# Are questions on statistical language valid and do we need statistical-language tag?

I am referring here to recent question: versus (vs.): how to properly use this word in data analysis that asked about proper usage of "versus" word in describing relations between variables in data analysis.

For me such a question is off-topic since it is about English language usage rather than statistics. My argument is that if the question would ask about describing the same things but in language other than English it would considered be bluntly off-topic (and possibly unanswerable for most of us).

Questions about word usage in statistical reports were closed several times before as off-topic, while the one I refer to seems not to provoke votes for being off-topic. So my question is: are this kind of questions off-topic or not? What are the criteria? And if this kind of questions are not off-topic, then maybe we need [statistical-language] tag as poster of the linked question suggested..?

• BTW there's a terminology tag which is already being used. – Scortchi - Reinstate Monica Apr 15 '15 at 20:30
• @Scortchi good point! However, the particular kind of questions as the one I linked somehow does not seem to fit for this tag - isn't it? Maybe a synonym tag is needed..? – Tim Apr 16 '15 at 6:17
• Why not? Google gives the definition:-"the body of terms used with a particular technical application in a subject of study, theory, profession, etc" & the tag wiki excerpt is:-"Indicates questions asking about the use and meaning of specific technical words/concepts in statistics". Both seem to fit with this q. (nomenclature is already a synonym - if people are creating different tags with similar meanings we can add to the synonym list.) – Scortchi - Reinstate Monica Apr 16 '15 at 10:14

Technical statistical usage sometimes departs from colloquial English usage. (Examples abound, such as the transitive use of "to learn" in the ML community and the use of singular nouns among statisticians to refer to entire families of distributions. And that's not even touching on the severely technical meanings of "confidence," "error," "fit," "probability," "sample," and so on! Statistics is a foreign language, no matter what your native tongue might be.)

Thus, to the extent a question might require statistical/ML expertise to understand or answer, and to the extent its answer might differ from colloquial usage, I think such questions are on topic. As a moderator, all I ask is that any answers supply objective arguments and refer to authoritative references for support.

• Isn't "to learn" transitive even outside ML? E.g. "to learn a poem". – amoeba Apr 15 '15 at 21:39
• @amoeba That is correct. However, "learn" is used both as a transitive and intransitive verb. The meaning "to acquire knowledge or skill" is intransitive, but this also seems to be the sense in which "learn" is meant even when it is used transitively in the ML community. The discord in phrases like "learn a hypothesis" or "learn a classifier" is jarring. – whuber Apr 16 '15 at 15:00

I've argued the other way in the thread cited, both by commenting against this point of view and by answering the question.

But we can agree that questions such as these do seem to straddle a border. The question is how much of the question is on each side.

My position is pragmatic. If a usage question is only in small part about meanings or senses that are statistical, then I agree that it belongs elsewhere. Where the primary thrust of the question is statistical, and we can reckon that the answer should be of use or interest to many people here, then it should stay. Thus a question about cause, say, would have some statistical flavour, but more crucially is common to many sciences, and indeed disciplines, so the grounds for keeping it would not be clear; but this question to me seems firmly in our camp.

The fact that the question is phrased in English is incidental here; that's the language of the forum any way.

The general question:

I think questions that are about statistical terminology (the correct use of statistical jargon) are on-topic.

If someone asks about "what does significance really mean in statistics?" then that's right on topic here, in my book. [If they ask about, say, the meaning of the word "linear", that's strictly a mathematics question but should probably be tolerated here because a complete answer relevant to statistics would highlight the use of the constant term (I think homogeneous coordinates may be the term for that) in linear models, which an ordinary mathematician's explanation might not bother to mention, but will be central to most statistics usage. Some judgement is required as to the degree of relevance.]

However, some questions ask about words or phrases that are just about ordinary English usage. They belong on english.SE (generally not ELL, unless it's clearly something that would be understood by an ordinary user of English).

The somewhat more specific question, and the issue that it raises:

There was a recent question about the statistical meaning of versus.

I believe the premise of the question is false -- there's no particular technical meaning of "versus" in statistics -- in statistical writing it's being used in its ordinary English sense (or, perhaps arguably, in some situations its ordinary mathematical sense), though sometimes subject to fairly ordinary English mistakes, or sometimes common habits in particular areas; that it might be used or misused in a specific way doesn't alter that it's English usage rather than a technical term (or at least not a stats one). The mention of this premise means the question falls in neither of the situations I mention.

Because the question carries the premise that it's a technical term (whether right or not), english.SE shouldn't accept it (they're not in a position to judge whether it's a technical term in stats), so it shouldn't be migrated by us. There may be some argument it really belongs on math.SE but I wouldn't lean that way myself.

So -- assuming it's decided that the premise is mistaken, which is certainly open to some debate -- the question then is 'does a question about a mistakenly-supposed technical term remain open here?'

Here's why I think that's a matter of judgement.

If the impression that the word is statistical jargon with a specialized meaning is very uncommon -- something idiosyncratic to the OP or their experience made them post -- then the question is unlikely to help other users who come here, and should probably close as off topic (preferably with a comment that it takes its plain English meaning so the poster understands where to ask further questions about the usage of the word absent the premise).

However, if it's something people might come to the same conclusion about then it should remain open here, and be answered. In addition, where there's uncertainty, the impulse should lean toward erring on the OP's side (i.e. interpreting the question somewhat generously).

I lean toward saying that the default position should be that such a question should remain open here, but case by case decisions would be needed. It's certainly one I'd prefer not to use my diamond on*, unless I was adding a fifth vote.

* (I wish I could take that diamond off and just vote as an ordinary user in such cases)

[In the particular case of "versus", I think Nick's answer in particular makes the question clearly belong here.]