Yesterday, the New York Times and CBS News (NYT/CBS) released a new poll showing broad support for health care reform, with 72 percent of respondents favoring the creation of a publicly-funded health insurance option. The conservative blog Powerline immediately took issue with the poll, arguing (wrongly) that the sample was skewed because 48 percent of respondents reported voting for President Obama last fall, while just 25 percent of respondents reported voting for Sen. John McCain (R-AZ). Powerline compared the NYT/CBS figures to the actual election results in which Obama won 53 percent of the vote and McCain won 46 percent.
Sen. John Cornyn (R-TX), apparently convinced by Powerline’s argument, cited the blog in two cable news appearances this afternoon to deny that there was any significant public support for the creation of a public health insurance option. “With all due respect to the New York Times and CBS, this polling sample was skewed,” he told MSNBC’s Andrea Mitchell. Similarly, on Fox News Cornyn said, “I think there’s been some particularly good blog coverage like Powerline blog talking how that sample was so skewed as to be meaningless.” Watch it:
Unfortunately for Cornyn, Powerline is wrong to conclude the sample is skewed based on the data they cited. As Slate’s Christopher Beam explained last week, the disparity between last fall’s actual vote tallies and the results reported by NYT/CBS yesterday comes down to respondents being too embarrassed to admit that they didn’t vote:
The main explanation for the gap, say pollsters, is people who didn’t vote at all saying they did. These people tend to say they picked the winning candidate. Just look at the Times and Journal polls, where about 80 percent of respondents said they voted in the 2008 election. In fact, turnout was about 61 percent. (A 20 percent gap is pretty standard.) Pollsters attribute the disparity to the social discomfort of having to admit, even to a stranger on the phone, that you didn’t vote.
Further as Beam explains, “Retroactive vote reporting tends to be a proxy for popularity. … In a 2006 NYT poll, more people said they voted for John Kerry in 2004 than voted for Bush.” If Powerline wanted a more reliable indicator of who was in the NYT/CBS sample, they could have looked at the proportion of respondents that identified themselves as liberal (27 percent) and compared that to the proportion that identified themselves as conservative (29 percent). Likewise, Powerline could have noted that the sample was 24 percent Republican and 38 percent Democrat — a fairly normal party identification advantage for Democrats at the moment.
To buttress their claim that the NYT/CBS poll was inaccurate, Powerline linked to a recent Rasmussen poll that found comparatively little support for the creation of a public health insurance option, with just 41 percent of Americans supporting such a move. But as Nate Silver documented last week, it is the Rasmussen poll — not the NYT/CBS poll — that falls outside typical levels of support for a public health insurance found in other recent surveys:
Powerline’s concerns about the accuracy of the latest NYT/CBS poll are unfounded, but that likely won’t keep Cornyn from continuing to refer to Powerline’s discussion as “particularly good blog coverage.”