Presidential election polls, particularly those showing unexpected trends, are almost invariably accompanied by ridiculous amounts of spin and over-reaction, and that’s looking to be the case for three new national polls (CBS/NYTimes, NPR and Fox) and one state poll (Quinnipiac in VA) released yesterday.
The national polls are a mixed bag, and all show a race within the margin of error, but Republicans are crowing over the CBS/NYT survey, showing Romney actually up by a point among RVs, amidst signs of deteriorating personal favorability for Obama and renewed economic pessimism. Taken at face value, the results would seem to cast some doubt on the belief that the attacks on Romney’s Bain Capital background and tax records, and his poor handling of them, are having a significant effect on public opinion. That’s even more true with respect to the Q-PAC poll of VA, where heavy advertising has given a thorough airing of the Obama campaign’s attack lines; yet Obama’s long-standing lead in the state has, in this one poll at least, vanished.
To put all this in context, however, I recommend you read two pieces by National Journal‘s Steven Shephard. The first, on the new national polls, notes the oddity that the NPR survey showing Obama still narrowly ahead deploys a “likely voter” screen (more about those later), which typically skews results a bit towards Republican candidates, while the CBS/NYT poll questions RVs from an overall sample of “all adults.” Perhaps most interestingly, Shephard suggests rapid erosion of Obama’s personal favoribility in the CBS/NYT poll heavily depends on a methodology that makes it easy for respondents to say they are not sure how they feel.
Shephard’s analysis of the Q-poll of VA is worth a read as well. The last Q-poll in the Commonwealth showed Obama up by 5%; the new findings appear to represent a combination of (a) a drop in Obama’s support levels among white voters, and (b) a sample where white voters make up a notably higher percentage of the electorate. This latter factor could make the survey a bit of an outlier, but in any case is a reminder that many so-called “shifts in public opinion” you hear about actually just reflect polls with different takes on the composition of the electorate.
And speaking of “different takes,” anyone paying attention to the back-and-forth spin on poll results should definitely read Mark Blumenthal’s refresher course at HuffPost on likely voter screens (which often produce better results for Republicans), and why most pollsters are reluctant to use them before relatively late in the cycle.
The bottom line is that poll averages remain the most reliable indicator of the contest, and the apparent lack of evidence of any big impact from the Bain/tax attacks on Romney is a reminder that voters do not follow these “stories” much at all until they have been aired repeatedly. I continue to believe the most important aspect of the Bain/tax issue is its utility as a set-up for later criticism’s of the Romney/GOP policy agenda as expressed in the Ryan budget.
