I think the change in the collective evaluation is due, in part, to the following: none of the vulnerable Republican incumbents on the Senate side seemed to have gained any appreciable traction thanks to Bush's rising job numbers and the falling price of gasoline. Perhaps pundits have found themselves surprised by this - and so are beginning to suspect that the GOP is in worse shape than they initially believed.

For my part, I am not surprised by the fact that Senate Republicans have enjoyed little-to-no bump from Bush and gas. And I do not think the final estimate needs to be rethought. (As a Wahoo during his tenure as governor and during his 2000 candidacy, I know that you underestimate George Allen at your peril; Tennessee flirted with Bob Clement in 2002 before giving Lamar Alexander 54% of the vote.) However, pundits do need to reevaluate what they take to be causal factors of vote choice. Regular readers of mine know that I have been on this topic for months - to little avail. Since the Spring, I have been arguing that the punditocracy has consistently misunderstood the relationship between national conditions and congressional vote choice. The economy and presidential job approval have a very complicated relationship with non-presidential elections.

We should not be surprised that neither of these factors are having an effect in the vulnerable GOP-held Senate seats. And if pundits are indeed puzzling over this lack of an effect, and they are thinking that maybe they need to change six months worth of confidence that the GOP would hold the Senate, maybe they should instead rethink how they understand congressional elections to actually work. For what we are seeing right now is consistent with what political scientists have found to be true: presidential job approval and economic evaluations have a complicated, and largely indirect, relationship with individual vote choices.

First off, the economy. What gives? Gas prices are falling. Consumers are more confident. The Dow is once again peaking. Why is this not helping Jim Talent in Missouri? It is not helping him because the theories of Anthony Downs and the rational choice theorists of the 50s and 60s - that the individual voter is "rational" and makes a selection based upon an assessment of his self-interest - have never found much empirical validation at the micro level. There has never been strong, consistent evidence to indicate that economic rationality exists in this way - though economic conditions do show a strong, statistically significant impact on net seat swings. Nevertheless, people do not seem to vote their pocketbook - when we control for other factors, like partisanship and demography, we see that pocketbook voting has little-to-no independent effect on individual vote choice. At best, we can say that long-term economic evaluations have an effect on partisanship, which in turn has a strong effect on vote choice. But that is about it. So - we should not be surprised that falling gas prices in Missouri have not helped Jim Talent. There is little in the scholarly literature to induce one to believe that a $0.30 drop in the price of regular unleaded would help him.

Second, presidential job approval. Unlike economic evaluations, this has been found to have some relationship to individual vote choice. However, it has by no means been a consistent effect, and it is not as strong as pundits take it to be. For the elections of the 1970s, for instance, scholars failed to find a strong relationship between presidential job approval and individual vote choice - though there was indeed a relationship between the two discovered in the 80s and as late as 2002. Nevertheless, the strongest effect that presidential job approval has is not direct, but rather indirect. Presidential job approval is what establishes, more than anything, the partisan political climate in this nation. When a President has high marks, his party seems to be strong. When he has low marks, his party seems weak. This affects the strategic considerations of ambitious partisans on the other side of the aisle. They will run when their opponent's party is disadvantaged. So - in the case of Talent, it is unsurprising that he is not getting much, if any, boost from Bush's higher numbers. Bush's numbers already had their largest effect when Claire McCaskill, without doubt a top-tier candidate, decided that the mood in her state was sufficiently anti-Republican to give her a good shot at victory.

All in all, then, I personally am not surprised to see Republican Senate candidates remaining relatively flat. And I would hope that the pundits who for months have been declaring this election to be a referendum on national conditions would take this opportunity to reevaluate the size and scope of the effect that national conditions have on congressional elections. It is more complicated than they have implied, and this recent, "peculiar" development demonstrates that point.

I am not saying that I can fully delineate how these variables factor into vote choices. I know that I cannot. I do not think there is any scholarly consensus on this matter, either. I doubt that even the best of the best in the academy could tell you exactly how the President and the economy factor into congressional elections. What I can say is that the theory upon which so many pundits have been relying, the idea of the electorate being magically transformed into a "nationalized" one that thinks about their pocketbook and Bush rather than the individual candidates in the race, is incorrect in important respects. They need to reevaluate. Otherwise, this will not be the first time between now and November that a result will surprise them so much.