Wikipedia talk:Identifying and using tertiary sources

Initial Village pump discussion

See Wikipedia:Village pump (policy)#Wikipedia:Use of tertiary sources for some early discussion of this essay. — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼ 15:46, 28 May 2015 (UTC)[reply]

Interpreting NOR differently

Currently, the essay states:

The distinction between tertiary and secondary sources is important, because WP:No original research policy states: "Articles may make an analytic or evaluative claim only if that has been published by a reliable secondary source." Thus, such claims cannot be cited to tertiary or primary sources.

I interpret the quoted line from WP:NOR a bit differently, and don't think it intends quite the restriction that is stated in the sentence I put in bold (for emphasis). For one thing, we need to remember that many statements of analysis or evaluation contained in reliable tertiary sources will essentially be summaries of statements of analysis or evaluation published by reliable secondary sources. I think that when a tertiary source is summarizing reliable published sources, then that analysis or evaluation contained in the tertiary source meets the requirement of NOR.

And in the rare cases when the analysis or evaluation isn't simply a summary of what the secondary sources say... there another thing to consider: sources rarely fit into neat little boxes... Just as a reliable secondary source may contain primary data... a reliable tertiary encyclopedia, almanac, etc. can contain Secondary level analysis or evaluation. In other words an encyclopdedia or almanac (etc) may be "mostly tertiary" and yet act as a published secondary source for certain specifics. In which case... when we cite the encyclopedia for those specifics, we are actually citing a published secondary source. Of course, we have to look at the reputation of the "mostly tertiary" source (its editors and its authors) when determining whether it should be considered a reliable secondary source for the parts of it that are secondary... but if deemed reliable, then we can cite the analysis or evaluation as a secondary source.

So, to sum up my thoughts... I think we have two situations where a "tertiary" source can be cited for analysis and evaluation per NOR... 1) where it is summarizing the analysis and evaluation of reliable secondary sources. 2) where it acts as a reliable secondary source itself. Blueboar (talk) 14:59, 28 May 2015 (UTC)[reply]

@Blueboar: I have three objections to the first point, but see a clear way past them:

The policy simply doesn't say anything like that. WP:NOR is unusually strident here (it's rare for policies to use emphasis like that), and unmistakeable: "only if that has been published by a reliable secondary source".
With tertiary sources we usually cannot tell from what sources they're drawing their own material (thus cannot evaluate their reliability). When we determine the reliability of a secondary source we're looking at different factors than when we determine this for a tertiary one. Reliability, like notability, isn't transferrable. I can write a terrible tertiary work using impeccable secondary sources, or a fantastic tertiary work using a wide range of secondary and primary sources, many of them questionable. It's a totally different kind of judgement and process. The tertiary reliability question ultimately boils down to "are they doing a good job compiling information", which says nothing about the acceptance and validity of specific claims they've catalogued from other, secondary sources.
Inclusion in tertiary sources is often either very indiscriminate (database of all species, discography of all rock albums), or extremely selective (coffee table book of baseball greats), and each editorial approach presents neutrality problems (programmatic undue weight, or explicit bias, respectively). (Secondary sources are almost always more middle-ground. A book on the causes of WWII is not going to entertain every single relevant notion ever mentioned anywhere, nor only mention the rationales that the author already agrees with; the bulk of a such a work will probably consist of examining a wide variety of non-trivial ideas.) The excessively inclusive tertiary works are great for being able to provide hard-to-find data (like the average length of of a particular nematode, or the track list on a specific release of an album) but useless for undue weight and notability analysis. The excessively exclusive tertiary sources, like coffeetable books, and abridged field guides, aren't much good for anything, except temporary "better than nothing" citations for non-controversial facts.

I think the way past this difference of opinion is this: If a tertiary source is actually engaging in analysis or evaluation, then clearly it not in fact tertiary, but secondary by definition (at least with regard to that particular material). This ties directly in with your second point. If a tertiary source isn't engaging in those things, but our article writing is, and citing that tertiary source for it, then that's original research. Acceptable analysis/evaluation is defined as coming from secondary sources. So, either way, I see no way to arrive at the conclusion that tertiary sources can source an analytic or evaluative claim.

One the second point, I agree, but the essay already covers this in the "Exceptions" section. So, taking these two points together, I'm not sure there's a true problem of conflicting interpretations. It may be that the wording just needs some clarity, and I've been working on it for a few hours. I'm sure it will need more, especially if we actually contemplate merging this with WP:Identifying and using primary and secondary sources, which I didn't even know existed (it's not used much, and under-linked, and full of errors). — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼ 16:31, 28 May 2015 (UTC)[reply]

I get what you are saying, but still disagree. One of the purposes of a tertiary source is to summarize material published in secondary sources. The best ones actually do cite their sources... but even if they don't, if the tertiary source is at all reputable, we can assume that the material it summarizes comes from reliable secondary sources (and thus has been published by a reliable secondary source. Which means it has passed the "only" wording of NOR)... Indeed, if we refer to an analysis or evaluative claim from an encyclopedia (or other tertiary source) we are required to cite that encyclopedia (or other tertiary source) for that analysis or evaluation... per WP:SAYWHEREYOUGOTIT.

No, tertiary sources do have their place... and while they may not always be the best source possible they are almost always at least marginally acceptable sources. And if a tertiary source contains an analysis or evaluation it's definitely not Original research to repeat that analysis or evaluation here on Wikipedia. Blueboar (talk) 21:10, 28 May 2015 (UTC)[reply]

On the first point, we're just talking past each other. I'll try again. Re: 'an analysis or evaluative claim from an encyclopedia (or other tertiary source)' and later 'if a tertiary source contains an analysis or evaluation...' – If an otherwise tertiary source does that at all, they are, for that claim, a secondary source, by definition. It's something that secondary sourcing does and tertiary does not. It seems to me that you are in one half of the argument trying to permanently categorize as tertiary any source that is usually tertiary, even when for a particular case it's secondary; yet in another half of the discussion you say 'a "tertiary" source can be cited for analysis and evaluation per NOR [...] 2) where it acts as a reliable secondary source itself, which I've also said myself. Thus, I'm honestly not seeing where the disagreement on this point can actually be real.

On the other point, the policy emphatically says twice that such analytic claims have to come from secondary sources (I didn't notice this the first time around). The second instance is at WP:PSTS: 'All interpretive claims, analyses, or synthetic claims about primary sources must be referenced to a secondary source, and this comes immediately after mention of tertiary sources: 'Secondary or tertiary sources are needed to establish the topic's notability and to avoid novel interpretations of primary sources.' So, there is no confusion between or commingling of secondary and tertiary sources in the policy. Tertiary are good enough for helping to establish notability, but only secondary are good enough (it said so two times) for "interpretive claims, analyses, or synthetic claims". It's very rare for a WP policy to say the same thing twice, so clearly this is meant to be taken seriously." Assumed" to be "published in" isn't the same as "referenced to" a secondary source, which is what WP:PSTS says more restrictively. But even if both lines used "published in" wording, I'm certain the interpretation that can just assume such publication would not have been upheld, e.g. at an RfC. It's just poor wording in one spot, meanwhile it's very clear that everything in the policy is about citing sources not assuming they may exist somewhere. Your WP:SAYWHEREYOUGOTIT point takes us right back to the previous paragraph: If a usually-tertiary source actually has any "interpretive claims, analyses, or synthetic claims" that we're using, then it is a secondary source; we're not citing a tertiary source as if it were secondary, we're citing secondary source material directly, that just happens to be surrounded by tertiary material in the same publication. PS: By defining all reliable tertiary sources as those which summarize the analysis and evaluation of reliable secondary sources (you said we can "assume" this), your conditional, 'a "tertiary" source can be cited for analysis and evaluation per NOR [...] where it is summarizing the analysis and evaluation of reliable secondary sources' would automatically qualify to every reliable tertiary source (i.e., the only tertiary sources acceptable on WP, since we don't allow unreliable sources). Thus, it would directly equate reliable tertiary and secondary sources, which the policy obviously does not do. — SMcCandlish ☺ ☏ ¢ ≽^ʌⱷ҅_ᴥⱷ^ʌ≼ 01:59, 29 May 2015 (UTC)[reply]