Just fired this off. Doubt it will do any good, but you never know.
I included relevent links to DonkeyRising, etc., but took them out for purposes of posting here. Just so you don't think I am plaigarizing =) We all know this story already...
Dear Mr. Halperin and various Googling Monkeys:
Since you at The Note are the pulsetakers of the national media, I think you owe it to them to Note the potentially significant problems with the recent Time, Newsweek and Gallup polls.
As you may be aware, the Newsweek poll significantly oversampled Republicans, yet did not use weighted results to account for this. If
they had, they would have found Bush up by 5 points, not 11. The Time and Gallup polls also oversampled Republicans, it appears. Kerry is
actually leading among independents, according to Gallup's internals, and he and Bush have identical horse race numbers among their fellow partisans. The only rational conclusion, therefore, is that Gallup oversampled Republicans -- how else could they find Bush ahead overall?
Back in June, the L.A. Times published a poll that oversampled Democrats, resulting in an unrealistic Kerry lead. You folks duly Noted that for several days running, and discussed the pros and cons
of using weighted polling results. I think you would do well to revisit this and ask the following questions in the wake of these
latest polls:
1. Following the L.A. Times poll, the Bush campaign made a concerted effort to discount it, with Matt Dowd calling it "a mess," forcing
the Times to issue a statement in its own defense. We see none of that from the Kerry campaign this week; instead, liberal blogs are doing the heavy lifting. Why is this?
- Why is the media obsessed with the horse race, and why does it highlight the likely voter Gallup model (Bush +7) and all but ignore the closer registered voter model (Bush +1)? Why did it all but ignore the polls that were taken at the same time as Time and Newsweek's, ARG and Zogby, both of which showed a much closer race?
- With the rising use of cell phones, the Zogby online polling experiment and the Rasmussen robo-calling technique, is it time to look again at the science and future of polling?
I read The Note every day, but too often grow
irked at the sense that, in fear of being labeled "liberally biased" by your conservative readers, you overcompensate and hew too closely to Republican talking points. Your documenting the possible problems with these polls would be one way of proving my theory wrong.