There is a best-practices tag which currently has no usage guidelines. Currently it mostly seems to be used on questions which are very broad, particularly where there might be multiple approaches (e.g. project management, data analysis, modelling strategies, how to write code).
I have had a couple of edit requests in the review queue asking to add the tag to various questions, and in the absence of usage guidelines it isn't completely clear where it would be appropriate.
On this question, I could understand why the tag was being added (the question includes the phrase "Is this an example of semi-supervised classification and what are good approaches one can take to this kind of problem?"). I rejected the edit in that case, as the question seems rather too specific compared to other questions with the tag. I felt that if this question deserved a "best practices" tag, then almost all questions should - after all, don't most questions (at least about plotting and data analysis; mathematical statistics or probability not so much) want to know what would be "best practice" in their situation?
On this question I felt that "what is a sensible way to approach" determining "if the missing data are 'missing at random' or 'missing not at random'" seemed sufficiently broad to make the tag worthwhile.
But on reflection it is hard to see how to make a principled decision about this, and I can understand that other people might not draw the same distinction that I did.
I'd like to see if there's community consensus on what usage guidelines should be for [best-practices]
. It seems to make particular sense in situations where there are a variety of practices but some of which are generally deemed superior to others (for instance, in regression modelling, stepwise regression is regarded as "not best practice"). But I can't see obvious black and white limits about where the domain of "best practice" should end. If it were applied to every question where, at some stage, an analyst has to make some sort of choice between methods, then it would become one of the most frequent tags on the site - but its value in identifying content related to professional practice (which is the use it currently seems to serve) would be almost entirely diluted. Although at the moment it does seem to serve some kind of (undefined!) purpose, I can even see an argument for getting rid of it entirely, if it turns out to be impossible to pin down.
Update (Dec 21): All threads that had [best-practices]
as the only tag have been re-tagged. This tag would now be ready to be burninated (by SE admins) if community agrees.
[rule-of-thumb]
actually is especially "meta" (any more than e.g.[intuition]
is) ... but I can see that being something other people might come to different conclusions on. So might be worth a post. (Personally I'd suggest keeping that one, I think it is rather clearer in purpose and more useful than "best practices".) $\endgroup$