Talk:Publication bias

Untitled
Publication bias is a well-documented problem in a range of disciplines. See, for example, V. M. Montori, M. Smieja and G. H. Guyatt, "Publication Bias: A Brief Review for Clinicians," Mayo Clinic Proceedings 75 (2000): 1284-8; A. Thornton and P. Lee, "Publication Bias in Meta-Analysis: Its Causes and Consequences," Journal of Clinical Epidemiology 53 (2000): 207-16

Heavy submission delays induced by precarity / mobility / lack of permanent positions / lack of resources / overload
Researchers are not free to publish all their results. They have a hierarchy who has to agree for paying the publication fees, and results should fit to the publishing scope of the hosting laboratory, and to the scope of the hosting group. This prevents a high number of manuscripts from reaching the submission stage.

We can also consider the high mobility demanded to researchers nowadays: by heading from one lab to another on the basis of short contracts, results obtained at a certain time can take several / numerous years to get submitted for publication.

Another effect can be: the host / funding institution may prefer its human resources to perform studies on the topics which are arranging / interesting for their field of action. This can induce a retention of manuscripts that could be important from the researcher to increase its position stability.

This also induces that a number of the works that were performed / prepared on the basis of public investment are NOT present in the corpus of publications/manuscripts. Which ratio of the prepared works actually reaches the stage of submission to a scientific journal, or to an archive repository? cybertib 20:36, 11 May 2019 (UTC+2)

Merger proposal
I suggest that File drawer problem should be merged with this article. The two terms are so closely related and the two articles are covering such similar ground that it doesn't seem useful to have separate articles. Publication bias is the more general term and this article is more developed so it makes sense to merge File drawer problem into here rather than vice-versa. --Qwfp (talk) 10:57, 21 February 2008 (UTC)

I agree that "File drawer problem" is an example of Publication Bias. --71.185.40.151 (talk) 11:29, 3 June 2008 (UTC) --SV Resolution(Talk) 11:30, 3 June 2008 (UTC)

No objection. 69.140.152.55 (talk) 03:53, 21 June 2008 (UTC)

Publication Bias – is it always positive outcome?
Ionnidis( Why Most Published Research Findings Are False) states "Claimed Research Findings May Often Be Simply Accurate Measures of the Prevailing Bias". In some cases, wouldn't the prevailing bias be for a negative result? If the prevailing bias is that Wonderbread-and-bologna sandwiches don't improve serum glucose in diabetics, won't it be difficult to get a paper published that claims a positive result? --SV Resolution(Talk) 13:54, 3 June 2008 (UTC)

The article File drawer problem states "'Publication bias' is a more general term, as it may include differences in the availability or accessibility of published papers due to the language, format or journal of publication.". What is the accepted definition of "Publication Bias" --SV Resolution(Talk) 13:56, 3 June 2008 (UTC)

It may not be always positive. I was listening to a radio programme, which was discussing this issue. It seems that academic journals favour results that are divergent from the norm. For example, a new theory or unexpected finding is published because it is unexpected/unusual. For a period of time, the prevailing bias amongst journals are studies which support this case. However, as the theory becomes (seemingly) well grounded, there is a prevailing tendency for the publication of research which negate the theory. In light of this, I call for a change in the definition expressed in the opening paragraph of this article.--62.252.10.14 (talk) 07:37, 26 May 2011 (UTC)

there are other kinds of publication bias
the file drawer problem is just one among many different kinds of publication bias. This entry should thus be kept. (and further developed). --BirgerH (talk) 13:55, 21 February 2009 (UTC)


 * I went ahead and merged it, there really wasn't enough to stand alone once the OR was removed. Fences and windows (talk) 00:20, 29 March 2009 (UTC)

And indeed there is a "reverse publication bias" in financial economics. Timmerman and Granger (2004) argue (in "Efficient market hypothesis and forecasting", footnote 1) that:

"In studies of market efficiency, a reverse file drawer bias may be present. A researcher who genuinely believes he or she has identified a method for predicting the market has little incentive to publish the method in an academic journal and would presumably be tempted to sell it to an investment bank."

This also provides support for keeping the more general term of "publication" bias, unless we are prepared to live with the implications (of academic journals in financial economics acting as the "file drawer" for the unsold stuff:)

I managed also to trace the original "file drawer" term to Rosenthal (1979) (Rosenthal, Robert, 1979. The file drawer problem and tolerance for null results. Psychological Bulletin 86(3), 638-641.) — Preceding unsigned comment added by 83.7.162.252 (talk) 12:09, 11 December 2011 (UTC)

the scale of the publication bias
Is the "1 in 3" number attributed to Dickersin (1987) correct or sufficient? The abstract of that paper states that the ratio of published-to-discarded rejections of the null (hypothesis of treatment ineffectiveness) is 55% to 14%. Surely it is equally salient news that the proportion of significant to insignificant results is four times higher in published studies than in those which had been consigned to the file drawer (or rejected by peer review)? 83.7.162.252 (talk) 16:25, 11 December 2011 (UTC)

There were actually 3 other studies that found no difference b/w rates of publication but they never got published ;) — Preceding unsigned comment added by 139.52.176.32 (talk) 16:39, 1 November 2012 (UTC)

That's hilarious. Thank you. — Preceding unsigned comment added by 86.134.208.61 (talk) 00:24, 4 January 2018 (UTC)

Effectiveness of pre-registration
I'm an outsider to pharmaceutical research, but I'd have thought that “[With pre-registration,] negative results should no longer be able to disappear” is too strong a claim: that one can still do studies not published or registered with such publications, and thus have any negative results disappear. Worse, it would seem that publication bias can still persist even within such publications, because one can choose to register a study only once one is reasonably sure that the outcome will be positive under carefully contrived circumstances while presenting the results as being general.

That is not to say that a pre-registration policy will have no effect on publication bias, merely that “[not] able to disappear” seems too strong a claim, that better wording would be something like “in an attempt to reduce publication bias”. Pjrm (talk) 04:45, 23 February 2009 (UTC)

I think the way of putting it that best keeps to the NPOV is "will still be recorded". It assumes nothing not already declared in the article (the existence of publication bias and the file drawer effect). —Preceding unsigned comment added by 67.86.45.4 (talk) 22:31, 28 July 2009 (UTC)


 * Sorry, I can't remember where, but I saw something about the pre-registration guidelines not actually being followed. So you still get publication bias, but worse, believe you aren't. Perhaps someone can hunt down more info on this. Could have been Ben Goldacre.

131.107.0.73 (talk) 11:44, 10 February 2010 (UTC)


 * It doesn't work. At the bottom of http://ajcn.nutrition.org/content/early/2016/08/23/ajcn.116.134932.abstract, I see: "This trial was registered at www.clinicaltrials.gov as NCT02616471."  So this should mean that there's less risk of publication bias, I suppose.  But I think it's common to find studies at clinicaltrials.gov that should have published results, but don't.  I remember noticing one: https://clinicaltrials.gov/ct2/show/NCT02118883 and I haven't looked at many.  Even though it's a legal requirement that studies registered at clinicaltrials.gov must follow (I reviewed the instructions there to publishers, and it's implied at Clinical_trials_registry.  There doesn't seem to be an effective enforcement mechanism.  :-(  Anyone have a friend at the DoJ or FDA or something?  {edit: Even when the results have been published, researchers sometimes haven't bothered to update the results - https://clinicaltrials.gov/ct2/show/NCT02616471 says no results published, but my first link above shows they have been published.} -- Elvey (t•c) 02:22, 25 September 2016 (UTC)

Is the person who discoverd publication bias worth a mention?
User:U3964057 is trying very hard to keep my well-sourced sentence "Publication bias was discovered by statistician Theodore Sterling, in 1959". out of the article. I wonder why. --Arno Matthias (talk) 09:48, 9 July 2015 (UTC)
 * Hi Arno Matthias. Thanks for coming to the talk page. As should be clear from my edit summary, my concern is not that the content is inappropriate for inclusion in the article, but instead that it is undue weight for the lead. My suggestion would be to kick off the 'evidence of publication bias' section with a sentence along those lines. What do you, or others, think? Cheers Andrew (talk) 01:59, 10 July 2015 (UTC)
 * Agreed; the lede should summarize what's in the body of the article. Fgnievinski (talk) 03:22, 10 July 2015 (UTC)
 * If you wanted the sentence in, you could have moved it to a more appropriate place. Instead you deleted it. This leaves but one conclusion... I will now put it in a third time and see what happens. --Arno Matthias (talk) 08:11, 10 July 2015 (UTC)
 * Hi Arno Matthias. I'm not sure where your concern about some ulterior agenda is coming from, but thanks for your latest edit. By removing the addition from the lead you have addressed my concern and it sits well with me. Cheers Andrew (talk) 08:55, 10 July 2015 (UTC)

WEIGHT - size of the problem obfuscated.
For no good reason, the article gives no clear picture of the size of the problem. For example:

0. Reference kicinski2015_19-2 states, In the meta-analyses of efficacy, outcomes favoring treatment had on average a 27% higher probability to be included than other outcomes. In the meta-analyses of safety, results showing no evidence of adverse effects were on average 78% more likely to be included than results demonstrating that adverse effects existed. In general, the amount of over-representation of findings favorable to treatment was larger in meta-analyses including older studies. This reference is, oddly, used to support this article content: The study showed that positive statistically significant findings are more likely to be included in meta-analyses of efficacy than other findings and that results showing no evidence of adverse effects have a greater probability to enter meta-analyses of safety than statistically significant results showing that adverse effects exist. The bits I underlined are vague and inferior to the use of numbers from the source. Can we fix this?

Also, the use of that reference to support this article content: A recent study showed that publication bias is smaller in meta-analyses of more recent studies, supporting the effectiveness of the measures used to reduce publication bias in clinical trials. feels

feels inappropriate.


 * 1) There's a copyvio issue - that article content contains these phrases which appear to be copied verbatim from the source: publication bias is smaller in meta-analyses of more recent studies and  supporting the effectiveness of the measures used to reduce publication bias in clinical trials.!
 * 2) We should state how much smaller the bias was found to be in recent studies!

--Elvey(t•c) 21:43, 16 November 2015 (UTC)
 * No responses. Made some improvements to partly address.
 * 0. Addressed: Added "27%" and "78%".  Did not include the CIs, but they're in the source.
 * 1/2. Removed for now. Without numbers, this is just handwaving.
 * --Elvey(t•c) 18:57, 20 November 2015 (UTC)

Several Error Corrections
The article contains several wrong citations and links. I corrected some of them, which I came across. I am aware of the previous discussions in this lemma, and I made sure not to deviate form the conclusions drawn during these discussions. All I did was correcting obvious "publication bias" in THIS lemma.

This citation is wrong. Nowhere in STERLING (1959) does the following sentence occur


 * "the literature of such a field consists in substantial part of false conclusions resulting from type-I errors"

That might be a mix up between STERLING (1959) and SCARGLE (2000).

Instead, the following one is in the summary of STERLING (1959):


 * "the literature of such a field consists in substantial part of false conclusions resulting from errors of the first kind in statistical tests of significance"

I am therefore replacing the sentence.

The article also says that:


 * "The term was first used in 1959 by statistician Theodore Sterling to refer to fields in which "successful" research is more likely to be published"

This also wrong, as STERLING (1959) did NOT use the term "publication bias" – he discussed the concept. Therefore I am modifying this sentence to:


 * "The subject was first discussed in 1959 by statistician Theodore Sterling to refer to fields in which "successful" research is more likely to be published"

The reference to SCARGLE (2000) is wrong, I corrected it and made sure the broken link works again.


 * Was: Journal of Scientific Exploration. 14 (2): 94–106.
 * Is: Journal of Scientific Exploration. 14 (1): 91–106.


 * https://www.scientificexploration.org/docs/14/jse_14_1_scargle.pdf

And I made sure the term "file-drawer" matches the writing of the cited authors ("file-drawer" vs "file drawer").

C.wolke (talk) 12:03, 11 May 2019 (UTC)

Publication Bias In Situ
Publication bias in situ relates to choices on how to analyze the data, and which results to report.

https://pubmed.ncbi.nlm.nih.gov/15296515/

Shouldn't the article have a section on this issue? Zvi Zig (talk • contribs 18:44, 18 July 2022 (UTC)

Wiki Education assignment: BIOL4410 senior seminar
— Assignment last updated by Kitterbitter160 (talk) 00:11, 19 October 2023 (UTC)