Not Using Wisconsin to Test Anything.

At the risk of undermining the united front we have here at Mischiefs of Faction, I want to cast some doubt on Seth's thoughtful piece comparing the two back-to-back Wisconsin gubernatorial elections.

The strategy Seth employed in that post makes some sense. It's what we might call a "most similar case design," in which we take two observations that differ on very little except for one important thing, and then we consider the apparent consequences of that one thing. Since the cases are so similar, differences in outcomes might be due to the very small differences we do observe. In this case, spending differed, first with a 2:1 Walker advantage, second a 7:1 advantage. But the outcome did not much differ, suggesting that the spending difference didn't have much effect. It's not a perfect design, but lacking much else, it's a good thing to look at.

The problem with a most similar case design is that even very similar cases can differ in important ways. In this case, the recall election was very different, even if it featured the same candidates in the same state. It was (a) a recall, (b) spurred by very intense and high profile conflict that (c) probably made a lot of people aware of things they were not aware of in the first election. The very nature of a recall implies that at least someone thought things were different enough that another election was in order. So it's not so similar after all.

For instance, I think it's reasonable to infer that the recall petition process itself mobilized and energized a lot of Wisconsin voters who may have been less excited about Barrett the first time, if only because it was a more routine election. If so, it might also be reasonable to infer that the money Walker spent was necessary to create a similar mobilization and energizing of those who were probably less excited about Walker the first time, again because it was routine. In general, I tend to think that campaigns matter mostly insofar as they bring voters around to where they would have been anyway. But that has to happen. The well-known relationships between party ID and the vote, and between economic conditions and the vote, might not exist without any campaigns. In the short-time-frame recall, maybe all that Walker spending was necessary to do that.

This is not to say that I think Walker bought the election. I think spending is a lot less important than most people do. But not because of this example.