4
$\begingroup$

(Note this previous, related discussion: Redundant tags: mixed effects and related models.)

The tag is very ambiguous. Unfortunately, this hasn't been addressed before and it now has 142 questions. An excerpt was just suggested:

For questions about hierarchical models, also called multi-level models.

However, it isn't at all clear that that's what the threads grouped under this tag are about. (It isn't clear they are consistently about anything.) For example, at least 33 are about clustering.

I'm wondering what we should do about this. It really seems to be a mess. Some options might be: retagging, making synonyms and/or merging, or deleting.

$\endgroup$
4
$\begingroup$

We have several well-established tags that start with "hierarchical":

In this situation having a generic tag is just plain confusing; everybody is supposed to be picking one of the more specific tags instead of using the obviously ambiguous generic one. So we should eliminate it.

It would probably require going through all 142 questions and retagging as appropriate. As always, we can manually retag only questions that are not about multilevel-analysis (based on the numbers above, this seems to be the most common usage). After this is done, the rest can be moved to multilevel-analysis via mod action (to save some time & to minimize flooding the front page).

Update (Nov 1, 2017): This has been finished, and the tag is gone.

$\endgroup$
  • 1
    $\begingroup$ Yeah, I was worried it might have to come down to going through all 142. I did just start retagging some that should obviously be [hierarchical-clustering] instead of [clustering] [hierarchical]. $\endgroup$ – gung - Reinstate Monica May 14 '17 at 19:05
  • $\begingroup$ If I were to retag these ones, I would leave [clustering] tag (and replace [hierarchical] with [hierarchical-clustering]); it's an old discussion about hierarchical tags (pun intended), but I find it useful e.g. to have clustering-related threads under [clustering] even if there is a more narrow tag. This is not very important though. $\endgroup$ – amoeba says Reinstate Monica May 14 '17 at 19:11
  • $\begingroup$ OK, so what I'm doing now is reading to see if each question is really only about the peculiarities of hierarchical clustering, or if it is about HC as an instance of clustering generally. I then tag w/ HC alone, or w/ both. In general, I'm not a fan of having to use 2 nested tags. $\endgroup$ – gung - Reinstate Monica May 15 '17 at 11:43
  • 1
    $\begingroup$ I found 8 that had "Bayesian" in them. Those have now been done. $\endgroup$ – gung - Reinstate Monica May 15 '17 at 17:14
  • 1
    $\begingroup$ All the threads from the search on "clustering" have now been edited. There are 99 <strike>red balloons</strike> threads left. $\endgroup$ – gung - Reinstate Monica May 15 '17 at 19:58
  • $\begingroup$ I guess all of them can be moved to multilevel-analysis now? Or do you want to go over them first to double-check? By the way, I edited the wiki excerpt & wiki of multilevel-analysis. I was hesitating about what to do with [nested-data] (that I created out of [nested] just some days ago) vs. [multilevel-analysis], but maybe it makes sense to leave them be as is. These concepts are strongly intersecting though. In fact, multilevel-analysis is somewhat vague. $\endgroup$ – amoeba says Reinstate Monica May 15 '17 at 20:14
  • $\begingroup$ Multilevel-analysis is somewhat vague. We can deal w/ that later, I suppose. I'm not sure if the rest can be merged now; I'm not sure if they are really all about MLA--I haven't yet gone through them to check. I've only done the ones that can be captured through simple, intuitive searches. $\endgroup$ – gung - Reinstate Monica May 15 '17 at 20:32
  • $\begingroup$ I still haven't gone through the remaining tags, but I'm not sure when I will. Do you think we should just go ahead and move them to multilevel-analysis now? What do you suggest? $\endgroup$ – gung - Reinstate Monica May 24 '17 at 16:44
  • $\begingroup$ @gung The last time we talked about it I took a brief a look at the remaining questions and had the feeling that there is some non-negligible fraction that are not really about multilevel analysis. So ideally I'd say we should look through them and remove [hierarchical] from everywhere where multivelel-analysis would be a misnomer. This would also make the case for the subsequent move stronger. But I did not come around doing it so far... $\endgroup$ – amoeba says Reinstate Monica May 24 '17 at 21:11
  • $\begingroup$ @gung I think this search stats.stackexchange.com/questions/tagged/… shows the threads that we should take a look at (and in many cases it seems remove the [hierarchical] tag), there is currently 40 of them. $\endgroup$ – amoeba says Reinstate Monica May 28 '17 at 20:44
  • $\begingroup$ FYI: I started going over the search results in my last comment above (42 of them at the moment) and am planning to retag all that are not about [multilevel-analysis]. After this is done, I think we can proceed with the merge as planned. $\endgroup$ – amoeba says Reinstate Monica Oct 29 '17 at 21:54
  • $\begingroup$ was our plan to make [hierarchical] a synonym of [multilevel-analysis]? (It would seem so...) I agree w/ your point that h is a bad tag to have around. If we make it a synonym, it will still show up when people start typing "hier" (I believe this is true even if merged). I wonder if retagging as m-a is better in the long run. $\endgroup$ – gung - Reinstate Monica Oct 30 '17 at 0:22
  • $\begingroup$ @gung Can't you (i) make the synonym, (ii) merge, (iii) delete the synonym mapping? $\endgroup$ – amoeba says Reinstate Monica Oct 30 '17 at 8:19
  • $\begingroup$ I don't know. I can try to find out. $\endgroup$ – gung - Reinstate Monica Oct 30 '17 at 11:25
  • 1
    $\begingroup$ It seems to have worked. The questions all have [multilevel-analysis] now, & [hierarchical] is just gone entirely. $\endgroup$ – gung - Reinstate Monica Nov 1 '17 at 0:16

You must log in to answer this question.

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