7
$\begingroup$

The homogeneity tag seems to map to "heteroskedasticity"

This is problematic because homogeneity of variance is not the only form of homogeneity people are interested in.

Consider, for example, this post, which I have been trying to improve (still a work in progress, I'm afraid, I started with trying to give it some context). He seems to be trying to follow the advice in this paper, or perhaps to follow the analysis in Wijngaard et al (2003)*.

The OP was asking about homogeneity, so wanted to use the 'homogeneity' tag (n.b., the tag doesn't specify '... of variance' so that's perfectly reasonable). But it just replaces that reasonable use with the misleading tag 'heteroskedasticity'.

Consider also the chi-square test of homogeneity of proportions. Someone might (again, legitimately in my opinion) want to use the homogeneity tag... with a rather unfortunate result (especially if they don't notice that it silently changes).

If the tag was 'homogeneity-of-variance' I think the argument for such a merge would be strong. But without the 'of-variance' I don't think they're necessary the same thing at all.

What should be done about it? I'm somewhat disinclined to simply say 'you can't use the homogeneity tag to talk about homogeneity'.

$\,$

* J.B. Wijngaard, A. M. G. Kleink Tank, G. P. Konnen, (2003),
Homogeneity of 20th Century European Daily Temperature and Precipitation Series,
Int. J. Climatol, 23, 679-692.

$\endgroup$
  • 2
    $\begingroup$ There's also homogeneity analysis (but see the R homals package), so I would be inclined to update the list of synonyms for the homogeneity tag in order to ensure that only heteroskedasticity <-> homogeneity-of-variance is suggested as tag synonym. $\endgroup$ – chl Jul 19 '14 at 19:44
  • 1
    $\begingroup$ It seems homogeneity fits in the meta tag concept. It does not help to much narrowing down an issue. Maybe in that post you could use tags r, hypothesis-testing and uniform. $\endgroup$ – Andre Silva Jul 19 '14 at 21:32
3
$\begingroup$

Without moderator action (I'm not clear on all the required steps), nothing is going to happen to the [homogeneity] tag. You do make a compelling argument for changing the existing tag to [homogeneity-of-variance], and then making that a synonym of [heteroscedasticity], but that also seems like it might be more trouble than it's worth.

Instead, if you have another, specific kind of homogeneity that you think merits a tag (e.g., [time-series-homogeneity]--possibly with a better name ;-), then why not create such a tag directly?

$\endgroup$
  • 2
    $\begingroup$ There's no problem in updating tag synonyms or editing current remaping. The problem really is what to do with homogeneity. $\endgroup$ – chl Jul 19 '14 at 19:46

You must log in to answer this question.

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