This page is an essay on WP:No original research, and WP:Identifying reliable sources.
|This page in a nutshell: Analysis and evaluation require reliable secondary sources, and we cannot cite tertiary sources for them. Tertiary sources differ from secondary ones by not themselves providing significant analysis, commentary, or synthesis. However, some tertiary sources are secondary in some applications.|
Generally speaking, tertiary sources (for Wikipedia purposes, as discussed at WP:No original research § Primary, secondary and tertiary sources, and WP:Identifying reliable sources § Primary, secondary, and tertiary sources) include any compilation of information, without significant new analysis, commentary, or synthesis, from primary and secondary sources, especially when it does not indicate from which sources specific facts were drawn. 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.
There are many types of typically tertiary sources:
Some of the above kinds of tertiary sources are considered forms of secondary literature in some disciplines, but remain tertiary (for most of their content) for Wikipedia's purposes including in those disciplines. Not understanding this is a common error by subject-matter experts new to Wikipedia editing.
The medium is not the message; source evaluation is an evaluation of content, not publication format.
Reliability of a tertiary source is principally determined by four factors: whether its producers (i.e. writers and/or editors) have subject-matter expertise, whether the underlying original sources of the non-novel material are clear, whether its producers are independent of the subject, and whether the work is generally regarded as reliable by others in the field in question (primarily a matter of authorial and publisher reputability). These factors counterbalance each other. For example, while typical mainstream dictionaries do not cite sources for specific entries, how authoritative they are considered can be gleaned from independent reviews of their content and editorial practices. Many tertiary works only cite sources in a general way, e.g. a bibliography. Beware tertiary works that have no indication of their own sources at all.
Another factor to consider with tertiary sources is they are often more error-prone than secondary sources, especially the more comprehensive they are. A database of millions of pieces of biographical data, each often taken from a single original primary source and added by a stressed and bored data-entry operator, is less likely to have gotten a particular individual's birth date correct than a book (secondary source) written about that person, drawing on multiple sources.
Tertiary online sources that are written in whole or in part by a general-public editing community are user-generated content, and are not reliable sources. This includes content farms, which have a paid but indiscriminate array of innumerable writers, and little editorial oversight, though many of them go to some lengths to disguise their nature.
Simple facts: A tertiary source is most often used for reference citations for basic and fairly trivial facts which are not likely to be disputed and which can be verified in other sources. Examples include various vernacular names for a species, the pronunciation of a foreign word, or a baseball player's statistics in a particular year. The WP:Good article nominations and WP:Featured article candidates processes tend to check that all statements in an article are sourced, and tertiary sources frequently are used for many non-controversial details.
Simple comparisons: Another common use is comparative, especially involving simple facts and basic concepts. An example is citing multiple dictionaries to show how interpretation of a term may vary. (Comparative use of tertiary sources for more complex or contentious material is ill-advised, as detailed below.)
Better than nothing: Tertiary sources are also commonly used when a secondary source has not yet been found. For example, a field guide about cacti has probably been reasonably well fact-checked, and can be cited as a source for the range of a particular species, if no source focusing on that species (and perhaps with more recent data) has turned up yet.
Older but still relevant details: Older tertiary sources can be used to source former, obsolete views or facts that need to be reported on in a Wikipedia article, for completeness, especially when it's difficult to find modern sources that even mention a long-replaced idea, name, person fulfilling a role, or whatever. As detailed below, there is a major difference between using a tertiary source to report obsolete facts as such, and trying to use them to preserve obsolete facts as still verifiable (e.g., you can use 19th-century encyclopedias to illustrate how seriously phrenology was once taken, but such sources cannot be used to try to contradict modern scientific works).
Analysis and evaluation: A tertiary source cannot be used, as a matter of policy, as a source for "an analytic or evaluative claim". This is left deliberately broad, so it is not subject to technicality gaming.
Controversial material: Any controversial, alleged fact is essentially unsourced if the only citation it has is to a tertiary source of questionable reliability (on the particular point or generally). As with secondary sources, this can happen for any number of reasons, including source obsolescence, lack of subject-matter expertise, conflict of interest, simple error, or presentation of a fringe idea as comparable to the generally accepted view, among other problems that can arise with a particular source. A tertiary source that is a compendium of factoids by an author with no known expertise, and which indicates nothing about the sources of its own information, is not a reliable source. Anyone could compile a large book of alleged facts, anecdotes, and folklore about any given topic, and probably find a willing publisher, without any fact-checking ever taking place. Note however that not all facts about a controversial subject are themselves controversial; there is no principle that a reliable tertiary source good enough for one article is not good enough for another because of the topic's notoriety, the amount of emotion editors bring to editing it, or the frequency with which our article on it is vandalized.
Complex or controversial comparisons: Comparative use of tertiary sources can be fraught with problems relating to undue weight, non-neutral point of view, novel synthesis, and lack of basic accuracy if the things being compared are subject to real-world contention, or are complex in nature. For example, a comparison between Christian, Judaic, and Muslim concepts of God is unlikely to produce encyclopedic results if based in whole or part on tertiary sources, which are likely to present a poorly nuanced view of complex theological questions and details of interpretation. Complex comparative work must actually be done in secondary sources cited by Wikipedia for those comparisons. The WP:AEIS policy does not permit Wikipedians themselves to engage in substantive "analysis, evaluation, interpretation, or synthesis" of facts or sources.
Over-inclusive works: Indiscriminate sources must be considered skeptically when determining both notability and due weight. Unfortunately, a large proportion of tertiary sources are indiscriminate. A guidebook that attempts to describe every restaurant in a city cannot reasonably help establish that a particular restaurant is notable. An index of every paper published about a topic in a given year tells us nothing about the critical reception of any given paper. The more inclusive, comprehensive, even "complete" that a work aims to be, the less useful it is for determining the notability of any subject it mentions. On the up side, the more comprehensive a work is, the more likely Wikipedia editors are to find reliable details in it about any subject within its purview. Thus, in a selection of tertiary sources for a topic, the source that is most reliable for WP:Verifiability purposes has a tendency to be the least valuable for notability and due weight analysis. The inverse is often not true; an exclusively selective, non-comprehensive source may well be very unreliable, too, simply because it was poorly researched and reflects a superficial, popular-opinion approach to its topic, as is often the case with coffee table books.
Better sources available: While a good tertiary source can usually be used without incident to source non-controversial facts, such citations can and should be superseded by ones to reliable secondary sources. WP:Identifying reliable sources tell us: "Each source must be carefully weighed to judge whether it is reliable for the statement being made and is the best such source for that context." It is extremely rare for a tertiary source to be the best such source, for anything, in any context; they're simply often the most readily available and easily digestible (being somewhat predigested). Sometimes a tertiary source can even be replaced with a primary one; for example, a dog breed's actual breed standard (the primary source) is more reliable for the breed's defined characteristics than a tertiary dog breed encyclopedia, though the latter might be very useful for differences and commonalities between varying standards published by different organizations, and may be a good source of additional details, like demographics and breed history. "Stacking" tertiary source citations after a sufficient secondary one is not advised; it does not add more verifiability to the claim in the article, but simply adds clutter.
Outdated material: An obsolete source cannot be used to "trump" newer reliable sources that present updated information, most especially when the older source states or implies a negative that cannot be proven but can be disproven easily by new data. A pertinent example (detailed here) is a prominent dictionary asserting that a specific phrase was first used in publication in a certain year, while later research found older examples, disproving this assertion (with its implicit negative, that there were no earlier cases). Because most tertiary works take a long time to assemble, or (in more dynamic media) are in a constant state of being incrementally updated, it is fairly likely that some particular pieces of information in such a work have already been surpassed by the newer work of others. Some information in tertiary sources may already be obsolete before they even see publication. Sometimes the very conceptual framework behind such a work becomes obsolete, given the passage of enough time, with enough advancement and reorganization in the field to which it pertains. E.g., a decades-old tertiary list of species within a genus, based on outmoded ideas of classification, cannot be used to contradict or seek undue weight against a widely accepted re-classification arrived at through more modern research. On the other hand, a recent high-quality tertiary source with clear and reliable sources may be of more value than an obsolete secondary one, especially in the sciences, where current understanding can be a fast-moving target.