Cherrypicking Hypothetical Polls
Posted on Wed Sep 26, 2007 at 12:14:48 PM EST
Tags: (all tags)
While decrying what he terms "poll literalism," Kevin Drum then engages in a curious bit of poll hypotheticalism:
First, withdrawal from Iraq. A recent New York Times poll showed that 65% of respondents want to withdraw either some or all of our troops from Iraq. Hooray! The country is with us! But then the Times asked a followup question: "What if removing troops meant Iraq would become more of a base of operations for terrorists, then would you still favor removing U.S. troops from Iraq, or not?" Guess what? Of that 65%, only 30% still favored removal. That's a huge drop based on a single hypothetical, and in a real campaign that hypothetical would practically blanket the airwarves. It wouldn't convince everyone, of course, but it would probably convince a sizable chunk. The odds are that in real life — i.e., during a campaign
It would probably convince a sizable chunk? How does Kevin know this? Excuse me, this is ridiculous use of polling. Can we do one with a Democratic hypothetical and see what the results would be? Of course not. What Kevin is pointing to is actually push polling. Push polling is not ACCURATE polling and it is impossible to draw any conclusions from it. But Kevin sees value in it, as he then defends the Biden push poll that Celinda Lake performed:
The reason it was derided (aside from the fact that Washington Post reporter Chris Cillizza failed to inform his readers that Biden was behind the poll) was because of the wording of one of the questions: "Some people say [your Democratic incumbent] is a strong supporter of Hillary Clinton and will support her liberal agenda of big government and higher taxes if she becomes president," the poll stated, before asking respondents whether they would still vote for their incumbent or choose a Republican candidate. Outrageous! And it is. On the other hand, that's exactly what Republican House candidates are going to say, isn't it? Which means that this poll, showing a 6-point lead for Democratic incumbents, is probably more useful than generic polls showing a 10 or 15 point lead.
Republican candidates say a lot of things. For example, in 2006, they said "cut and run" 8 million times regarding the Iraq issue. They said the very hypothetical Kevin has such confidence in from the Times poll.
This is weak analysis from Kevin on ow polling works and frankly, well below his standards. Bad show.
| < Warren Jeffs Convicted | Open Thread Diary Rescue > |
