Wikipedia talk:Primary Secondary and Tertiary Sources/Archive 5

Archive 1Archive 3Archive 4Archive 5

Peer Reviewed Publications Should Always Treated As Reliable Secondary Sources

WP:Sources states

In general, the most reliable sources are peer-reviewed journals and books published in university presses; university-level textbooks; magazines, journals, and books published by respected publishing houses; and mainstream newspapers. As a rule of thumb, the greater the degree of scrutiny involved in checking facts, analyzing legal issues, and scrutinizing the evidence and arguments of a particular work, the more reliable it is.


Clear enough, but I've seen wikilawyers use the claim that peer reviewed studies are a "primary source" to oppose and block reporting of studies in Wikipedia, arguing that text books or even newspaper articles covering the study (if they exist) are preferable because these are secondary sources. This has the effect of closing the door to reporting any newer studies which have not been included in text books or in media reports. This "block" thereby limits the scope of the encyclopedia as a comprehensive summary of known information on a subject.

I therefore believe the defininition of a secondary source should be clarified to include all peer reviewed journal articles, except in those rare circumstances where the article has been retracted by the journal...but even in those cases it may be especially important to report that the article was published and retracted.

Further reasoning is as follows:

1. Lab notes and raw data are clearly primary sources. Published raw data collected by national regisitries, surveys and the like are also primary sources.

2. Once these notes and data (whether collected by an indivdiual research or third party compilations of data) have been analyzed and intepreted by a researcher, or team of researchers, and this synthesis and analysis and review of data and literature has itself been reviewed by peers with expertise in the field and published in a peer reviewed, indexed journal, that publicaion is now an reliable secondary source, by definition. Afterall, the information has been vetted by experts and considered at least worthy of consideration as having added information useful to the field. That is roughly speaking the standard and purpose of peer review. While peer review is sufficient, I add "indexed journal" as an additional level of quality verification since the academic journal indexing houses verify that the journal, even a new journal, has editorial processes worthy of the journal being indexed.

3. A core policies are reliability and verifiablity...not truth. That some wikipeida editors do not find peer reviewed studies convincing does not justify wikilawyering efforts to block their inclusion in articles. Peer reviewed sources are and should remain highly valued and not subject to suppression by ideologically driven editors.

I've read previous discussions of this subject and believe the policy is currently correct, but it is frequently being misinterpreted and should therefore be clarified. --SaraNoon (talk) 17:03, 9 August 2008 (UTC)

The devil is in the detail. For example articles in old peer reviewed journals are not as useful as more recent peer reviewed journals, but some very recent articles in peer reviewed journals may be rejected by most scientists working in that field in the coming year or two. This paragraph has to be read with common sense and trying to make it more detailed will probably detract from its usefulness as a policy. I do not know if such points have been added to a guideline, but that is probably where such details should be put. -Philip Baird Shearer (talk) 17:35, 9 August 2008 (UTC)
Peer review does not guarantee reliability in general. If you and your flat earth society peers agree that the earth is flat, that does not make "the earth is flat" a reliable statement. Journals associated with notable professional societies are generally a reliable source, but it really depends. The acid test is essentially if articles from the journal are routinely cited in other journals, especially broadly recognized ones. SDY (talk) 17:52, 9 August 2008 (UTC)
Flat earthers do not get their "studies" published in peer reviewed, indexed journals of geology. So that's a red herring argument. I added the provision also that the journal must be indexed by the indexes covering the field of the subject. If you're adding that the journal must be cited by other journals in the field (proving that it is accepted by other journals in the field as a journal, not a basement publication of flat earthers) I'm fine with that.
For wikipedia, "reliability" does not apply to whether the claimed fact is true but whether it was published in a reliable source -- which peer reviewed journals are, by defintion. If for example, the wikipedia article and source accurately report that an article in Flat Earthers Review" states that the earth is flat, that is a true statement (that they claim it, not that it is flat). I'm all in favor of having facts and analyses opposing a postion of one source included in articles. What I object to is proponents of one view deleting peer reviewed material simply because it doesn't agree with their view.
Regarding old and newer studies, again if they are peer reviewed and an editor wants to include them, I think that's fine and those edits should be respected. On the other hand, that doesn't mean that contributor should object when other editors put his sources in perspective by adding material showing that the analyses in the older cite have been supplanted by newer research or that the latest new research has been questioned by some experts (giving citations, of course.) My general view is that the more information and citations given the more valuable the article is to those starting to research a subject. As noted above, my main concern is with the wikilawyering of some editors who try to twist these common sense definitions into excuses for deleting material from reliable sources, ie. peer reviewed journals, because they are more concerned about pushing a perspective or agenda of their own rather than letting Wikipedia be a record of the literature...and even the back and forth disputes...of the experts in the field as recorded in the peer reviewed journals. --SaraNoon (talk) 18:08, 9 August 2008 (UTC)
If it is a real journal, I agree with you totally. The only qualification given, though, was that it was peer reviewed, which means nothing if the quality of the peers is dubious. SDY (talk) 22:39, 9 August 2008 (UTC)
Agreed in principle, but drawing from original articles which are highly technical has problems. Reviews can also be technical, though, so that's not something that should be directed entirely at original articles. Reviews are, in many cases, a better source, but reviews are often less common than original articles, and, in some cases they are more prone to error, since the reviewer cannot be an expert on everything she is looking at. II | (t - c) 19:35, 9 August 2008 (UTC)
I agree with and appreciate your points that (1) review articles are often less common, or perhaps even unavailable, and (2) that review articles may have their own limitations, errors, omissions, and POV.
Regarding the concern that some papers may be very technical, I don't see that as a core problem. If the wikipedia aritlce deals with something which is nuanced and technical the readers are looking for some such details. What is very important, however, is that each editors should know his or her own limitations and not try to summarize something they don't understand. At the same time, all editors should generally respect that some, perhaps even many editors, are contributing to the article precisely because they do understand the technical details and are qualified to reflect the findings of a paper, generally by quoting the original author, et cetera. That doesn't mean an expert editor should engage in OR, but it does mean that an expert editor should not be deprived the right to cite a peer reviewed study just because another editor objects to the study because it presents material that goes against some preferred POV.
While policy is that the burden of proof is on an editor to supply a reliable source (in this case a peer reviewed study), once that is done the burden of proof for changing (much less excluding) the contribution falls on editors who believe the contested contribution mistates or misrepresented the peer reviewed source. If there are misrepresentations, the solution in such cases is to correct the article to reflect what the peer reviewed source really says. What I really object to is people who just delete reliably cited material with excuses in the vein of:"I don't believe it," or "I've found another expert who disagrees," or "You don't have the approval of myself and other editors (consensus) for making this addition to the article."
It is the unwarrented deleting of reliable sources which underlies much of the edit warring around here. It is my hope to see peer reviewed sources more firmly identified in policy as reliable secondary sources to reduce the risk of such edit warring and to "force" those disagreeing on a citation to work at ways to properly report the findings or fact without deleting the citation itself. How a reliable source is reported, and in proper balance with other sources, is certainly a fair editorial decision that requires consensus. The fact that a peer reviewed source is a reliable secondary source should not be an issue, however. --SaraNoon (talk) 20:05, 9 August 2008 (UTC)

We do say that peer-reviewed journals etc. are the most reliable sources. The problem may lie in the inclusion of "published experimental results by the person(s) actually involved in the research" under the subheading of "primary sources". This would include results published in peer-reviewed journals, given the present wording. I can see the concern that published experimental results are sometimes found to have been in error, or prove irreproducible by others; so there may be good reasons for caution. On the other hand, if some groundbreaking research results are published in Nature, or some other mainstream, peer-reviewed scientific journal – say, the discovery of a new particle or element – are we really saying that we should not mention this in Wikipedia? And what about sociological studies, for example, published by the sociologist who has done the fieldwork? Surely, such studies are reliable sources, certainly more reliable than newspaper articles.

Where did you have these problems, SaraNoon? Which scientific fields are involved? Jayen466 22:58, 9 August 2008 (UTC)

Peer-reviewed sources are generally primary sources and are always subject to the WP:PSTS issue. ScienceApologist (talk) 04:42, 10 August 2008 (UTC)

  • That's an interesting view, given that we have "In general the most reliable sources are peer-reviewed journals and books published in university presses ..." prominently displayed in policy. Could you explain? Jayen466 18:13, 10 August 2008 (UTC)
PSTS is very carefully worded so that only the following parts of peer reviewed papers are classified as primary sources:
"written or recorded notes of laboratory and field research, experiments or observations, published experimental results by the person(s) actually involved in the research;"
Any parts of that make analytic or synthetic claims are classified as secondary sources in PSTS. This is a reasonable division, in my opinion. — Carl (CBM · talk) 18:24, 10 August 2008 (UTC)
I disagree with ScienceApologist for exactly the reason raised by Jayen. I also agree with Carl, and underscore that nearly every peer reviewed paper that includes "published experimental results" (for example) also include analytical and sythentic claims and some review of the literature. So in many if not most cases where a peer reviewed article contains primary source material it also includes secondary source material. It is therefore appropritate as a source, but editors should take care in using any of the primary source material (ie. a data point, such as the melting point of zinc) to not draw inferences or make their own analysis, but it is clearly fair game to summarize (or better) quote the analysis of the published researcher as that is from the secondary source portion of the paper.--SaraNoon (talk) 15:57, 11 August 2008 (UTC)

If a peer reviewed article is later challenged by another such article, per NPOV we ought to mention both, in particular when it concerns scientific disputes. WP:V and WP:OR do not preclude the inclusion of competing views, on the contrary. ≈ jossi ≈ (talk) 23:49, 10 August 2008 (UTC)

among the problems

One is the low quality of many so called peer reviewed sources, and the other is the very high quality of some that are not. I have reviewed for journals where the clear expectation of the peer reviewers was to find some reason for accepting anything that might be submitted from anyone in the industry or profession or academic field, and the only things ever rejected were off the wall stuff from people without any familiarity at all with the subject. -- and even then, if they were short of papers for an issue, this was interpreted pretty flexibly. Some "peer-reviewed" journals in such cases publish the proceedings of symposia, under the authority of the symposium convener--who in turn sometimes has their main purpose the need of getting enough papers to hold a symposium. I have also reviewed for journals in the position of accepting the top 5% of papers , or only those papers which in addition to being excellent, are considered to be of broad interest. But even these journals of the highest quality always have a few papers that it turns out nobody else even bothers to cite again--and even the lowest quality journals sometimes publish a paper that turns out to be of major significance--sometimes where one of the editorial board is induced to submit some actually good work to help the journal along.
On the other hand, some non-peer reviewed sources--in my subject, Library Journal, in science Chemical and Engineering News, and Physics Today--as examples only--are technically professional magazines, where decisions are made by the editors of the journal, responsible in some cases to professional societies, in some even to commercial publishers. Such journals typically contain state-of-the-art summaries that are in practice treated by academic and professionals to be of the highest authority and are very widely used as standard treatments. I've written for one or two such journals, and the editing was the most careful I ever received.
I think one must evaluate individual cases. I have seen here at Wikipedia many instances where a person with a lack of sophistication attempts to override the consensus of the academic world on an academic subject with popularized sources that at best can be used for showing what the popular misconceptions of the subject may be. In the other direction, have also seen people using isolated peer reviewed journal articles to support the most dubious ideas. Even more, I have seen people selectively picking out unrepresentative peer-reviewed sources for quotations to justify a particular not-universally accepted position. In a controversy, there's nothing easier to find than a distinguished scholar in an unimpeachable journal saying that his position is widely held,and everyone else has been refuted. In anything interesting , one can do this for both sides of the question.
With respect for the weighting of primary articles in the scientific literature versus review, Sara is generally correct that there is no simple rule. Many primary articles provide an encyclopedic evaluation of the prior literature. Many review articles are purely POV, published because of a friend who's the journal editor. Though many review articles are themselves peer reviewed , sometimes quite heavily, they remain to be understood in most cases as being the personal view of the people who write them. To illustrate the lack of separation, PubMed (and other indexes) regularly class as review articles any article with above a certain number of references, on the empiric grounds that it can be used by someone seeking a review of the literature. DGG (talk) 22:36, 9 August 2008 (UTC)
I have to ask... Why is this being discussed here? How does all this relate to No original research? No matter where a paper (even one supporting a completely rejected theory) is published, it isn't OR ... unless the author of the paper is the person who is adding it to Wikipedia. If it has been published elsewhere, it can not be original research to state what it says here. Doing so might violate several other policies... but not WP:NOR. This sounds like a topic that would be better discussed at WP:RS or WP:V. Blueboar (talk) 01:21, 10 August 2008 (UTC)
true, it really belongs in RS. However, there is actually no such thing as simply following the sources to provide V. In anything really interesting and contentious, you can find sources to support almost anything, and the judgement as discretion to select them is very close to original research in many cases. To provide one's own synthesized opinion is not actually any worse than to find from among 20 opinions the one that coms closest that one wants to say--it is just a less wikipedia=worthy way of giving one opinion. either can be honest NPOV work, either can be POV-pushing. The way the theme got to this was i think the discussion of whether reporting primary published scientific papers was OR as distinguished from citing reviews. RS or no RS, the proper use of sources is not intellectually different from much OR.DGG (talk) 02:09, 10 August 2008 (UTC)
From the WP:SYNTH aspect of OR, you mean? I find SYNTH, as presented, is easily misinterpreted -- giving rise to the problems you discuss.--Father Goose (talk) 22:19, 10 August 2008 (UTC)

@DGG" Though many review articles are themselves peer reviewed , sometimes quite heavily, they remain to be understood in most cases as being the personal view of the people who write them. ... which makes these personal viewpoints perfectly acceptable in Wikipedia articles if properly attributed. ≈ jossi ≈ (talk) 23:52, 10 August 2008 (UTC)

Why Peer Review/Secondary Sources Discussed Here

Regarding the thread above, I brought it up here because I've seen WP:PSTS, which is part of No OR, cited by wikilawyers as an excuse to exclude any mention of citations -- even from major journals. They simply argue, as ScienceApologist did above, that peer reviewed journal articles are generally primary sources.

This conflict over how peer reviewed sources (whether from a fantastic journal or a "any opinion goes" journal) is therefore a source of ongoing of editing conflicts and edit warring. Clarifying this definition would help to reduce such editing conflicts. In my view, this clarification should be done in a manner which encourages inclusion of more material and tries to discourage exclusion of material.

In short, I can see why the definition at WP:PSTS can be interpreted in a way that conflicts with WP:Sources. Therefore, WP:PSTS it should be clarified to reiterate that peer reviewed sources are highly prized as verifiable sources with an explanation that they often combine both primary and secondary material. In that regard, editors should be reminded that any use of the primary material (raw data) should be used with caution and not with OR, and that it is generally best to rely on the scondary source materail from the peer reviewed source, which is usually in the form of the textual description of results and discussion. The "background" sections are also valuable secondary sources as they often provide a review of the relevant literature undergirding the new paper.

Some of these points should be developed in guidelines explaining this policy. But as this is a policy page, it should be clarified in a way that reduces conflicts between interpreting WP:PSTS and WP:Sources.

Regarding guidelines, as stated above, I believe in inclusion rather than exclusion. If some editor thinks that a third rate paper from a third rate journal should be cited in an article, I would be all for keeping the citation, but might insist that it be just that, an additional citation associated with better cites to the same point raised. After all, even a third rate paper is verifiable and rather than edit war, it is best to find at least a nominal place to include it out of respect for the other editor's insistence that it should be included.

My main point then is not that every peer review paper deserves equal weight or discussion. Not at all. But I do object to POV pushers deleting peer reviewed papers simply because the findings and opinions expressed conflict with what the editors insist (usually without sound basis) is the "accepted" viewpoint....often relying on well known newspaper or magazine articles as "reliable secondary sources." In my view, peer reviewed sources should always be more highly preferred than newspaper articls...unless you are looking for a synthesis of what so and so said at this political rally, of course.--SaraNoon (talk) 16:22, 11 August 2008 (UTC)

--SaraNoon (talk) 16:22, 11 August 2008 (UTC)

How about this as a draft proposal? Perhaps someone can tighten it up.
Secondary sources are accounts at least one step removed from an event.[3] Secondary sources may draw on primary sources and other secondary sources to create a general overview; or to make analytic or synthetic claims.[4][5] Peer reviewed studies including original data are one step removed from the gathering of the data in that they generally provide an analysis and synthesis of the data along with an expert opinion, all of which has been reviewed by third parties. The raw data in a peer reviewed paper, however, should be treated with the care due to primary source material. SaraNoon (talk) 16:29, 11 August 2008 (UTC)
I think these matters are too subject-specific to belong in policies. The fact of the matter is, journal articles can be primary sources in some fields and contexts and secondary in others. Moreover, journal articles are not the only example of peer-reviewed sources. A one-size-fits-all decision in a policy page would simply not be appropriate on such a matter. Making a universal rule based on evidence taken from a narrow context leads to mistakes, in policy-making as in science. Best, --Shirahadasha (talk) 04:03, 12 August 2008 (UTC)
The current WP:PSTS statement can be read in a way that contradicts reliable source policy and is therefore a cause of much edit warring. Some clarification, including perhaps additional guidelines, would benefit the project. Peer reviewed sources, whether in journals or elsewhere, should be definition be recognized as reliable, verifiable sources which for the most part are also reliable secondary sources. Policy should reflect this and put the onus for excluding any particular peer reviewed source on those arguing for exclusion, not on those bringing material to the project. Such a clarification would emphasize the goal of including material so that the arguments will be limited to how they are included rather than if they should be included.--SaraNoon (talk) 17:48, 12 August 2008 (UTC)
The onus of providing reliable sources has long been with those who wish to add material, not with those who may question it (see WP:V). By extention, when a conflict occurs about a source (such as whether the source is infact reliable, or whether a statement being derived from the source violates NOR), it is up to the person who wishes to add the material to demonstrate that the material conforms to our policies. Blueboar (talk) 18:30, 12 August 2008 (UTC)
SaraNoon, this has been a longstanding weak-spot within the policy article, and has been resistant to change, but most people except Wikilawyers ignore it. The distinction between primary and secondary is a subtle one derived from historiography, actually all sources are both primary and secondary in some sense. So don't worry about reading the section too closely--rather, follow the general principles of the article, which is to avoid original research in any form, and make sure everything is verifiable, which means that you use reliable sources. If your source is reliable, and you cite it fairly without synthesis, it isn't original research. Period. Don't worry about whether a source is primary or secondary, because the answer is, it's both. COGDEN 21:01, 12 August 2008 (UTC)
Shirahadasha, could you please give an example of a journal article published in a mainstream, peer-reviewed academic journal that you would consider a primary rather than a secondary source – just to help understanding and communication here around this issue. Jayen466 18:24, 12 August 2008 (UTC)
You and SaraMoon may wish to read over this discussion, as it is fairly typical of discussions about peer-reviewed journals and scientific articles that have repeatedly occurred on this page. Vassyana (talk) 23:38, 12 August 2008 (UTC)
coGDEN, I agree with you that it should be clear enough already for reasonable editors, but the fact is, as you noted, it is a weak spot and is being used for wikilawyering and obstruction and disruptive edits by "tag teams" are especially difficult when they resort to this "you can't use that source because it's a primary source" nonsense. Obviously, this is especially problematic on any article that is even remotely controversial where one editor, or teams, want to 100% suppress verifiable sources simply because they don't like the evidence or conclusions of those sources. Rather than go to dispute resolution constantly over these matters, at least some edit warring could be avoided and people could focus on how to include sources rather than if sources should be included, if this was properly clarified in policy. I think my suggested change comes close to tightening this up. If it is not clarified in policy, it should at least be clarified in some higher level guideline. Would you support some kind of tightening up?
One of my pet peeves is that a truly comprehensive encyclopedia would give readers tons of citations to follow-up on. Even if a cite was not worth discussing, if an editor believes it should be included and can add the cite to an appropriate passage, other editors should respect that instead of deleting it because it leads to an article, evidence, and synthesis that undermins their POV. But I've seen open edit warring just to keep a disfavored cites (much less discussion) out of an article even though they are cites to major peer reviewed journals. Lack of clarity on this issue leaves room for disingenous efforts to block reliable and verifiable information from having any place in Wikipedia. That's a crying shame and a real disservice to the project.
Reading Vassyana's recommended log of a previous discussion, I see the argument raised that peer reviewed articles should only be cited in a manner that conforms with some other secondary review. For example, one editor writes: "I am proposing that journal articles should not be cited in a manner that directly contradicts the findings of more secondary sources." In other words, one editors preferred secondary sources can be used to veto or silence peer reviewed journal articles cited by other editors. Even if the latter are newer? Or if the preferred secondary source represents just one of many views presented in sedoneary sources. This is all fertile ground for edit warring.
A clear policy, "Though shalt not delete reliably cited peer reviewed articles (but you can clarify or cut back material from such sources" would help to force editors to respect the sources brought forth by other editors and work together to fit them in with some due weighting, of course, which is easier to hammer out once the relevent sources are all cited. --SaraNoon (talk) 21:32, 14 August 2008 (UTC)
Thanks Vassyana, that was helpful, and food for thought. Jayen466 22:17, 15 August 2008 (UTC)

Merge of Articles

Wikipedia:Classification of sources is a previous attempt to draft a guideline which should be reviewed and merged into this attempt.--SaraNoon (talk) 21:29, 24 August 2008 (UTC)

PSTS Confusion and Disagreements

I agree with DGG's classification of sources at Notability_(Geographic_locations), and I added a note to that effect there. Census reports, for example, are compilations of census forms.

But then I started looking at the sources listed in the notes (2 through 4) used in WP:PSTS I see the definitions and examples are all over the place. For example, this one lists Encyclopedia Britannica as a secondary source, while most put encyclopedias as tertiary.

Actually, I would actually agree that some encyclopedias, like Britanica, are compiled secondary soruces since they commission experts to write the articles, which are therefore essentially review articles by an expert who synthesizing his or her interpretation of a body of knowledge. In my book, I wouldn't object to WP editor starting an article using a Britannica source (though defined as WP as tertiary), though it is always better, in my opinion, to move sources back to peer reviewed scholarly articles whenever possible.

To me, in regard to definining the quality of sources for WP source material, tertiary includes not only distance from the original source but expertise. A newspaper article reporting what a scientists says at a press conference is a great secondary source documenting that scientist's statements. The same article providing a summary of what "most scientists believe" or what "most research shows" is reflecting a tertiary view, expressing the view of a reporter with limited expertise who has most probably looked at only a few sources and spoken to a few scientists. But a Britanica article by an expert, while in a compiled encyclopedia, is closer to being a reliable secondary article.

And there is part of the rub. Primary, secondary and tertiary vary depending on use, authorship, ane even intent.

  • Gettysburg Address, primary. No dispute
  • Contemporary newspaper articles about it, secondary.
  • But those same 1864 newspaper articles about the Gettysburg address are treated as primary sources when used by a modern scholar to write a dissertation on media treatment and public reactions to Lincoln and the Civil War.
  • And even that modern scholar's papers will be treated as primary source material two hundred years from now when scholars are investigating the 21'st century's understanding of Lincoln.

Another source of confusion: many sources contain both primary, secondary, and even tertiary material. For example, peer reviewed science article typically includes introduction which will contain a literature review of studies related to the subject of the stdy and in a concluding section where the authors discussion, synthesize and intepret the results of the study. Whether the results reported in the tables are primary material (since this is the first reporting of the findings) or secondary material (since this is a synthnsized statistical analysis of raw data which is not reported, but is the true primary source) depends on one's preferences for defining primary and secondary sources. But the presence of primary material does not change the fact that the introduction and conclusions are secondary source material representing the peer reviewed opinions of an expert in the field.

The current NOR page contributes to much edit warring and wikilawyering because there is no universally clear definition for primary and secondary sources in all contexts and fields. In addition, it makes two "definitive" but (at least to wikilawyers) competing statements: "Wikipedia articles should rely on reliable, published secondary sources" and "Primary sources that have been published by a reliable source may be used in Wikipedia, but only with care."

A policy that breeds confusion and contention isn't very helpful. The main point of NOR is to stick closely to the source, whether one believes it to be primary or secondary. That should be the focus, not trying to define secondary or primary sources (where there is plenty of room for contention).

I will post a recommendation for clarifying policy in the near future. At the very least, the policy should state that secondary sources are preferred but primary can be used (provided they are used without adding interpretations) within the same sentence. But first, what are your thoughts? --SaraNoon (talk) 15:14, 20 August 2008 (UTC)

Actually, the Gettysburg address isn't a source at all, as it was simply a successsion of sounds that dissipated into the air. Primary sources would be Lincoln's written text, if it survives, & any notes taken by anyone who was present, whether such notes were written down at the time or later. Peter jackson (talk) 17:13, 20 August 2008 (UTC)
There are five copies written by Lincoln.[1] All would be primary sources. Other than that observation, do you generally agree with my assessment?--SaraNoon (talk) 17:22, 20 August 2008 (UTC)
As I see it, the problem this page is trying to address isn't the use of primary, secondary, or tertiary sources, but the misuse of them... there are situations where all three are appropriate to use in Wikipedia. If you are writing an article on the Gettiesburg address, it is appropriate to quote and cite the original text (the primary source)... to cite a noted historian's analysis of it (a secondary source)... and to note the date it was given by citing another encyclopedia (a teritary source). What would not be appropriate would be to include our own analisis of the address.
We get so wrapped up in definitions that we keep forgetting why they are there in the first place... to tell editors not to enguage in Original Research. If our examples are confusing the issue... let's remove them. Blueboar (talk) 17:54, 20 August 2008 (UTC)
SaraNoon, if we use the primary/secondary distinction, we ought to describe what primariness is, and what secondariness is, and how to tell the difference. I don't think we should classify types of sources (like peer-reviewed journals or newspapers) as one or the other, because whether a given source is a primary or secondary source depends on how it is used. For example, a contemporary newspaper about the Gettysburg Address could be either primary or secondary. It would be a secondary source for the text of the Address, but a primary source of any commentary or observations about the Address. (In this case, using the primary source for the Address would be far preferable to using some 19th century yellow journalist's version quoted from memory.) COGDEN 18:53, 20 August 2008 (UTC)
Yes, I would suggest dispensing with the definition lists and simply creating thorough explanations as to what constitutes a primary or secondary source, and how both should be used properly. Kaldari (talk) 18:59, 20 August 2008 (UTC)

I believe the SYNTH section is already far more important than PSTS, which has problems we're talking about. I'd like to move PSTS below "Citing oneself" without any other changes. Doing so not only helps to bring readers to the SYNTH section first, but may help alert those watching policy to look in on this discussion regarding rethinking the PSTS section to join in the discussion.--SaraNoon (talk) 14:44, 21 August 2008 (UTC)

No objections to SaraNoon's proposal on my part... but may I suggest that, before people attack this problem, they review the archives for this page. There have been several attempts to clarify or modify PSTS in the past (all met with a lot of resistance), and it I think it will help the discussion if everyone is aware of the arguments (for and against) that have been made before. Blueboar (talk) 14:57, 21 August 2008 (UTC)
part of the problem comes from the specialized use made of "primary source" in the experimental sciences, s compared to history or to literature. Strictly speaking, the primary source of scientific research is whatever record the observations--traditionally the laboratory or field notebook, now often the automated record of the experiment--sometimes even a physical specimen. But scientists use the term "primary literature" to refer to the published papers that result from analyzing the work represented by these experiments and calculations--all textbooks present it this way. And then they use secondary literature for review articles and advanced textbooks further evaluating and summarizing them. In terms of their use at Wikipedia , the difficulty is that to the extent that these "primary" articles contain facts that need evaluation they are not usable here except under special circumstances, such as the interpretation being uncontroversial or obvious, but to the extent they contain facts evaluated by experts they are usable here as reliable secondary sources. Unfortunately, the extent to which any one paper will be one or the other is subject to no fixed rule. DGG (talk) 18:48, 21 August 2008 (UTC)
I have reviewed at least much of the previous discussion. It only confirms to me that because there are so many nuances, the bulk of what is now in PSTS should be moved to a guideline status for PSTS so more examples can be presented and also so this "measure" will not be central policy which wikilawyers can misuse to trump commonsense. In place of this large section on PSTS I think a shorter section refering to the guidelines and re-emphasing the focus on NOR would make the policy both more brief and focused. But for now, I'd just like to try "demoting" PSTS to a lower section in the policy to spark more feedback while also not disrupting the existing policy.--SaraNoon (talk) 00:57, 22 August 2008 (UTC)
I’d suggest instead writing a new guideline “Interpreting WP:PSTS”. Agreed, there are so many nuances, and examples would be a good idea, but disagree that it should not be policy that, in simple terms, secondary sources must be used, supplemention with primary sources is OK. --SmokeyJoe (talk) 01:46, 22 August 2008 (UTC)
I agree with Joe, both that we need such a guideline, and also that for the ordinary sort of Wikipedia article, the simple policy as he words it is exactly the sort of basic instruction to give. DGG (talk) 23:17, 25 August 2008 (UTC)

Primary and tertiary sources

I've added a section for these types of sources, since they really do have an appropriate role to play in encyclopedia writing. DurovaCharge! 08:50, 3 September 2008 (UTC)

Thanks. I think that is a good contribution.--SaraNoon (talk) 19:35, 4 September 2008 (UTC)

Should use of primary sources be legitimised by citations of the primary source in existing secondary-source literature?

I like this draft. One suggested addition: I wonder if we should add something to the effect that, where there is a substantial body of secondary-source literature available, a primary source should only be cited if that same primary source is also cited and discussed in one or (preferably) several secondary sources. This is to avoid giving more weight to it than it has in the established literature. Jayen466 23:42, 7 September 2008 (UTC)

What would be the purpose of that restriction, and how would its limits be defined? Seems to me that the stipulation of uncontroversial facts covers potential abuse. If the primary sources conflict on a point, then don't cite them. Instead use secondary analyses by recognized experts. Or is there another angle that doesn't anticipate? DurovaCharge! 00:02, 8 September 2008 (UTC)
Here is one real-life scenario I've come across (not from English Wikipedia): A late 19th/early 20th-century European author had said somewhere, in some recorded speech, something uncomplimentary about a novel written by a black author. An editor wanted to cite this passage in the article on the European author to give the impression that the author was a racist. The secondary literature characterised the author, based on his overall oeuvre, as a figure who was ahead of his time in supporting the modern notion that all races were equal. The generalised scenario here is an editor finding a passage in any author's work or recorded correspondence that they strongly agree or disagree with, and wishing to cite that in the article. My contention is that this should not be done if (1) there is a substantial body of secondary literature on the author, and (2) the passage has not been an object of interest in that body of literature. Our article should mirror the primary objects of interest of the existing secondary literature.
Another, hypothetical example might be academic studies reporting evidence that Transcendental Meditation or Sahaja Yoga have measurable benefits for certain medical conditions, say arthritis. While a mention of these studies may be appropriate in the articles on TM or SY, I would argue that they should not be mentioned in the article on arthritis, unless the results reported have caused significant interest in the existing literature on arthritis.
To sum up, I want us to think about whether any due-weight justification for the inclusion of a primary source in an article should be based on secondary sources attributing significance to that primary source. Jayen466 00:42, 8 September 2008 (UTC)
User:MastCell gave a couple of real-life examples here which I think described a similar concern. Jayen466 01:07, 8 September 2008 (UTC)
That's a very good consideration to be raising at this time. I'd definitely exclude that under WP:NOR and the exclusion against interpretive use of primary sources. Yet I wouldn't object to raising something like that as a specific example. No matter how we structure this, people are going to try to tiptoe around that clause. DurovaCharge! 01:38, 8 September 2008 (UTC)
In general, I disagree with Jayen that we can define this well enough in policy, as you still face the difficulties regarding defining primary and secondary material. Any such discussion belongs in guidelines, rather than policy. Taking Jayen's, arthritis example, if a new peer reviewed study finds and reports and comments on a link, that observation and commentary is from a reliable, peer reviewed source and inclusion is not original research, it is keeping an article up to date. Attempts to exclude the material until it is peer reviewed a second time at some future date when someone publishes a review of the literature poses an unnecessary obstacle to reporting the facts and opinions reported. While editors can and should debate how the material should be included, policy should not be shaped to put an undue burden on those supporting inclusion.--SaraNoon (talk) 16:54, 8 September 2008 (UTC)
Yes and no. If it is a new study, published by a leading scientist in a first-rate journal, there may be some wiggle room – though I am sure if it's important, there will be immediate coverage of it in journals like Nature, New Scientist etc., providing a rationale for justifying its inclusion. If on the other hand, it is a 20-year-old study that appeared in a third-rate journal and has "sunk without a trace", i.e. without eliciting further comment in the literature, or if it is an isolated passage in a letter Goethe wrote several hundred years ago that has not attracted published attention, it is different. I accept it is difficult; all the more reason to think hard about it. Perhaps a judiciously worded reference to observing "due weight" in the citation of primary sources may do. Jayen466 02:23, 9 September 2008 (UTC)

Most reliable

In general, the most reliable sources are (a) peer-reviewed journals and books published in university presses, ...

I don't think this is true. If one is looking at the ICJ Bosnian Genocide Case is the most reliable sources "peer-reviewed journals and books published in university presses" or the documents released by the ICJ and other court's interpretations of those documents? If one wishes to list the people currently on trial at the ICTY, is the ICTY web site less accurate for such information than "peer-reviewed journals and books published in university presses"? Are peer-reviewed journals and books published in university presses, speculating on the legal position of prisoners held in Guantanamo Bay detention camp before the the U.S. Supreme Court ruled, in Boumediene v. Bush, on June 12, 2008, or are newspaper reports published the day after the ruling more reliable? If one wishes to see the accepted interpretation of the clauses Geneva Conventions are "peer-reviewed journals and books published in university presses" more reliable than the ICRC commentaries? --Philip Baird Shearer (talk) 13:51, 10 September 2008 (UTC)

Despite the circumstance you point out, you haven't convinced me that this is not true in general.--BirgitteSB 17:59, 10 September 2008 (UTC)


Reliable here, as elsewhere in Wikipedia, does not mean reliably accurate but rather certified as having been read by a knowledgable third party before it was published. I'm personally quite open to changing this and the other section you cited above. I retained these from prior incarnations of PSTS to show respect for the history of this policy. I think the proposed ranking of the quality of secondary documents being vetted by peer reviewer down to a city desk editor provides some guidance to people judging the general merit of sources, but it really has little or nothing to do with the question of avoiding OR. Certainly primary source documents can be used in ways that do not involve OR, and many if not most editors know how to do so.--SaraNoon (talk) 22:45, 10 September 2008 (UTC)
This phrase is part of at least two policies (WP:NOR and WP:V), and for a very good reason. I would not be in favour of touching it. Given that we are writing an encyclopedia, we should be reflecting scholarly research where it is available. Jayen466 23:34, 10 September 2008 (UTC)

Repetitious

The page seems to repeat itself a lot. I'm tempted to be bold and merge some of this, but as this page is "under discussion", I'd like some input first. - jc37 10:23, 1 October 2008 (UTC)

Go ahead, but remember that the top section is a substitute for PSTS in policy and the rest would be on a separate page as a guideline. So there should be a lot of duplication between the two sections, but not within each section.--SaraNoon (talk) 16:12, 3 October 2008 (UTC)