The Median Voter and the Mixed Voter

Our minds work by making models of the world and using them to predict how things will happen. These models are powerful because they’re so deeply-ingrained we don’t even realize we’re using them. They just seem like “common sense.” In politics, there are two major models for how voters think, which I’ll call the median voter model and the mixed voter model.

The median voter model says that politics lives on a line from left to right. Voters are scattered across this line and vote for the politician that’s closest to them on it. Politicians get elected by “positioning” themselves closest to the most voters, which usually means in the “center” of the line.

There are some complications, though. Because primary voters tend to be “more extreme” (i.e. Democratic primary voters are all on the left, Republicans on the right), politicians take a more extreme tack during the primaries, before heading back to the center for the general. And because they don’t want to seem like flip-floppers, they’re somewhat constrained by the primary positions they take.

But, in general, this model is pretty widely-accepted in politics. So widely that it’s not even thought of as a model — it’s implicitly assumed by all the things political commentators say. Comments like “He’s moving to the center to pick up votes,” “[X] couldn’t get elected in that district, so how is someone more extreme going to make it?” only make sense because we all have this model in our heads.

But, while I haven’t studied the question in detail, there doesn’t seem to be much evidence for this model. Even intuitively, it doesn’t make sense: does the average person really develop a location on a one-dimensional issue spectrum and then figure out where various politicians stand on that same spectrum? The notion seems almost ridiculous.

UPDATE: Andrew Gelman has studied the question in detail, and concludes that the median voter theorem doesn’t seem to be true: “My research with Jonathan Katz suggests that being a moderate is worth about 2% of the vote in a congressional election: it ain’t nuthin’, but it certainly is not a paramount concern for most representatives. … Incumbent congressmembers almost always win reelection. And, when they don’t, they’re often losing as part of a national swing (as in the 1994 Republican sweep or the 2006/2008 Democratic shift). And when an incumbent does lose unexpectedly, it can be for something unrelated to their votes (remember the “check kiting scandal” of 1992?).”

The mixed voter model, promoted by George Lakoff and most prominently adopted by Howard Dean, says that voters aren’t rational, coherent evaluators but a bundle of feelings, prejudices, and contradictions. Politicians get elected by playing on the feelings voters already have that would encourage a voter’s’ support. Thus, instead of moving to the center to get more votes, Lakoff argued politicians should actually become more extreme. Conservatives won votes by appealing to people’s sense of order; liberals would have to respond by appealing to their sense of empathy. Moving to the center by promoting a compromise position that prevented rhetorical appeals of either type, was doomed. This model was used to explain why radical conservative politicians kept winning elections against moderate and centrist Democrats, when there was no evidence of a conservative electorate.

This model makes much more sense to me, although again I haven’t seen too much specific evidence for it. But it’s still pretty rare and seems deeply-counterintuitive to most practitioners of politics. But whether it’s right on the details or not, it’s clear that unless we evaluate and question these models and think about them critically instead of just assuming one is true, it’ll be hard to make much political progress.

You should follow me on twitter here.

July 28, 2009