On Thu, 28 Oct 2004, Miles Jackson wrote:
> Again, if the difference between the sample percentages is less than the
> margin of error, the poll's not "calling it" for anybody. People often
> make that mistaken assumption, but that's not a flaw in the research or
> the statistical procedure here. Any outcomes within the margin of error
> are feasible; thus the results don't necessary show a systematic error.
> (There may be one, but the evidence is ambiguous at best.)
You're exactly right, Miles. They don't *necessarily* show a systematic error. But they present a prima facie case for wondering about one -- esp. when considered in the light of the problems we are facing now, when the majority of the same polls, using the same methods, seem to be systematically oversampling Republicans to a comical degree. It makes one wonder whether whatever distorting force is causing that was at work to a lesser extent four years ago.
As you say, there may well be no connection between the two phenomena. They could both be false, and they could even both be true and have nothing to do with each other. But in an article whose entire raison d'etre is to speculate theoretically about possible sources of bias, I can't see how the lop-sidededness of the previous poll could avoid being theoretically speculated about -- unless the discussion was in bad faith, and the real aim was to damp all speculation. Which I think it was. Pollsters have a material interest in dealing with these issues in private. But they have even more of a material interest in not talking about them in public.
Michael