All of the questions that ask about whether a source is "reliable" all have the same answer. There's no substitute for doing the analysis yourself, and to me these questions seem to ask for permission to skip analyzing the information.

In the don't-ask section of the help center which talks about constructive subjective questions, we see that they should:

invite sharing experiences over opinions

I would like to edit these into questions and ask the question that I think should have been asked in the first place: about how to evaluate the source instead.

There is a valid question lurking in Checking Daughters/Sons of the American Republic (DAR/SAR) lineages?, because the standards of research required for getting in have probably become more rigorous over the decades as the standards of proof have been refined. But answering the question on those grounds ducks the real question about how to evaluate the information.

Your thoughts?

  • I agree - are you proposing a wording change to the help center on-topic page (that we can change)? We cannot modify don't ask. If so, perhaps we can ask answerers to propose a wording change that they would like to see. Or are you just seeking endorsement for re-phrasing questions like these by editing them to make them more answerable. If the latter then that is precisely what I see as being the purpose and extreme value of our wiki-style editing.
    – PolyGeo Mod
    Apr 24 '17 at 23:09
  • I don't know what the best solution is. That's why I wanted to hear from the community
    – Jan Murphy Mod
    Apr 24 '17 at 23:30
  • 1
    As someone who has not been active very long, I am a rather confused at this point. On one hand I understand the advice of doing my own analysis, but on the other hand, I do not want to have to do everything from scratch. Surely there are some repositories that are more reliable than others. It would be really useful for me to have some indication of the error rates. If I know that collection X has severe errors in every other record, I will approach it differently from collection Y with, say, major errors once every 100 records. Statistics along that line would be extremely valuable for me. Apr 25 '17 at 20:01
  • @user3697176 I understand that you haven't been doing this very long because you've just referred to repositories and collections as if they were the same thing. So I'm a bit confused about what you're asking for.
    – Jan Murphy Mod
    Apr 25 '17 at 22:01
  • I meant collections both times, as you surmised, and I am mostly interested in things available on the internet . What I am looking for is statements along the lines of: Scanned records in FamilySearch.org are generally trustworthy, but there are occasional transcription errors. US Census records are generally trustworthy, although some spellings are highly imaginative. Ancestry.com is almost pure fiction, etc. (I am basing this on my own very limited experience, so please correct me if I am wrong.) It would be very nice if there were on-line databases containing this kind of meta-information. Apr 26 '17 at 0:51
  • That's easier to answer in the negative. Ancestry's US City Directories for instance. The indexing is horrible -- there are entire half-pages of the directories for which there are no entries in the index. But the same answer remains -- there is no substitute for going out & looking yourself. Safer to assume everything is bad until proven otherwise.
    – Jan Murphy Mod
    Apr 26 '17 at 0:58
  • 3
    What you can look for are "known issues" articles (see the FamilySearch Wiki for each specific online collection), 'scope and content' articles from archive catalogs and other descriptive information created by archivists, and scholarly articles from researchers who are doing population studies. But I don't think that's the kind of "reliability" assessment you're asking for.
    – Jan Murphy Mod
    Apr 26 '17 at 4:19
  • @user3697176 This article may be of interest. Thomas W. Jones, PhD, CG, CGL, “Perils of Source Snobbery,” OnBoard 18 (May 2012). bcgcertification.org/skillbuilders/skbld135b.html
    – Jan Murphy Mod
    Apr 26 '17 at 21:49

I'm unclear why we should single this out among the (many) markers for a poor questions. And we already have mechanisms to fix them (edit) or indicate disapproval (down-voting).

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .