I posted my question on stats.se yesterday but it was moved to maths. I actually thought about putting it on maths.se first, but since it was mathematical statistics, I thought I would post it here. I can understand there is a grey area, and probably my question /is/ better suited on maths.se because it can (I imagine) be answered by a mathematician who doesn't know anything about statistics. Is that the right way to view it ?

On the other hand, I also thought that, since there is a conditional distribution involved (which I think is irrelevant to the answer, but not completely sure about that) it might get better answers on stats.se ? FWIW, the answer I got on math.se isn't completely satisfactory to me (no disrespect to the answerer of course, I'm pretty sure it's trivial to him/her and it's just me who can't see the wood for the trees, and will hopefully come back later with a follow-up), which also makes me wonder if it might be better here....

  • 3
    $\begingroup$ Hi Joe - it seems to me that the true spirit of your question is how/if you can differentiate under the integral sign and I agree with you that the fact that the integrand involved is a density function is not particularly relevant. Thinking of it that way, there's not really any statistical content, so I'd tend to agree that the move to math.se was the right thing to do. Also, note that the person who answered your question is a regular poster on the stats.se so you might've gotten the same answer either way :) $\endgroup$ – Macro Jul 14 '12 at 13:06
  • $\begingroup$ @Macro thank you. It was the conditional part that I didn't think was relevant. I thought it was relevant that it involves and distribution and density, because of the relationship between them. I want to differentiate the CDF and get a result involving the PDF. Of course the relationship between them is purely mathematical, so I do take the point. $\endgroup$ – Joe King Jul 14 '12 at 13:24

Yes, that's the right way to view it: questions are migrated to sites where they are (a) on topic and (b) likeliest to get a great answer in a timely fashion. Sometimes, as in this case, the very act of migration gives a clue as to how the question can be answered!

Your moderators frequently have to field questions of mathematical statistics, or other theoretical questions, that ultimately have purely mathematical answers. My philosophy as a moderator is to keep the question here if it or any of its answers can be construed as having some statistical interest (such as in using statistical thinking to motivate a particular line of attack). If the question otherwise is (for example) purely one about how to perform a particular mathematical manipulation, such as (in this case) how to apply the Chain Rule or (as in other common cases) how to apply optimization software, then migration seems the best course of action.


You must log in to answer this question.

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