The conservative Power Line blog today has been vigorously attacking the CBS/New York Times poll showing overwhelming support for a public option — criticism that was repeated on TV today by Sen. John Cornyn (R-TX). But a close examination shows that their supposed deconstruction of the poll doesn’t actually have any merit.
Power Line says that this poll is worthless, because respondents were asked who they’d voted for in 2008, with the answer coming up as Obama 48%, McCain 25%. “Since Obama won the election by a 53%-46% margin, the poll obviously skews left,” they say.
A big problem occurred to me, though, one known to anyone who’s read about polling for a long time: The who-you-voted-for question is good for a lot of things — except for finding out how people actually voted. It’s really an indicator of people’s willingness to say they voted for the incumbent (regardless of whether they’re telling the truth) or to say they voted for the challenger. All this poll really tells us is that some people are eager to say they voted for Obama, and others won’t readily admit they voted for McCain.
I checked with Prof. Larry Sabato, and he agreed with me that this is not in any way a good reason to impeach the credibility of the poll — though he did add a caveat that due to the ongoing complexity of the health care debate, the listed support for the public option might not be reliable, either!
Check out Sabato’s comment, after the jump.
Eric–You are closer to the truth. While the winner is popular, there is almost always a bump for the winner in self-reported responses. Forget about Americans rooting for the underdog. Once the underdog has clearly lost, they go for the winner disproportionately. The car bumper sticker test proves it. The winner’s stickers are almost always left on for months, while the loser’s stickers are scraped off the day after the election. A famous example, albeit exaggerated, came in a poll (Gallup, I believe), taken in 1964. Respondents were asked for whom they voted in 1960. Turns out, Kennedy had won a massive landslide, well over 60%, instead of a 119,000-vote squeaker victory and a shade under 50%.
I said you were ‘closer’ to the truth. Here’s why. Health care is a complicated subject except for wonks. Every word in a polling question matters, since many people are literally being informed by the question itself. That is, they don’t know enough about the subject of health care reform and the various alternatives to have formed an opinion. The question itself can create public opinion. So we need to be very cautious in interpreting polling on the subject. For years I’ve said, let me word the questions and I’ll get you any answer you want.
