The following, apparently based on YouGov/Economist poll data, that had I found on Matt Karp’s Twitter feed, is a remarkable graph.
The implication of spatial politics is that, the electoral game is necessarily zero sum in nature: you gain from one side, but you lose from the other, and you gain the most from those whom you are closest to, while losing the most from those whom you are farthest from. In the usual spatial models, the very idea of variance is rarely acknowledged, except as a nuisance.
Seen through the lens of the spatial models, Clinton appears to be the most “liberal” politician of the three: she is supported by strong Democrats, less so by weaker Democrats, and even less by leaners. Independents don’t like her much, but the reaction that she engenders among the Republicans is confusing. Maybe that just means that all Republicans are the same? (Certainly, this is claimed by a lot of spatial modeling types who study polarization–all Republicans bunch up with similar DW-Nominate scores, and all that).
Things get stranger with Barack Obama–beloved by both strong Democrats and leaners, but less so by weak Democrats. Perhaps it doesn’t really mean that much–maybe all the Democrats are the same, too, not unlike the Republicans.
Then there’s Bernie Sanders: who is actually more popular among the leaners and not especially disliked by most Republicans. And this is the guy who is supposed to be the flaming extreme liberal, while Obama and Clinton are, respectively, more centrist?
The problem, I think, is that the idea of spatial models is fundamentally wrong even if convenient and simple. While the idea that people have political viewpoints that correspond to a number line is neat, it is a terrible way to conceptualize how real people see politics. For many, geometric ideology is not all that relevant–it is obscured by huge error terms, which are not errors, as it were, but substantial uncertainty. With large enough uncertainty, politics ain’t so spatial any more, and that ain’t a bad thing.
The trouble with Clinton, indicated above, isn’t so much that she is more or less liberal. She is what the market says she is: i.e. people who are responding to the surveys think she is “liberal,” meaning that she should be supported by people who call themselves “liberals” and hated by those who call themselves “conservatives.” Without knowing what “liberal” or “conservative” really means, we just buy into the market’s designation and stick that on the politicians. But the dirty secret, of course, is that, even if markets might be right on average, they are not only wrong, they are also often stupid. Conventional wisdom often beats out real insights just because people believe them. This is where things become problematic.