Wikipedia talk:Scientific citation guidelines

Archives

Archive 1 Archive 2

This page has archives. Sections older than 60 days may be automatically archived by Lowercase sigmabot III when more than 4 sections are present.

There is a request for comments at [ Wikipedia talk:Identifying reliable sources (medicine)#What does MEDRS cover? ].

At issue is whether the lead paragraph OF WP:MEDRS should remain...

"Wikipedia's articles are not medical advice, but are a widely used source of health information. For this reason it is vital that any biomedical information is based on reliable, third-party, published secondary sources and that it accurately reflects current knowledge."

...or whether it should be changed to...

"Wikipedia's articles are not medical advice, but are a widely used source of health information. For this reason it is vital that any biomedical and health information is based on reliable, third-party, published secondary sources and that it accurately reflects current knowledge."

This has the potential to change the sourcing policy from

I made a minor modification in Convex curve. This article is now citing more than once a book (by Alfred Gray) which is also available online. The point is that the online version has one PDF file by chapter. Do we have to make one link to each PDF that is actually used, or is it correct to provide a link to the URL on which all the PDF can be downloaded (as I did) ?

Laurent.Claessens (talk) 12:57, 13 April 2016 (UTC)[reply]

As a general principle, I try to arrange the reference so that it is as easy as possible for someone to go look up that reference and verify the passage referenced. That would argue for linking to the actual PDF; right now I cannot tell which PDF out of the 60 or so that p.163 occurs in. --Mark viking (talk) 16:35, 13 April 2016 (UTC)[reply]

I in general like to link to the HTML "landing page" that links to the PDF, if one is available. If the HTML page lists multiple PDFs, you must be clear about which one is the correct one, perhaps by providing a chapter number. Linking directly to the PDF is permissible too. Antony–22 (^talk⁄_contribs) 20:15, 13 April 2016 (UTC)[reply]

A suggestion: pull all of your citation templates into their own "Sources" section, and then use {{

Harv}} templates to create short cites that link to the full citation. One advantage of this is you avoid duplicating full citations (as has already been done in that article), and another is that it is easier to make all citations consistent. (The duplicated citations are not consistent.) Then each citation of a particular chapter can use Harv's |loc= parameter to create a link to that chapter. Note: you could also link to Google Books, which gives the reader direct access to some of the pages.~ J. Johnson (JJ) (talk) 22:05, 13 April 2016 (UTC)[reply

]

Where did this idea that general references should be placed at the beginning or end of paragraphs come from? That just seems like more micromanaging of editors to me. It may not be appropriate to place a general reference inline at all, let alone in a specific position in the paragraph. If the idea is to identify general references to the reader, then a much better approach is to list them under a specific section, or a subsection of the bibliography if the article has one. SpinningSpark 19:13, 10 January 2017 (UTC)[reply]

For one thing, the

WP:DYK rules require that each paragraph have at least one inline citation. So if an article is going to be linked at DYK, giving a general reference inline is a good idea. —David Eppstein (talk) 19:26, 10 January 2017 (UTC)[reply

]

I don't think DYK should be driving Wikipedia guidelines. In any case, it does not say anything about general references, and it only says one cite per paragraph as a rule of thumb. It does not say anything about positioning in the paragraph. SpinningSpark 20:50, 10 January 2017 (UTC)[reply]

A number of articles on planetary science or discoveries by spacecraft cite press releases or media reports, rather than the actual journal publication. (I.e. stories about the published paper, rather than the paper itself.) Is there a guideline about which is preferred for Wikipedia? Fcrary (talk) 22:53, 19 May 2017 (UTC)[reply]

This page seems out of step with some of our current editing practices. Reading it it seems to discourage, or at least minismise, the need for inline citations. Things like However, in many articles it is cumbersome to provide an in-line reference for every statement don't really gel with current practices. It also really reads like a personal essay than a how-to guide. AIRcorn (talk) 23:52, 18 January 2020 (UTC)[reply]

I agree. Featured articles, which are articles of the highest standard, have multiple inline citations, and almost all information is referenced and cited. Veritas cosmicus (talk) 01:08, 21 January 2020 (UTC)[reply]

DYK standards require at least one inline citation per paragraph (for scientific articles as for anything else except fiction plot summaries, not counting paragraphs such as those in the lead that summarize more detailed later content) and I think that is a good place to set the line for a minimum level of sourcing to meet best practices (better than going by what FA does, as very few articles will ever make it to GA status). One-paragraph stubs may be able to get away with just having references at the bottom but I wouldn't want to see that for anything longer. I agree with Aircorn that this does seem out of step with current editing. I don't think it can be brought back into step with small edits; the difference between current practice and what is described here is too great. And I don't see any great need to have separate standards for scientific articles (except

WP:VPP, not just by consensus of people who might happen to have this watchlisted. —David Eppstein (talk) 01:16, 21 January 2020 (UTC)[reply

]

@David Eppstein and Veritas cosmicus: It looks like not many watch this page, which may explain a few things. I am happy to start a RFC on making it historical. It will at least bring other editors in. AIRcorn (talk) 04:21, 27 January 2020 (UTC)[reply]

@Aircorn and @David Eppstein, should we get back to this? I'm concerned about its promotion of academic scientific priority. I don't want people citing outdated papers to "give credit" to someone; I want them to WP:MEDRS#Use up-to-date evidence. Or should we try to re-write it? I think only about a quarter of it would require removal or very substantial re-writing. WhatamIdoing (talk) 08:00, 10 March 2024 (UTC)[reply]

I on the other hand do not want Wikipedia articles to promote incorrect folk theories of priority in the face of clear earlier precendents. But as I said already four years ago, I think this guideline has fallen by the wayside and should be marked historical. That would take care of it saying or not saying whatever it should say or not say about the purpose of citations, more definitively than trying to update it. —David Eppstein (talk) 08:22, 10 March 2024 (UTC)[reply]

None of us want inaccurate information in an article. I'd prefer "discovered by Sal Scientist^{[modern source]}" to "discovered by Sal Scientist^{[original paper]}". The original paper doesn't actually support a statement like that (it proves only that Dr Scientist wrote about it, not that it was first). WhatamIdoing (talk) 18:22, 10 March 2024 (UTC)[reply]

But it's important to include it so that people can verify for themselves that Sal's paper says what we are saying it says, and find out in what terms she expressed it. That doesn't mean omitting the modern citation; we can cite both. —David Eppstein (talk) 18:31, 10 March 2024 (UTC)[reply]

Really? If you have a (truly) reliable source that says Sal discovered _____, then you also need Sal's original paper? Do you expect reliable sources to be unreliable? WhatamIdoing (talk) 19:17, 10 March 2024 (UTC)[reply]

I think it's helpful to readers to be able to READ the sources we are telling them about, rather than telling them "there's a source about this but I'm not going to tell you how to find it". There are lots of reasons they might want to read it. Verifiability is one. Just in the past couple of days I found myself going back to the 1880 original source for a mathematical concept to find additional citations for claims elsewhere in an article that were not already clearly cited. Being able to go read that source was helpful, for verifiability, exactly the main purpose for including citations. Not including that source and just saying who invented it with other sources would have made the citation much more difficult to track down. Searching Google Scholar for the author's name gives over 400 hits. —David Eppstein (talk) 19:41, 10 March 2024 (UTC)[reply]

99.7% of the time, readers don't want to read any ref, much less one behind a statement like "Leukemia was first described by anatomist and surgeon Alfred-Armand-Louis-Marie Velpeau in 1827. A more complete description was given by pathologist Rudolf Virchow in 1845". And if they did read it, the paper would have solely historical value. The original papers cannot verify the primary claims in either of those two sentences (i.e., "first described" and "more complete description"), and they have no medical/modern information value at all.

Having a citation to Velpeau's 1827 "Sur la résorption du pus et sur l’altération du sang dans les maladies" case study also doesn't mean that they can read it. The source may be offline-only or paywalled.

It sounds to me like your reason is that a tiny minority of readers, specifically in the field of mathematics (which does not have such a significant problem with old papers becoming outdated), might occasionally be saved a quick trip to their favorite search engine if we add citations to historical papers. However, reader convenience isn't mentioned in this page. Instead, it has a whole section dedicated to "giving credit to the original discoverer".

I would not ban including original papers, but I generally prefer to put them in a

WP:Directly support the content they're following. WhatamIdoing (talk) 20:33, 10 March 2024 (UTC)[reply

]

Historical value is value. It sounds to me like you have in mind a very specific use case of readers and are vastly overgeneralizing to the idea that all readers are like that (like you?). —David Eppstein (talk) 21:09, 10 March 2024 (UTC)[reply]

Do I? Well, here's the citation:

Velpeau, Alfred-Armand-Louis-Marie (1827). "Sur la résorption du pus et sur l'altération du sang dans les maladies". Revue Médicale Française et étrangère (in French): 216–240.

What kind of value would you get out of that? What kind of value can you imagine anyone getting out that, that they wouldn't get by spending a minute or two with their favorite web search engine? (That's how I found it, so presumably other people can, too.) WhatamIdoing (talk) 03:36, 11 March 2024 (UTC)[reply]

I don't know what value you get generalizing from one example. To make it two, here's the citation that I did get value out of, directly about verifying the content of an article and improving its sourcing:

JSTOR 2369261. I'll refrain from speculating about what might be different about these two cases, because the sample size is still too small to be meaningful. —David Eppstein (talk) 07:00, 11 March 2024 (UTC)[reply

]

The rule has been written very generally, to encompass not merely mathematics but also subjects like medicine that have changed dramatically. Based on this conversation, I suspect that antique papers primarily (only?) have relevance in mathematics, and that any advice we give should be specific to mathematics. WhatamIdoing (talk) 16:43, 11 March 2024 (UTC)[reply]

From a sample of two, you can infer that one of the two is a very field-specific special case that needs an exceptional carve-out, and the other one is the general case that covers everything else? Your powers of inference are more discerning than mine. —David Eppstein (talk) 17:35, 11 March 2024 (UTC)[reply]

From a sample of two, plus pre-existing knowledge that the advances in the hard sciences, applied sciences, and social sciences – but not the formal sciences, whose knowledge is not technology-dependent – has invalidated or radically transformed nearly all of the work done before the 20th century (and a good deal of the 20th-century work, too).

Consider computer science: Babbage's difference engine is a wonder of the computing world. Do you need to read the original publications about it to understand anything about computing in the present century? To understand trigonometry? About how to build a calculator? So far, we have established that there is possibly some value to seminal antique papers in the formal sciences. We have no evidence that this is true outside the formal science, and no clear reason to believe that all sciences are the same as mathematics.

In fact, I suggest that the value of original papers is more significant outside of the sciences – in philosophy and the humanities, than in chemistry and biology. WhatamIdoing (talk) 18:36, 11 March 2024 (UTC)[reply]

I disagree strongly with the first posting. However, in many articles it is cumbersome to provide an in-line reference for every statement is imho exactly on point and should not be changed. I can't see how it would out of step with current editing practices. Featured articles are hardly representative for current editing practices. The idea of sourcing every single statement (as in every sentence or half sentence) is rather cumbersome and often impractical in particular for human written texts and creates a ridiculously bureaucratic hassle for authors/editors. --Kmhkmh (talk) 19:52, 10 March 2024 (UTC)[reply]

I think it might depend on how one understands "every statement". There is a sort of

WP:MINREF material WP:Glossary#cited

.

At the moment, given that this page is marked as a guideline, I understand that all featured articles for scientific topics are required to comply with its advice. WhatamIdoing (talk) 20:37, 10 March 2024 (UTC)[reply]

Sure, but in my experience/perception we probably get the first case, when we change the guideline and I really don't see that from the perspective of featured articles, but the bulk of regular articles and authors working on them, where imho a paragraph based citation is good enough in most cases. The last thing we need there, is bot runs and wikignomes formally tagging articles as insufficiently sourced due paragraph based citations or needlessly turning paragraph based citations into line or sentenced based citation (and doing that potentially without really reading sources and content). It will in doubt just create more "bureaucracy" driving editors away.--Kmhkmh (talk) 00:44, 11 March 2024 (UTC)[reply]

There is certainly a lot of that going on. I ran across an article on my watchlist that I hadn't looked at in years. There were about 25 consecutive edits in 25 consecutive months without even half a sentence being added or removed. It was all trivial copyediting and formatting. High-volume editors rarely seem to write content. WhatamIdoing (talk) 03:30, 11 March 2024 (UTC)[reply]