see what you're saying, but personally I don't think so. One is about how people's minds work in the normal scheme of things. The other is about things going wrong medically or mentally, and how people deal with it.

Okay. I don't feel hugely strongly about this one either. My personal viewpoint is that it is all on a continuum, and I am mindful of the stubby sections, but again not a deal-breaker by any means. Casliber (talk · contribs) 19:57, 5 July 2010 (UTC)

Any other concrete examples which can be added to this section would be helpful. The subheadings are a little stubby for the most part, and this might end up being the most interesting bit of the article for many lay readers.

How much more content would you say it needs? Is it that you want more sections under "Consequences", or for the existing sections to be longer?

Overall, nearly there. Casliber (talk · contribs) 04:24, 5 July 2010 (UTC)

Thanks. I urge you to reconsider the merging of the final paragraph of the lede ("Confirmation biases contribute to overconfidence in personal beliefs...") into the former para. That final paragraph sums up the consequences of confirmation biases. The previous sentences are about causes of the bias. As well as the semantic reason for keeping these separate, it seems to me that the lede is more punchy and less off-putting with the shorter final para. You're the reviewer and it's up to you, but that particular split seems better the way it was. MartinPoulter (talk) 17:03, 5 July 2010 (UTC)

I agree. Though two sentences is indeed a short graf, the lede was stronger that way.—DCGeist (talk) 18:04, 5 July 2010 (UTC)

Okay, on thinking about it, I can see where you are coming from and am happy as a reviewer that it is not a deal-breaker for getting over the line, so rearrange away. Casliber (talk · contribs) 19:57, 5 July 2010 (UTC)

• Comment There are quite a few redirects that no longer link to any section in the article and some others that probably should. — Dispenser 18:09, 27 June 2010 (UTC)

Thanks for this excellent tool. The links to non-existent sections are now fixed, and some redirects to the article now go directly to the relevant section. I don't understand how to fix the duplicate the ref issue, since the code in question is produced by the citation template. MartinPoulter (talk) 08:02, 3 July 2010 (UTC)

• Support Comment. A good article - my concern is clarity of prose, mainly in the "Biased search for information" section:

• "Rather than searching through all the relevant evidence, they frame questions in such a way that a "yes" answer supports their hypothesis and stop as soon as they find supporting information" I didn't udnerstand this sentence. Does it mean "Rather than searching through all the relevant evidence, they seek evidence that supports their hypothesis and stop as soon as they find such supporting information"?

Maybe this sentence is trying to do too many things. It's crucial that people ask questions to which a "yes" answer rather than a "no" answer confirms their hypothesis, so your suggestion would miss out what's important. It's more like, "they frame questions in such a way that an affirmative answer supports their hypothesis and stop as soon as they find affirmative answers". Not sure how to do this in a less repetitive way. MartinPoulter (talk) 19:00, 3 July 2010 (UTC)

I find that no clearer. Actually, as I found myself suggesting below, wouldn't this para be clearer if the entire sentence ("Rather than searching through all the relevant evidence, they frame questions in such a way that a "yes" answer supports their hypothesis and stop as soon as they find supporting information.") was deleted? Maybe I'm missing something , but I think the whole explanation works without it. hamiltonstone (talk) 12:21, 4 July 2010 (UTC)

It's important in conveying the positive test strategy that the reader be told that people prefer affirmative answers to negative answers, and I don't think the "For example" sentence can do this on its own. I've reworded the paragraph. MartinPoulter (talk) 19:15, 4 July 2010 (UTC)

OK, I made an extra tweak and I'm happy with that. hamiltonstone (talk) 23:30, 4 July 2010 (UTC)

• "However, they reject tests that are guaranteed to give a positive answer, in favor of more informative tests" How can anything be more "informative" than something "guaranteed to give a positive answer"? I didn't get this sentence at all.

A test which is guaranteed to give one answer is totally uninformative. Thus any question which can reveal information about a topic will be more informative than that. The point is that though people search for information in a way that fixes on one hypothesis, in this experiment they were not so stupid as to choose a pseudo-test which is guaranteed to affirm that hypothesis. Would it help if "informative" were replaced by "diagnostic"? MartinPoulter (talk) 18:45, 3 July 2010 (UTC)

Actually it might help if the whole sentence was deleted. Now that I think I have understood you, I would suggest that it actually is a detail that qualifies the simple explanation, and in doing so makes that explanation of the basic phenomenon harder to follow. hamiltonstone (talk) 12:14, 4 July 2010 (UTC)

Having spelled out this tendency in hypothesis-testing, I think it's fair to people to then say that the tendency has a limit. I see now that I was trying to pack too much into one sentence. I've reworded the paragraph. While I don't expect this fully answers the concerns above, could you bear with me and say what problems you see with the new version? MartinPoulter (talk) 19:15, 4 July 2010 (UTC)

Made a slight change - that's all good. hamiltonstone (talk) 23:30, 4 July 2010 (UTC)

• "However, a later experiment gave the subjects less presumptive questions to choose from, such as, "Do you shy away from social interactions?"" - the significance of the "However" is not apparent to the reader until s/he reaches the end of the para. I think the point is that this later round of tests weakens the case for this type of confirmatory bias. If I have read it correctly, then the sentence beginning "However" needs to be revised to make clear where the argument is going. Something like "However, later experiments have called into question the strength of this confirmatory bias effect."

"However" was useless and misleading in this context, so I've removed it. MartinPoulter (talk) 18:45, 3 July 2010 (UTC)

OK, but I htink you've muissed my main point: the para lacks an explanation of the significance of these later experiments, hwich seems to be that the preference for positive tests (which leads to a biased search for information) is actually weaker than the preference to ask more neutral diagnostic questions, when people are given that choice. This seems to be significant, and the reader's attention should therefore be drawn to it in a more explicit manner. hamiltonstone (talk) 12:24, 4 July 2010 (UTC)

The confusion for me is that the final sentence of the paragraph seems to make the point clear and with the appropriate caution. "This pattern, of a main preference for diagnostic tests and a weaker preference for positive tests, has been replicated in other studies." Are you suggesting an improvement to this sentence, or that it needs to occur earlier in the paragraph? The suggestion you make above is in a very general kind of language that I hope we can avoid if possible. MartinPoulter (talk) 18:44, 4 July 2010 (UTC)

Exactly yes - I wanted something earlier in the para. I've now written a new intro to the para. Hope you think it is OK. With that change I'm now happy with the article and am supporting. hamiltonstone (talk) 23:41, 4 July 2010 (UTC)

• The description of the "bingo basket" experiment seems confusing. It refers to "the basket" (implying there is only one), but then to "each basket" (implying there are two). Can someone look at this again?

• I've been trying to handle the description of the "bingo basket" experiment. I did a nice edit on it a few days ago that somehow got completely lost. I've taken another crack at it. Please read it again and let us know if any confusion remains.—DCGeist (talk) 01:56, 2 July 2010 (UTC)

• That's better. hamiltonstone (talk) 03:18, 2 July 2010 (UTC)

• Rest of the article seems good. Well-referenced, neutral, good illustrations etc.

Would hope to see this improved and passed, though it has been here a while. Not sure what other 'comments' editors' thoughts are. hamiltonstone (talk) 00:36, 2 July 2010 (UTC)

Just an update: I've had a very hectic work week, so have had only a few minutes of WP time. I'm determined to press ahead and fix the remaining issues. I hope to get some serious time to do this over the next few days. I'm really glad to see that comments are continuing to come in, and I will address them fully. MartinPoulter (talk) 08:49, 2 July 2010 (UTC)

OK. A minor note: at some point in the last couple of days "their estimate of the probability increased." changed to "their estimation increased." I don't know why this was done, but the earlier text made sense; the latter doesn't. hamiltonstone (talk) 12:14, 4 July 2010 (UTC)

That was me, and it was done in the pursuit of concision. I've restored the clearer phrasing.—DCGeist (talk) 15:33, 4 July 2010 (UTC)

Support: Lovely to see a quality article on a serious intellectual topic such as this. Looking forward to more.—DCGeist (talk) 20:29, 4 July 2010 (UTC)

The above discussion is preserved as an archive. Please do not modify it. No further edits should be made to this page.