Rolling with the Polling
|GOP (R)||Dems (D)|
|ABC News /Wash Post||45||51|
|USA Today /Gallup||44||51|
Whatever the outcome of the Nov. 7 midterm elections, they’ve been a real boon for pollsters and math fans. In the last few days alone, there have been seven new “generic Congressional” polls released, surveys that ask prospective voters whether they’ll choose the Democratic or Republican candidate in their district.
The polls—from Fox News, CNN, USA Today/Gallup, ABC News/Washington Post, Newsweek, Time, and the Pew Research Center For The People And The Press—all show Democrats leading Republicans, but by widely varying margins ranging from four points up to 20. The average difference is about a dozen points.
Of course, they’ve received wide play in the media. On Nov. 6, the front pages of USA Today, The New York Times, and The Washington Post all had the word “polls” displayed prominently in front-page headlines or subheads. In post election post -mortems that inevitably address media coverage, one oft-heard complaint is that journalists pay excessive attention to the strategic horserace aspects of a campaign at the expense of the underlying issues. This year, there are a number of reasons why polls seem to be such a dominant part of the campaign narrative.
For one, the basic media story line is relatively uncluttered by complexity. The election has been portrayed largely as a referendum on one man, George W. Bush, and on one main topic—the war in Iraq. (The exit polls should tell us whether that was essentially accurate or simplistic.) Another reason for the proliferation of polls is the sheer size of the battlefield in this fight for control of Congress, one which stretches from Florida to Montana and includes dozens of House seats and about 10 Senate slots. Adding to that is the sense that many of these contests, even in their final hours, are fluid and too close to call. And besides, phrases like “Tennessee Toss-Up”—which is how CNN has been characterizing that Senate fight—seem to practically roll off the tongue. A third is the expense and logistic impossibility of doing carefully drawn accurate polling in all the contested races.
In the meantime, the polls just keep coming. Aside from the generic Congressional question, a review of two political web sites – Real Clear Politics and Electoral Vote Predictor – turns up more than more than 90 polls (some by the same companies) released since Nov 1. looking at various contested Senate, House and gubernatorial races. With this much variety, one thing is probably a mathematical likelihood. One of them is bound to be right.