Thursday, July 13, 2006

The Generic Ballot

Mark Blumental, the Mystery Pollster has an interesting discussion (with many links) about the value of the generic ballot polling question. You know the one that asks if the election were held today, would you be more likely to vote for the Democratic or Republican candidate. He points to a poll done in three congressional districts by Democracy Corps, the outfit run by the dynamic trio of Stan Greenberg, James Carville and Bob Shrum. What is interesting about this poll is that, in the midst of a bunch of other questions of likely voters about the respondents views of Bush, the Congress, the Republican and Democratic parties, the pollsters asked the generic question. With no names mentioned, things do not look good for the Republican incumbents in these three districts.
Generic Ballot
NJ 07
D 45
R 44

PA 07
D 51
R 39

OH 01
D 47
R 44

So, if you just looked at the generic poll number for these three districts, you'd think that those incumbent Republicans congressman are in a lot of trouble. They may well be. But, look at how the numbers switch when the respondents are given the names of the candidates.

NJ 07
Linda Slender-D 41
Mike Ferguson-R 50

PA 07
Jo Sestak-D 41
Curt Weldon-R 51

OH 01
John Cranley-D 43
Steve Shabot-R 52

This represents quite a switch from the generic ballot. It could be that this represents the power of incumbency and that likely voters respond to the name that they're familiar with. Or it could be that when they're just asked out of the blue if they're going to vote for the Democrat or Republican, they don't plug in the names of the candidates in their district, whom they might not even know, but they think of a generic Republican, say George W. Bush and a generic Democrat, whether it be Bill Clinton, Al Gore, or Ted Kennedy. And when Bush is unpopular, their responses lean Democratic. Or it may just be that this far out ahead of the election, the question is meaningless

Blumenthal, who is a Democratic pollster, but quite fair and nonpartisan in his analyses, points to comments made by Gallup pollsters about the cushion for Democrats in their polls.
In another report released in February, Gallup's David Moore put it more plainly: "Our experience over the past two mid-term elections, in 1998 and 2002, suggests that the [registered voter] numbers tend to overstate the Democratic margin by about ten and a half percentage points." Similarly, taking a somewhat longer view ("most of the last decade") Gallup's Lydia Saad reported last September that "the norm" a five point Republican deficit among all registered voters that "converts to a slight lead among likely voters." Make "some adjustments" to the generic vote, she wrote, and one can "make a fairly accurate guess about how many seats each party would win."
So keep all those considerations in mind the next time you hear gloom and doom about how the Republicans are going to do this fall based on some generic ballot poll question, don't start running for the exits. When you figure in the Democratic skew and the margin of error, you're really not receiving as much information as the way that the media that write up the polls imply. At this point out, those results don't mean all that much.

A couple of months ago, Jay Cost argued against placing much credence in the generic ballot number because he maintains that it has very little prognosticative value this far out. Cost makes several acute points, but one that we should remember is that the average voter does not think of their vote as part of a vote for partisan control of the Congress. They are more likely to vote on the individuals than on whether their vote will turn the House over to Nancy Pelosi or leave it in the hands of Denny Hastert. Political junkies like myself and my readers might think that way, but then again, we're not the average voter.