Choosing between self-archived articles and the publisher's version

With preprint and postprint options for self-archiving now often available to authors of journal articles, it seems like the range of potential access points has expanded from what used to be the case in the past. Someone trying to track down an article might find it:

• as a preprint in an institutional repository or a discipline-specific repository (the author of the article would have put it there prior to the peer-review, acceptance, or publication of that article in a journal)
• as the official version published by the journal and found in print or online (this assumes that the reader is affiliated with a library that has an online or print subscription to the journal. or that the library has a database with full-text access to that journal, or that the library is willing to get it via ILL)
  
• as a postprint copy of the article, usually found in an institutional repository or a discipline-specific repository (again, it is up the author to see that it gets deposited, although as Dorothea Salo explains well, this task is usually handled by the person managing the repository, a situation referred to as "mediated deposit.")
  

I've previously written about the challenges of knowing what discovery tools work best for finding articles in OA repositories; now I'd like to focus on the complexities of how to help a patron who has found two or more different versions of an article (for example, a postprint located in an OA repository and the published version in a database) and wants to know which one he/she should prefer. My instinct is to go with the published version, as that may be the one most likely to be free of any possible typographical errors (although one can imagine a scenario where a journal version had a typo and the author deposited a corrected postprint edition in an OA archive). Does anyone have any experience helping researchers make these kinds of choices?