[Goose]

I don't really know what ecological regression is and I have no basis of knowing how common or expected it would be to share your underlying data with someone who isn't in the same field, whose apparent "thing" is debunking the statistical method you used. I would imagine many professors would just ignore that email but my academic expertise is limited so maybe I'm mistaken.

You are probably correct that ignoring such a request for your data isn't uncommon, at least initially. It is probably also not uncommon to disregard criticism of your work from people you think are "unqualified" and are "out to get you".  I suspect MTL felt exactly that way about computer people questioning some of his publications. However, once a criticism is in the public domain, failure to follow up and debunk it with data is risky. If you believe the "debunking" of your statistics is incorrect, you should hoist the critics on their own petard. If you can't, it may come back to bite you. If you can't, you should also consider how strong your position is.

But probably not if the NY Post is what's being cited.

I totally "get it" that the NY Post has its "agenda". However, AFAIK nobody has disputed that Michael C. Herron and Kenneth W. Shotts did in fact lodge the criticisms cited. If they are totally off base, it should be simple for Gay to cite the work that debunks them. It may exist. I didn't find it, but I don't claim to be a statistician.

I do know that when statistics are used to "prove" an assertion, it is often necessary to examine whether the assumptions of a given statistical method are in fact met. I also know from experience that this step is often omitted or not done rigorously Sometimes it is very hard to do, as confounding factors that may not be obvious initially pop up after the analysis is "done". For that reason, IMO Gay should have co-operated and supplied the data, even after the fact. Better still, she should have pointed out where and why the critics were wrong. Yes, you don't have to do that, but if you don't and subsequently get in a public administrative role, it may not end well for you.

The problem is there are several general categories of the situation. The work may be basically "correct" and the criticism debunked. The work may arguably be in error, but it is at worst an honest difference of opinion. The work may also be knowingly flawed. Not providing the data or debunking the criticism makes it look suspiciously like the last alternative.

[Goose]

I don't really know what ecological regression is and I have no basis of knowing how common or expected it would be to share your underlying data with someone who isn't in the same field, whose apparent "thing" is debunking the statistical method you used. I would imagine many professors would just ignore that email but my academic expertise is limited so maybe I'm mistaken.

You are probably correct that ignoring such a request for your data isn't uncommon, at least initially. It is probably also not uncommon to disregard criticism of your work from people you think are "unqualified" and are "out to get you".  I suspect MTL felt exactly that way about computer people questioning some of his publications. However, once a criticism is in the public domain, failure to follow up and debunk it with data is risky. If you believe the "debunking" of your statistics is incorrect, you should hoist the critics on their own petard. If you can't, it may come back to bite you. If you can't, you should also consider how strong your position is.

But probably not if the NY Post is what's being cited.

I totally "get it" that the NY Post has its "agenda". However, AFAIK nobody has disputed that Michael C. Herron and Kenneth W. Shotts did in fact lodge the criticisms cited. If they are totally off base, it should be simple for Gay to cite the work that debunks them. It may exist. I didn't find it, but I don't claim to be a statistician.

I do know that when statistics are used to "prove" an assertion, it is often necessary to examine whether the assumptions of a given statistical method are in fact met. I also know from experience that this step is often omitted or not done rigorously Sometimes it is very hard to do, as confounding factors that may not be obvious initially pop up after the analysis is "done". For that reason, IMO Gay should have co-operated and supplied the data, even after the fact. Better still, she should have pointed out where and why the critics were wrong. Yes, you don't have to do that, but if you don't and subsequently get in a public administrative role, it may not end well for you.

The problem is there are several general categories of the situation. The work may be basically "correct" and the criticism debunked. The work may arguably be in error, but it is at worst an honest difference of opinion. The work may also be knowingly flawed. Not providing the data or debunking the criticism makes it look suspiciously like the last alternative.