|This page in a nutshell: The general problem addressed in this document is what editors should do with concerns about potentially inaccurate source material.|
WP:Editing policy states, "on Wikipedia a lack of information is better than misleading or false information". To this end, potential inaccuracy is a consideration for each and every source brought to an article.
Verifiable material may or may not be accurate
Editors sometimes think that verifiable material should be accurate, but verifiable material may or may not be accurate. A famous example of verifiable material that is potentially inaccurate is the front page of the Chicago Tribune on November 3, 1948—we have an article about this headline at "Dewey defeats Truman". In this case, we have a retraction from the newspaper which provides strong evidence that the material was inaccurate. But many published errors have not resulted in retractions.
As Carl Sagan pointed out in his The Demon-Haunted World, experts can be wrong or not even experts in the field in question. This means that using the fact that a source is verifiable to say it is accurate is the argument from authority fallacy.
The difference between "potential inaccuracy" and "inaccuracy"
There are few situations in life in which we have total knowledge, or in which we have language that is not subject to re-interpretation. From a practical viewpoint, there will always be a level of uncertainty in concluding that material is inaccurate.
It should be noted that just because sources are in conflict does not mean that one or more has to be inaccurate. They can be portraying the subject from different points of view, and essentially be accurate within their respective POVs.
So it simplifies the analysis to discuss the likelihood of "potential inaccuracy" rather than the likelihood of "inaccuracy".
Should inaccurate material be excluded from the encyclopedia?
Editors may tend to think that inaccurate material should be excluded from the encyclopedia, because we want an accurate encyclopedia, but closer analysis reveals a more complex picture. Readers may want to be aware of apparent inaccuracies or patterns of contradictions as part of their reading. Apparent inaccuracies of a lesser note can be relegated to a footnote. Ultimately, with allowing for due weight considerations in how the material is presented, and notwithstanding copyright violations, the only reason to exclude verifiable material from the encyclopedia is because it is insignificant.
Approaches to reporting potentially inaccurate material
Potential inaccuracy is a reason to reduce the due weight that is assigned to such material.
As listed below, there are three main editorial approaches to reporting potentially inaccurate material: inline attribution, footnotes, and exclusion due to insignificance.
As with other editorial decisions, editors must consider the forms of evidence that are available.
Levels of exclusion regarding potentially inaccurate material
- We don't use Wikipedia's voice to say it, instead we use inline attribution.
- We mention the anomaly in a footnote.
- The potentially inaccurate material has so little prominence (WP:DUE), that we don't mention it at all.
Examples of forms of evidence regarding potentially inaccurate material
- Editor's opinions are one form of evidence, because as long as there is a consensus that such evidence is enough, that is ok. "Obviousness", such as when editors agree there was a typo in an otherwise reliable source, fits here.
- Inductive reasoning based on reliable source statements. However, we are not a part of the scientific process, so such reasoning should only require a high school education.
- Older source material tends to be more inaccurate than newer source material.
- Retractions by the publisher are strong evidence of inaccuracy, but not absolute (e.g., a retraction may be politically motivated).
Examples of verifiable yet potentially inaccurate material
Note: These are examples, see the article for the current resolution regarding the issue.
- 1930 Palm Island tragedy: We have two spellings, "Prior" and "Pryor", in reliable sources. (1) Only the earliest newspaper articles used the spelling Pryor. (2) An editor claiming to be the nephew of Prior reports that "Prior" is the correct spelling (see Edit History). Most but not all editors agree that only one of the two spellings can be correct.
- 1930 Palm Island tragedy: Sources geographically distant from the event use the spellings "Patterson" and "Paterson". Most sources use "Pattison". Most but not all editors agree that only one of the three spellings can be correct.
- Conspiracy theory: "The first recorded use of the phrase "conspiracy theory" dates back to a history article from 1909." (Knight, Peter. Plots, paranoia and blame. BBC News, 7 December 2006). Knight is a senior lecturer in American Studies from the University of Manchester being quoted in a well-respected paper--RS through and through. Another reliable source that implies 1909 is the Oxford English Dictionary. However such statements are inconsistent with other reliable source evidence that the phrase was used earlier. The phrase "conspiracy theory" occurs before 1909 in:
- Garrison, George Pierce (1906) Westward extension, 1841-1850 Edited by Albert Bushnell Hart LLD Professor in history in Harvard University, p. 31,
- Rhodes, James Ford, (1895) History of the United States from the compromise of 1850 New York, Harper,
- (1891) The Economic review: Volume 1 Christian Social Union (Great Britain) Oxford University Branch, p. 540,
- Ellis Thompson, Wharton Barker The American: a national journal: Volumes 19-20 10 May 1890, p. 67,
- McCabe, James Dabney (1881) Our martyred President ...: The life and public services of Gen. James A Garfield, p. 556,
- (1871) The Journal of mental science: Volume 16 Association of Medical Officers of Asylums and Hospitals for the Insane (London, England), Medico-psychological Association of Great Britain and Ireland, Royal Medico-psychological Association, p. 141.
- Erin Burnett: An interviewer stated that Erin Burnett attended St. Andrew's Episcopal School (Maryland). St. Andrew's School (Delaware) is a boarding school much closer to her hometown. It was posited by an editor that it is physically impossible for her to have attended the school named in the interview, as her hometown is on the opposite side of the Chesapeake Bay, and the school does not board students. Later, a copy of her high school yearbook was found as a source.
- Al Davis: Multiple sources after Davis' death credit him as being the first National Football League owner to have hired a Latino head coach, Tom Flores. Earlier sources noted that Tom Fears was the first Latino coach in the league. All sources are from reliable news agencies. One source says Fears was the first, having started coaching in 1967, while Flores was second in 1979. Editors are able to corroborate the starting dates by researching a reliable statistics source, but this source makes no claims about the first Latino hired.
- Dewey defeats Truman
- The Moon is made of green cheese
- Myth of the Flat Earth
- Early March 2012 tornado outbreak: That "A baby was blown ten miles by this tornado" was reported around the world and added to this article. More at New Pekin, Indiana.
Appendix: Reliability in the context
Reliability in the context is subtly different from inaccuracy, and the difference is the difference between a verifiable source with potential inaccuracy, and an unreliable source that fails WP:V. Evidence of inaccuracy may be used to argue to the unreliability of the source in the context.
The content guideline Wikipedia:Identifying reliable sources makes these statements:
- The reliability of a source depends on context. 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.
- Proper sourcing always depends on context; common sense and editorial judgment are an indispensable part of the process.
- Deciding which sources are appropriate depends on context. Material should be attributed in-text where sources disagree.
- Whether a specific news story is reliable for a specific fact or statement in a Wikipedia article will be assessed on a case by case basis.
Examples of reliability-in-the-context issues
- 2011 Norway attacks: The claims that “all roads into Oslo's downtown area were closed” and “public transport into and out of the city was also halted” were disputed based on personal experience and the argument was made that the refs were in a foreign language and that the authors used terminology unlike that which the local media used. See also, Talk:2011_Norway_attacks/Archive_2#Transportation.
- Alexander Hamilton: a distinguished American historian, talking about a different subject, made assertions about what George Washington and Alexander Hamilton did in 1796. They would be true of Washington and his Secretary of the Treasury, and are based on Washington's collected papers, which includes letters to and from the "Secretary of the Treasury." The source forgot for the length of a sentence that Hamilton had resigned in 1794, and the Secretary in question was Oliver Wolcott, his successor.
- Pluto: From 1930 to 2006 Pluto was classified as a planet but during 2006 the IAU formally established a definition of planet which resulted in Pluto falling out of the planet category and into the category of dwarf planet. So any source calling Pluto a planet must be viewed in the context of when it was written rather than modern context.