CARLHAYDEN
Atlas Icon
Posts: 10,638
Political Matrix E: 1.38, S: -0.51
|
|
« on: September 08, 2005, 10:53:06 PM » |
|
Thank you for posting this thread.
First, I agree with you general points.
Second, as I previously noted, there are all kind of problems with using random telephone numbers, as I noted about fifteen months ago.
Third, I think we disagree about Democracy Corps, which to me has pretty consistently overstated Democrat preference in their polls.
Fourth, do you know the Mystery Pollster? He has some good stuff, IMHO.
Finally, it seems to me that the structural dynamics of opinion formation make polls on candidates/issues with very limited data bases (i.e. the respondents know little about the candidates/issues) highly unreliable unless this is factored into the analysis.
Specifically if an issue (to take an example) is favored by 50% in an early poll, and opposed by 40%, but among those who can answer factual questions about the issue the support is 40% and opposition 50%, bodes badly for the prospects of the issue to be carried.
|