I have spent a lot time over the last few days pouring over the U.S. House polling numbers on Pollster as we have been at work on some sort of summary scorecard. But before we plunge back into the micro, district-level analysis, I thought it would be useful to do a bit of a review of the way things look at the national level. We do seem to be facing a surge of voter discontent with the Bush administration and the Republican Congress, but forecasting how that surge will impact the race to control the House is something of a puzzle.
Last week, I spoke at a forum sponsored by Campaigns and Elections magazine and was fortunate to share the stage with Democratic pollster Mark Mellman. He summed up this puzzle with a metaphor:
There's a big anti-Republican wave out there. But that wave will crash up against a very stable political structure, so we won't be sure of the exact scope of Democratic gains until election night. We really don't yet know which is ultimately more important -- the size of the wave or the stability of the structure.
So let's take a closer look at what we know now about the size of the coming wave and the stability of the structure.
Generic Ballot. National pollsters measure Congressional vote preference with a question commonly known as the "generic ballot," usually some variation of the following: "If the election for U.S. House of Representatives were held today, would you vote for the Democratic or Republican candidate in your congressional district?"
The goal of the question is to measure the total national vote cast for Congress. The good news for Democrats is that they have been leading on the generic ballot by margins of 10% or more since August, with a recent bump up in the aftermath of the Foley page scandal. However, the Foley bump is just icing on the Democratic cake. When my colleague Charles Franklin compared the trend in the generic ballot to past elections back in August (see especially the powerhouse graphic below), he noted that Democrats have not had a lead that "approached 10%, let alone exceeded it" since 1994.
That lead is important because historically, as Franklin showed in a subsequent post, the better a party does on the generic ballot the higher its share of the national Congressional vote. Unfortunately, however, that question proves to be a very blunt instrument in predicting seat changes, showing a huge historic variation of 45 seats on Franklin's chart. And as he points out, since 1946, the generic vote favored the Democrats just before 7 of the 8 elections that installed Republican majorities in the U.S. House.
There are two reasons for this imprecision. The first is that the generic question omits the names of the candidates, so the answers probably tell us more about general attitudes toward the two parties than about actual vote preference. Surveys I have seen recently that asked both the generic and actual votes produced a mismatch at the individual level of roughly 25%.
The second problem: Forget the survey question, for the moment. The national vote for Congress - the thing we are trying to predict -- is itself an imperfect predictor of seat changes. Franklin has the details (and as usual, the killer chart), but the following table shows some of the noise. Notice, for example, that the Democratic share of the Congressional vote was lower in 1998 than 1996, but they managed to pick up five seats. Similarly, Democrats improved their total vote margins between 2002 and 2004, but still managed to lose a net three seats.
Franklin also provides evidence that number of seats won by each party has "been MUCH less responsive to changes in votes since 1994 than in the previous 46 years . . . I would be very reluctant to assume that the historic relationship between votes and seats is still true."
The weak relationship between the vote and the seat count beaks down because of the "stability in the system," that Mark Mellman talked about. We'll get to that.
Given all this imprecision, why pay attention to the national generic ballot at all? Because comparable national surveys are conducted more often, with larger sample sizes and more rigorous methodologies than many of the surveys we are seeing at the district level. If a last minute change occurs in the national political environment ("the size of the wave") the national surveys will show it first.
[Editor's note to self: This is a blog, not a book. so I'll stop here and pick up with a look at the "enthusiasm gap" in the next post].