4
$\begingroup$

A purely mathematical question about linear regression was migrated to stats.SE.

https://stats.stackexchange.com/questions/14928/meaning-of-p-values-in-regression

  1. In this particular case, how did that happen? Is there a vote procedure, or did a moderator decide, or (something else)?

  2. In general, I understand that there may be some users in favor of instantly migrating statistics questions to stats.SE. I think that there should at least be some delay to see what direction the math.SE discussion takes. Some other opinions on this appear in comments below the migrated question. Question: is there a pre-existing consensus or policy on how to handle the statistics questions (where entirely mathematical)? If not, how are the migration decisions made?

[EDIT: the migration in this case was done after an answer was posted in math.SE. The question, one answer and some of the comments, were all moved together to stats.SE.]

$\endgroup$
  • 5
    $\begingroup$ At the least, it seems that the question shouldn't be migrated until we math.SE folks have had a chance to answer it. If the question receives a good answer here, why should it be migrated? $\endgroup$ – Mike Spivey Aug 29 '11 at 2:54
  • 4
    $\begingroup$ @Mike: because it sets a precedent for other people to ask statistics questions here, some or many of which may be better served at stats.SE. $\endgroup$ – Qiaochu Yuan Aug 29 '11 at 2:54
  • 3
    $\begingroup$ What is the problem with asking purely mathematical questions here about statistical procedures (in this case, the precise meaning and properties of p-values associated to linear regression)? Many such questions are not necessarily better-served on stat.SE, and the OP or others can always flag the post for migration in case of not getting a good answer on math.SE. But when a good answer appears immediately on math.SE isn't that a sign discussion should continue here, at least for some time? $\endgroup$ – zyx Aug 29 '11 at 3:09
  • 4
    $\begingroup$ It seems to me from its contents that this particular question rather belongs to stats.SE than here, so it seems reasonable to migrate it. I have no idea how large the intersection of user bases is, but those people on math.SE interested in the question can follow the migrated question anyway and answer it there, so I don't see a real problem here. In fact, I even think that migration is to the best of the OPs interests, as more people will see the question. If the OP wants the question to be here then s/he should say so and I don't think anybody will object. $\endgroup$ – t.b. Aug 29 '11 at 3:36
  • 1
    $\begingroup$ @Theo: the OP made comments about restricting the question to "keep it mathematical", i.e., on-topic for math.SE. He has posted 6 questions on stats.SE, so was well aware of the existence of two sites, and apparently did post with the intention of getting an answer from math.SE per se. The judgement of the content of the question and where it belongs (coming without an account on stats.SE nor answers to statistics questions here) seems to be that anything statistics-related does not belong here, not that mathematical statistics is better served there. $\endgroup$ – zyx Aug 29 '11 at 3:56
  • $\begingroup$ Yes, I don't have a stats.SE account nor do I feel qualified to answer statistics questions so I don't do that. However, the question clearly is statistics-related and yes, I do think that statistics-related questions should be on stats.SE by default. If the OP is aware of stats.SE and asks here because he seeks a more mathematical explanation then why doesn't he say so and make that more explicit? If that is the thrust of the last sentence of the question (and you may well be right on that) then it could be made clearer and more specific. The question can still be migrated back, I think. $\endgroup$ – t.b. Aug 29 '11 at 4:19
  • $\begingroup$ By the way: had I seen the question before it was migrated I would probably have left a comment suggesting that stats.SE might be a better place to ask this question and asked if the OP wants a migration (that's what I have done on previous occasions). $\endgroup$ – t.b. Aug 29 '11 at 4:25
  • 3
    $\begingroup$ One more thing: even after the reformulation of your comment there seems to be an insinuation that I have something against statistics questions. I object to that. $\endgroup$ – t.b. Aug 29 '11 at 4:37
15
$\begingroup$

Let's start by putting this question on a better footing. There is no evidence of "haste," the migration was not "instant," nor is the original question "purely mathematical." ("What do these p-values mean individually about those parameters?" asks for a statistical interpretation, not about the mathematics of regression.)

Timeliness in migration is a good thing. It can be confusing when a question accumulates comments and answers on one site and then is moved to another. That the regression question was identified and migrated within an hour of its posting testifies to the watchfulness and care of the math site's moderators. Let's not attribute their dedication and professionalism to "haste": that's an unfair, unwarranted characterization.

A sounder, politer way to raise the concern expressed in this thread might be, "help us understand how the math and stats communities make decisions to migrate questions between them."

This is something that high-rep members and moderators in both communities have to consider routinely. Ultimately each decision comes down to determining which site will serve the question better. The math site excels in obtaining rigorous, carefully qualified, purely mathematical answers to precise, well-framed questions. The stats site does well at explaining applications, interpreting results, and helping the OP find a way to ask the question they really meant to ask, but didn't know how.

Some questions sit on the boundary. This is natural, given the mathematical theory that imbues all of statistics and the existence of mathematical statistics as a branch of pure mathematics. In those situations, either site would be a good home. Thus it's not a problem if some of these boundary questions get moved to the other site. As a moderator of the stats site, when I encounter such questions I may make a quick search of both sites to see where similar ones have been asked, how many people have been interested in them, and to gauge the qualities of answers they may have gotten. I want the question to go where it will get a great answer, quickly, not just a "good" or adequate one.

Two aspects of a thread that would seem to be part of migration decisions are actually of little import. The presence of answers in itself means little. However, if either (a) a question has been around for a few days and has accumulated no answers or (b) a question is collecting answers that are clearly off the mark or less than helpful, then that suggests a need for migration. Popularity (measured in votes, views, or favorites) is irrelevant. A question can be clearly off topic on one site and still be popular there. Off topic questions ought to be migrated, closed, or modified to be appropriate for the site they're on. This of course is an ideal and in practice we don't always get it right, but we come close.

Incidentally, the migration process automatically moves all comments and answers along with the question. It has to: where else would this ancillary material go?

$\endgroup$
  • 1
    $\begingroup$ I agree with most of what you have written, but I think it better to leave it to the community and the OP as much as possible to decide where the question is better served, unless the question is really off-topic. Migration means that the question is off-topic. If we think an on-topic question can get a better answer on another site we can comment and inform the OP and if he/she feels that the answers haven't been satisfactory then the question can be migrated on OP's request or just cross-posted also on the other site explaining why the answers on the original site were not satisfactory. $\endgroup$ – Kaveh Sep 3 '11 at 6:14
5
$\begingroup$

If a question already has a (credible) answer here, I think moderators should rely on the voting system for migration, rather than migrating unilaterally. The fact that someone here answered a stat question is strong evidence that the question is on scope here.

$\endgroup$
  • 3
    $\begingroup$ I disagree. People here love answering questions, even if they're slightly to seriously off-topic. (That isn't necessarily a bad thing, but I'm not going to take the presence of answers as evidence that the question is more suitable to this site than to stats.SE.) $\endgroup$ – Qiaochu Yuan Aug 29 '11 at 15:41
  • $\begingroup$ @Qiaochu, are you saying no parts of statistics are part of math (i.e. they are completely off-topic on Math.SE) or just saying that the question is better served there? I would strongly disagree with the first one, but if you meant the second one (i.e. the question is not off-topic on Math.SE) then I think it is better to let the community or the OP decide where the question is better served. $\endgroup$ – Kaveh Sep 3 '11 at 6:05
  • 1
    $\begingroup$ @Kaveh: the latter. I'm willing to concede that it might be better for the community to decide, but I don't know what our policy is. Nobody's complained about migrations here before that I can remember. $\endgroup$ – Qiaochu Yuan Sep 3 '11 at 6:11
  • $\begingroup$ @Qiaochu, I see, I didn't meant it as an objection to your decision but as having the discussion for the future decisions. :) $\endgroup$ – Kaveh Sep 3 '11 at 6:16
4
$\begingroup$

You can simply read the original question to see who migrated it -- click the "from math.stackexchange.com" link to go back to the source:

https://math.stackexchange.com/questions/60425/meaning-of-p-values-in-regression?noredirect=1

migrated to stats.stackexchange.com by Qiaochu Yuan♦ 7 hours ago

This question belongs on our site for statisticians, data analysts, data miners and data visualization experts.

$\endgroup$
3
$\begingroup$
  1. For migration to meta and to stats.SE there is a voting procedure (it's an option under voting to close), but in this particular case I was responsible for the migration.

  2. I'm not aware of one. If I see a question that I believe the community at stats.SE is capable of giving a better answer to than the community at math.SE, I migrate it.

$\endgroup$

You must log in to answer this question.

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