CBS/NYT/Quinnipiac Swing State Polls & Party ID

By Sean Trende - August 1, 2012

‹‹Previous Page |1 | 2 |

Moreover, party ID is sensitive to question ordering. CBS/NYT/Quinnipiac asks about party ID after a battery of questions, including ones about the state of the economy; whether Obama and Mitt Romney care about the needs of people; who would do a better job on health care; and so forth. The exit polls ask the party ID question much earlier in the process, immediately after the “how did you vote” questions (which are upfront). All of this has an effect on how people think of themselves. In particular, the health care and “cares about you” questions can prime respondents to think of themselves as Democrats.

Moreover, while the sample plausibly over-samples Democrats, it doesn't seem to be overly liberal. This is something that is more consistent over time and that is less susceptible to question wording and ordering. The following chart shows the percentage of self-identified conservatives, liberals, and moderates in each state in the 2008 and 2010 exit polls as well as the recent CBS/NYT/Quinnipiac polls:

As you can see, here the results are actually closer to the 2010 electorate, with the exception of Ohio. This suggests that the ideological orientation of the surveys isn’t particularly skewed (we’ve seen other polling suggesting that somewhat more people self-identify as conservative), but that Romney is doing unusually poorly among self-identified moderates. In addition, Obama’s approval rating in the polls is broadly consistent with his national approval ratings -- around 48 percent in each state. (I’d also note that I’m unimpressed with critiques about the McCain/Obama responses: More people tend to recollect voting for the winner than actually did.)

Overall, my recommendation is simply not to pay much attention to non-demographic sample issues.  Instead, I’m drawn to two competing explanations here. First, random sampling error can create outliers, and it is possible that one of these results is just an outlier (in fact, it is possible that all three are outliers, just very unlikely).

Second, survey design, especially the choice of a likely-voter screen -- and choices in how hard to “push” undecided voters toward giving an answer -- can create quite different results in polls. Remember, these are the first polls where Quinnipiac pushed undecideds and screened for likely voters, so they are the first polls where we’ve seen any “house effect” that might be present in its likely-voter screen, or that was present in its decision not to push undecideds.  It's also the first time Quinnipiac partnered with CBS/NYT, and its methodology may have shifted a bit from earlier iterations.

In all three states polled, the RCP Averages (Florida, Ohio, Pennsylvania) include at least one poll of likely voters from a nonpartisan source that is roughly consistent with the CBS/NYT/Quinnipiac result. They also include polls that are not consistent with the CBS/NYT/Quinnipiac result. Overall it looks like CBS/NYT/Quinnipiac’s system places it on the more pro-Democratic side of the “house effect scale,” but not outrageously so.

Either that, or it is a harbinger of polls to come. 

‹‹Previous Page |1 | 2 |

Sean Trende is senior elections analyst for RealClearPolitics. He is a co-author of the 2014 Almanac of American Politics and author of The Lost Majority. He can be reached at Follow him on Twitter @SeanTrende.

Mitt Romney for Mayor
Carl M. Cannon · November 16, 2014
A President Who Is Hearing Things
Richard Benedetto · November 12, 2014
Bret Stephens' Call for Robust U.S. Foreign Policy
Peter Berkowitz · November 16, 2014

Latest On Twitter