# What to do about the [control] tag?

We have a tag with 141 threads. It has no wiki excerpt and is a synonym of it. The absence of a wiki excerpt is wise since the threads range from control groups, control variables, control assays, control charts (although we do have for those too), and so on.

It seems to me that the position could be simplified by removing the tag where another one would be more specific like or even or one of its cousins. If that leaves a core with a common meaning the tag could then be defined but it may be possible to eliminate it altogether.

• A quick glance at the most upvoted [control] Qs shows that most of them are about "controlling for variables" in regression. I find the term "control" very unclear (I initially thought this tag is about control theory); in fact "statistical control" is IMHO not much better. Wikipedia calls it en.wikipedia.org/wiki/Controlling_for_a_variable - arguably too cumbersome for a tag name. I'd say we need to come up with a reasonable tag name, rename this tag accordingly, and then go through all 141 threads and remove the tag from the ones that are not about "controlling for a variable". – amoeba Dec 14 '17 at 20:14
• I don't see a particular problem with the length of controlling-for-a-variable; typing control would bring it up (so you don't usually have to type all of it) and it's then clear which thing it is about. – Glen_b Dec 14 '17 at 23:52
• @Glen_b There used to be 25 chars limit but I think it's lately been removed. If there are no better suggestions then we can go with that. – amoeba Dec 15 '17 at 13:47
• @amoeba there are only 141 of them so it might almost be as easy to re-tag them each manually which would also give an opportunity to flag any for closure. I assume renaming the tag requires special powers? – mdewey Dec 15 '17 at 16:23
• Yes only mods can rename, but I would still suggest to do it this way. One can then look through all 141, but we won't need to retag the ones that are fine. – amoeba Dec 15 '17 at 16:33
• @amoeba I believe it's 32 now but I am not sure – Glen_b Dec 17 '17 at 7:39
• (I am going to post another answer, but did not have time for it yet.) – amoeba Dec 20 '17 at 9:20
• @amoeba have you had any more thoughts or shall I try to push this one forward with a suggestion as answer? – mdewey Jan 7 '18 at 14:58
• @Glen_b I have now completed re-tagging and there are about 65 left which are about controlling for a variable. Can you do the re-naming as discussed above please? – mdewey Jan 26 '18 at 15:21
• @mdewey Sorry can you be explicit about the renaming you're asking for? These are hard to untangle if I screw up. – Glen_b Jan 27 '18 at 4:12
• @Glen_b the last suggestion was [controlling -for-a-variable] which got three upvotes. I will ping amoeba too. – mdewey Jan 27 '18 at 14:25
• @amoeba do you agree with [controlling-for-a-variable]? – mdewey Jan 27 '18 at 14:26
• Don't have any better suggestions, so yes. Also, if we come up with a better name later, we can always [ask a mod to] rename it again. – amoeba Jan 27 '18 at 16:31
• @mdewey By my understanding the proposal was to make anything that's currently got the control tag become controlling-for-a-variable. That's done now; I didn't synonymize them (forcing every future use of control to be controlling-for-a-variable), but that should be doable if you need it. – Glen_b Feb 6 '18 at 0:51
• @amoeba I have written a brief excerpt. – mdewey Feb 6 '18 at 15:39

So as to move this forward I propose the following course of action.

I shall go through the (now) 144 questions and (1) re-tag any of them with a more appropriate existing tag, and (2) remove the control tag where it seems redundant given the other tags.

If it then seems to me, having read them, that @AdamO's proposed wiki does fit all the ones which are left then I will edit the wiki unless he does it first. If on balance the earlier suggestion from comments on my question, controlling-for-a-variable, seems appropriate i shall come back here and edit this.

======== edit 24 January 2018 ===============

I have now gone through them, deleted redundant tags, re-tagged with more appropriate (mostly , ) and there remain 95. Of these eight are, I think, about control theory.

Estimating a first order plus dead time model

Gradient descent for a noisy system

Using dynamic programming with n-step ahead predictors

Template learning in Neural Network for control applicatication

Existence of ε-optimal Borel measurable policies in stochastic control

Deriving the particle filter with driving-force/inputs/control-signal

IoT - Sensor Fault Detection and Isolation Techniques using Fuzzy-Model

Why assume controls are independent of state estimation in Kalman Filter?

Since control theory is a mystery to me would someone like to confirm that. If they are I propose to re-tag them . It would then need a wiki.

Of the remainder there are quite a few about control groups and the rest about control variables. I think these are sufficiently distinct in questioners' minds for us to have separate tags. If people agree i will re-tag the control group ones and write a wiki and the remainder can be renamed as a block by someone with the right powers.

============ edit 6 February 2018 ================

This has now been completed with re-tagging and a new tag for as discussed in comments on the original question.

• Apologies for not following up on that. I still think that [control] is a bad tag name, and Adam's proposed wiki excerpt sounds way too broad for me. I think we could create [control-theory] if needed and rename [control] into [controlling-for-a-variable] or something like that. To begin, you could retag all Qs that are NOT about controlling for a variable. Is that what you were having in mind? – amoeba Jan 16 '18 at 15:40
• @amoeba yes, that is my idea. It will then be easier to see the wood for the trees. I hope. – mdewey Jan 16 '18 at 16:00
• 1,2,7 are not about control theory as far as I can see (I'd remove control tag from them). The rest are. – amoeba Jan 24 '18 at 12:52
• Also, I agree that [contol-group] and [controlling-for-a-variable] should be two different tags (or possibly not exist), and should NOT be lumped together in one tag. – amoeba Jan 24 '18 at 12:54

I think an ideal schism would be "control-theory" garnering its own tag, all other questions remaining as "control". The Kalman-filter is essentially a statistical, data analysis tool, hence on-topic for the site, although very few questions deal with this method. The Kalman-filter uses control in the sense of controls on a dashboard, external inputs which can be manipulated in a dynamic system.

While multivariate adjustment in observational studies and random allocation of control in experimental studies seem to have little relation, they are related. Both seek to reduce forms of variability and bias in analyses. It's unfortunate that "control variables" is still preferred terminology in a lot of experimental science. However, the implication is partially correct: you get efficient and unbiased inference by in some sense accounting for these variables, whether by stratification, adjustment, block-randomizing, weighting, or matching. If I had to give a Wiki about control, it would be something along the lines of,

Methods in experimental design and statistical analysis for reducing variability and eliminating bias.

• I think there are some which your definition would not cover like the ones falling under the tag [quality-control] but they could be dealt with separately I suppose. – mdewey Dec 16 '17 at 14:37
• I created the tag [control-theory] and re-tagged the ones which amoeba agreed were pertinent. If you know of any others the tag is there now. You might also want to edit my naive attempt at a wiki excerpt. – mdewey Jan 26 '18 at 15:23