Whether they are trying to lay the foundation for a “stolen election” claim, or an enduring “conservative majority” myth, or are simply trying to cheer themselves up, the growing tribe of Poll Denialists rely on an awful lot of dubious or downright ignorant assumptions in dismissing polls they don’t like as biased.
Here are three, just in case you encounter one of these birds and want to go to the trouble of trying the art of persuasion:
1) “Weighting.” Most denialists assert or assume that pollsters showing a significant pro-Democratic turnout gap are “weighting” the sample for partisanship before or after conducting the survey. TNR’s Nate Cohn demolishes this one pretty thoroughly today:
Most pollsters don’t weight their polls to match a preconceived electorate. Instead, they take a demographically representative sample based on actual figures from the US census and then let respondents speak for themselves about whether they’re voting for Obama or Romney. For illustrative purposes, consider the Bloomberg/Selzer poll. They started by taking a sample of all American adults, weighted to match the demographics of all adults in the US census, like, race, education, and marital status. To produce a likely voter sample, they then would have excluded adults who weren’t registered to vote and then asked a series of questions to help determine who was likely to vote.
As Nate explains, whether or not someone voted in 2008 is typically part of (but not the totality of) the test for whether a registered voter becomes a “likely voter” in a given poll, but there’s no “2008 turnout model” that guides the initial sample, which is based on census numbers, not prior election results. It’s the Denialists, not their targets, who are into “weighting,” because they demand a partisan distribution that resembles their version of the “true” divisions in the country.
2) Midterms Versus Presidentials: A lot of Denialists observe that the partisan distribution of voters in current polls resemble that of 2008 rather than 2010, and consider that a “proof” of bias, since 2008 was a “Democratic wave election” and 2010 is a “Republican wave election,” and a more recent one at that. They typically ignore the eternal difference in turnout patterns between midterm and presidential elections, which varying significantly regardless of any other factor (e.g., this or that politician’s or party’s popularity, the economy, etc., etc.). Midterm electorates are much older and whiter than presidential electorates, which has become particularly significant in the last couple of cycles as the two parties increasingly reflected a country polarized on age and ethnic lines. To put it another way, the day after the 2008 election Republican midterm gains became exceptionally likely, and the day after the 2010 election the winds shifted and Democratic prospects for 2012 improved. “Reweighting” polls to pretend away that demographic shift makes no sense at all.
3) Partisan ID Variables: The reigning assumption of Denialists is partisan ID is a zero-sum game that translates directly into a horse-race advantage. This ignores the underlying partisanship of half-or-more of self-identified independents. A combination of “brand erosion” for the GOP since 2006 and the self-conscious (if ultimately meaningless) independence of Tea Party folk has meant that an increasing number of reliable Republican voters are now identifying themselves as indies. This has two superficially confusing effects: Democrats obtain an ID advantage over Republicans, while Republicans do better among independents. None of this involves a distorted sample. “Reweigthing” samples to make Rs and Ds equal (or even to give Rs an advantage, as Rasmussen does) double counts an awful lot of conservatives who vote R but identify indie.
There’s another myth in circulation in the conservative echo chamber that is not so much a denial of the accuracy of polls, but a challenge to their predictive value: wildly exaggerated versions of the dubious “incumbent rule” holding that undecided voters always break decisively against incumbent candidates like Obama. As Dave Weigel pointed out yesterday, Dick Morris pursues this delusion to a particularly ludicrous extent.
None of this should matter after Election Day, but again, to the extent that a losing GOP can be expected to claim that adverse polls (plus “voter fraud” and perhaps Romney’s failure to run a sufficiently ideological campaign) affected the actual results, it’s worth exploding the Denialist case early and often.
Feed the Political AnimalDonate
Washington Monthly depends on donations from readers like you.