Over on Polipundit, I have published a Guide to the National Polls for the 2004 Presidental race. Since I am trying to be objective over on Polipundit, I did not grade the polls, except to make a few remarks about their more distinctive qualities. Here, I'm going to grade them, red marker pen in hand. Before I do, here's the scale:
A : Top Quality, solid work and great reports, I'd pay for this kind of information
B : Good work, good report. Not complete as I'd like, and maybe makes a few assumptions.
C : Satisfactory. Nothing to complain about, but not very impressive, either.
D : Needs Improvement. They need to show their work, and their report needs more information.
F : Failure. Such shoddy work, I wouldn't use their report if they paid me to take it.
OK, now on to the polls:
ABC News/Washington Post: Good data, cumbersome format for their report. C+
American Research Group: Methodology not explained, poor detail. D-
Associated Press/Ipsos : Good data, solid methodology. Reports need to be easier to access, and demographic voting results should be published. B-
Ayres McHenry: Poor detail, no methodology, partisan agency. F
Battleground Poll : Good methodology, details unsatisfying. C-
CBS News, and CBS News/NY Times : Excellent detail, extensive reports, convenient and consistent. A+
Democracy Corps : Partisan Propaganda, but at least there are some details in their reports. D+
Fox News/Opinion Dynamics : good detail, reasonable methodology, but could and should improve on both counts. B
Gallup: Solid data, great details. A
Harris: decent methodology, but does not release necessary details. C-
Investor’s Business Daily/Christian Science Monitor: No data, no support, all pretense. F
Investor’s Business Daily/TIPP : No data, no support. F
LA Times : Poor detail, methodology not established. D
Marist College Institute for Public Opinion: No methodology cited, no details. D
NBC News: Good data verification, reports do not have much detail. C
Newsweek : Excellent consistency, detailed reports. A
NPR-POS/GQR : Small sample base, few details, suspect firm with ties to partisan group. C-
Pew Research Center: Solid information, excellent detail. A
Quinnipiac University: Few details, no details on methodology. C-
Rasmussen: Many polls, useful for daily tracking, but no details or methodology. D
Survey USA: Lots of state data, but no demographics. Has a lot to say, but some of it doesn't stand up to inspection. B-
TIME : Tracks trends well, report a little incomplete, but consistent. C+
Wall Street Journal: Co-sponsors with other polls, has no known standards for weighting or methodology of its own. D
Zogby: A biased poll spokesman, non-random respondent pooling, no explanation of his weighting, and results which often are out of synch with other national polls. F