Please post your tag synonym suggestions as new answers in this thread, one answer per suggestion. Upvote answers where you believe that the suggested tags should be made synonyms, and downvote answers where you believe the tags should remain separate. Well upvoted suggestions will be eventually implemented by the moderators (and then the corresponding answers will be deleted).
10 Answers
GPT is just one well-known example of a Large Language Model (LLM), but has no wiki and only 6 questions. LLM is more general, and has a wiki.
-
1
-
3$\begingroup$ We also have voting-system which is distinct and for which I have just proposed wiki info. $\endgroup$– mdeweyFeb 9, 2021 at 15:28
-
2$\begingroup$ Looking at this again I see we also have survey-weights which I would have thought was a proper subset of survey-sampling $\endgroup$– mdeweyMar 6, 2021 at 13:41
-
1$\begingroup$ @mdewey: I think I will ask this as an independent question, to get more people involved. Seems complicated! $\endgroup$– kjetil b halvorsen ModMar 8, 2021 at 2:02
independence (1381) $\leftarrow$ non-independent (574)
EDIT:
non-independent usage guidance (there is no wiki entry):
Data, events, processes, etc, are non-independent if knowledge of 1 provides some information about the state or value of the other.
independence usage guidance (more info at wiki entry):
Events (or random variables) are independent when information on some of them tells you nothing about the probability of occurrence (/ distribution) of the others. Please DO NOT use this tag for independent variable use [predictor] instead.
These are literal opposites and I don't see any need for them both to exist.
Not sure, but perhaps:
mixed-model (5495) $\leftarrow$ random-effects-model (1450)
I suppose one can have a model with only random effects and no fixed effects, which would make it different from a mixed model, though I don't think the conceptual distinction is so large that it would merit a separate tag. And I think all the questions I've seen with this tag are about using random effects in mixed models.
-
1$\begingroup$ I agree the distinction is not large enough to merit two tags. $\endgroup$– kjetil b halvorsen ModJul 9, 2022 at 0:35
-
1$\begingroup$ It's worth noting that these have different meanings between statistics & econometrics. $\endgroup$ Jul 11, 2022 at 17:38
-
3$\begingroup$ @gung-ReinstateMonica That's interesting - the distinction is not evident in the tag info. If the distinct econometrics definition is sufficient reason to retain both tags, it may be worth a tag edit to clarify. $\endgroup$– mktJul 11, 2022 at 18:01
-
1$\begingroup$ Skimming over them, a lot of the
[d]
Qs don't seem to be about[ds]
. Maybe someone should go through them & retag some first? $\endgroup$ Jul 20 at 11:34 -
2$\begingroup$ The decision is problematic at the very first place. It is not a formal term and is prone to be used in a random way by a random user. If the decision-theory is more apt at any place, only it should be used there. Let the decision tag be exiled in favor of the more appropriate [decision-theory]. If the former is used somewhere as the latter's synonym, then those posts ought to be re-tagged, imo. $\endgroup$ Jul 20 at 16:53
-
$\begingroup$ @User1865345-solidarityMods That is basically what I am proposing here. If you support it, you can upvote the answer. $\endgroup$– mktJul 20 at 16:56
-
1$\begingroup$ Done @mkt. Looking closely, decision has been used rather in an indiscriminate way wherever there is decision in the statistical method(s). $\endgroup$ Jul 20 at 16:57
-
2$\begingroup$ @gung-ReinstateMonica You're right, many of them tagged
decision
seem to be about decision trees and so ought to be taggedcart
instead. I will retag those. $\endgroup$– mktAug 3 at 6:16 -
2$\begingroup$ @mkt, thanks. Once they are all appropriately retagged, the mods can make the synonym, merge them, then delete the synonym and
[d]
will disappear. $\endgroup$ Aug 3 at 11:57
pooling (266) $\leftarrow$ pooled-model (6)
EDIT: pooled-model has no usage guidance or wiki. pooling has usage guidance:
Pooling, eg for variance, is used when several groups or populations are assumed to have a common property (a common parameter value) and the information from all the groups or populations are used together to estimate that common property.
I think this usage guidance is broad enough to cover pooled-model as well.
-
$\begingroup$ It would probably help if you provided a little more information (are there wikis? how many overlap, how many seem to be about the same topic, etc.), and made a case for the synonym. $\endgroup$ Jul 28 at 11:33
-
There seems to be a fair bit of overlap between:
p-value (2573) and statistical-significance (5,774)
And the usage of hypothesis-testing (9591) overlaps a fair bit as well, though in principle it does differ more.
Is there a case for merging the first two? And perhaps editing the tag info for the third to make a distinction?
-
1$\begingroup$
ss
vsht
has been discussed before. I see a subtle distinction, but that doesn't mean they will be used well by typical askers.pv
->ss
may be a viable merge. $\endgroup$ Jul 15, 2022 at 13:57 -
$\begingroup$ I'm not sure here ... why is p-values more related to statistical significance than to hypothesis testing? By the way, p-values have their own existence, now, without regard to where they come from ... and some uses of p-values can be seen as descriptive. $\endgroup$– kjetil b halvorsen ModJul 19, 2022 at 17:12
-
1$\begingroup$ @kjetilbhalvorsen To your first point, Bayesian hypothesis testing exists although I doubt it appears often in questions with that tag. And as for the second, I'd be curious to see if that is a common view. I would have thought statistical significance and p-values would be hard to divorce from each other. To me, they seem much closer to each other than to hypothesis testing (which could itself be made a synonym quite reasonably, I think). $\endgroup$– mktJul 19, 2022 at 17:29
-
1$\begingroup$ All reasonable comments, I will think more about this and come back later ... we could also compare posts under the tags and see if there is much sign of difference in use ... $\endgroup$– kjetil b halvorsen ModJul 19, 2022 at 17:33
There are now two tags, pre-post-comparison and change-scores essentially covering the same topic. I propose we let pre-post-comparison be the main one (even if is is newer and less used), as it is the most general one.
-
1$\begingroup$ (+1) I recently created
pre-post-comparison
because, as you note, there are several closely related topics (with quite a few threads) that are not covered by the much more specificchange-scores
. $\endgroup$– mktAug 21, 2022 at 17:51 -
$\begingroup$ I've come across questions under the
mixed-model
tag that might be appropriately tagged withpre-post-comparison
too. But the terms "pre-post" (and variants) don't appear in the body of the question because the OP didn't think about it: 1. And here is a simplyregression
example: 2. It maybe challenging to apply thepre-post-comparison
consistently? Or perhaps I misunderstand its application? $\endgroup$– dipetkovAug 21, 2022 at 18:40 -
2$\begingroup$ @dipetkov I frequently add relevant tags that the OP missed or is unaware of (often because they're unaware of the terminology). I also leave a comment suggesting that they look at the top questions in the tag if it's a question that is likely to have been addressed previously. $\endgroup$– mktAug 21, 2022 at 19:42
-
1$\begingroup$ To my mind, change scores & pre-post comparisons are (at least subtly) different things. $\endgroup$ Aug 21, 2022 at 20:17
-
1$\begingroup$ @gung: They are certainly subtly different, I see that change-scores is a subset of pre-post-, and for a long time that tag has been used that way. See its tag wiki $\endgroup$– kjetil b halvorsen ModAug 21, 2022 at 21:21
-
1$\begingroup$ @gung: I still thinks that for the purpose of organizing posts here, it is better to synonimize. What do you think? $\endgroup$– kjetil b halvorsen ModOct 31, 2022 at 1:17
-
$\begingroup$ @kjetilbhalvorsen, to me they are at least subtly different. However, I didn't downvote this suggestion or give a firm thumbs-down. So if others are going for it, I'm not stopping it. $\endgroup$ Oct 31, 2022 at 13:46
feature-scaling (120 posts, no wiki) must be synonymous to either normalization or standardization (both with wikis), or both ... ?
-
6$\begingroup$ Scaling seems more generic. Maybe
[n]
&[st]
should be grouped under[fs]
, & we could even create a[centering]
tag to be a synonym as well? Dunno what's best here... We probably don't need 3 tags for this, though. $\endgroup$ Feb 24, 2022 at 14:00 -
$\begingroup$ @gung-ReinstateMonica Is centering the same thing as scaling? I always thought scaling was a specific application of the multiplication function, but centering is a specific application of the addition function. (Leaving aside that multiplication itself is repeated addition. :) Do you have a new insight for me, gung? $\endgroup$– AlexisJun 12, 2022 at 3:32
-
1$\begingroup$ here is my standard answer on the specific use of these terms, @Alexis, but I'm guessing
fs
is a generic term. I think these should be grouped together somehow, but how best to do it is unclear to me. The best superset name would probably belinear-transformation
, but that doesn't seem like it would be a good choice for our situation. $\endgroup$ Jun 12, 2022 at 11:47 -
1$\begingroup$ I agree with gung --
[normalization]
$\to$[feature-scaling]
and[standardization]
$\to$[feature-scaling]
seems like the better solution. $\endgroup$– Sycorax ModJun 2 at 20:07
multilevel analysis $\iff$ mixed model
-
3$\begingroup$ If you agreed with @dipetkov, it would be better to delete this suggestion, & then add another. It's better not to change the suggestion entirely. $\endgroup$ Aug 14, 2022 at 6:54