In some instances the question doesn't itself fit a particular tag, but the accepted answer/popular answers do. If tags are a useful guide for others then it could make sense to re-tag based on the answers as well as the questions?

For example, often the question might not specify the language, but then R based code examples are given in the answer - should these be tagged R?

Or the answer to the question might involve proposing the use of a particular statistical method, that would not be known to the question asker. So if the answer involves, try using X, should the question get tagged with X?

share
    
Sorry if this is a bad question, I asked it in good faith. Could someone quickly comment to explain to me what I did wrong? Is this explained somewhere else? –  Corone Jan 29 '13 at 15:03
    
From the Meta FAQ: "Voting here works a bit differently from the main site. On Meta, voting is often used to express agreement or disagreement, not to point out a lack of quality or helpfulness. Please don't be concerned if you receive downvotes – members of the community may simply disagree with your bug, feature request, support issue, or the nature of the discussion." –  Scortchi Jan 29 '13 at 16:33
2  
I think it's a sensible idea in general. After all if you're searching for topic X, answers about X could be just as useful to you as questions about X. But R in particular should be an exception or everything would end up getting tagged with 'R'. –  Scortchi Jan 29 '13 at 16:47
    
Ok thanks @Scortchi, that makes me feel better! Yes I see the point about R, it would get silly. For more specific topics though, such as if someone has a problem, and the answer suggests using a GLM, then that could get tagged GLM, even though the initial question didn't mention this. –  Corone Jan 29 '13 at 16:59
2  
Scortchi, nearly everything is tagged R. I am considered an R expert by the community, even though the last time I actually ran it was in November. I think the move often suggested in the comments on the main site is to tag the specific R package, such as lmer for mixed-model, for instance. –  StasK Jan 31 '13 at 20:02
    
Since I asked this a while back, and now something has prompted me to want an answer, I've created three CW answers to see if we can get a poll going. –  Corone Mar 5 '13 at 16:55
3  
I can't find it now, but I recall this was discussed a couple of years ago: the conclusion was that tags should reflect acceptable answers. I don't think a vote is appropriate because questions about how tags function are primarily site-wide; for consistency, this should be addressed on meta rather than here. –  whuber Mar 6 '13 at 22:03
    
Ok, cool. Do you want to write that as an answer, then I'll flag it as accept and delete the other answers? –  Corone Mar 7 '13 at 6:45
    
I assume when you say meta, you mean stack exchange meta. I'm not sure it obvious that all SE sites would use the same conventions for tags given the diversity of content. –  Corone Mar 7 '13 at 6:48
    
@whuber, sorry realised I didn't tag you in my comment reply, so you wouldn't have seen. –  Corone Mar 8 '13 at 21:45
add comment

3 Answers

up vote 3 down vote accepted

Adding this comment as an answer to try and encourage finding consensus on this:

@Scortchi says:

I think it's a sensible idea in general. After all if you're searching for topic X, answers about X could be just as useful to you as questions about X. But R in particular should be an exception or everything would end up getting tagged with 'R'.

This suggests any good answers about X should result in tagging the question.

Vote this answer if you agree...

share
    
I've marked this as accepted based on @whuber 's comment above –  Corone Mar 11 '13 at 9:06
add comment

An issue I found here (Should the gmm tag be added?) though is that what if the reference is in some or one of the answers, but not the accepted answer.

Vote for this answer if you think that retagging based on the answer should only consider the accepted answer.

share
    
I don't think the accepted answer should be considered the only answer that counts, in general. While I don't necessarily think that every two sentence answer should influence the tags used, there are often very good, notable answers in addition to the accepted 1. I think sometimes the accepted answer is middling, & some of the other answers are much better & should be give precedence in assessing the value of the thread & how it should be tagged. –  gung Mar 5 '13 at 19:36
add comment

For completeness:

Tagging should only refer to the contents of the question. The primary purpose of tagging is to help people filter topics they feel they can answer and contribute to. When searching to learn you can just as easily use simple text search rather than tag searching.

Vote for this is you agree...

share
3  
There is certainly a limit to how much influence the answers should have, but I don't agree that only the question matters. –  gung Mar 5 '13 at 19:38
add comment

You must log in to answer this question.

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