Left-right ideology is by far the strongest predictor of party identification. Education, in contrast, has very little connection to partisanship.* However, when you look at opinions about specific issues, ideology and education are both extremely important. For example, conservatives and the less educated are both sharply more hostile to immigration than liberals and the well-educated. (See here for some simple regressions).
If you think about parties as coalitions of people with similar policy preferences, this is pretty weird. Based purely on the data, we could just as easily have a "smart party" and a "stupid party" as a "liberal party" and a "conservative party." But we don't.
OK, so maybe, per Robin Hanson, partisanship isn't really about policy. But then what is it about? The most natural alternative to me is that parties are coalitions of people who feel comfortable with one another. Sure, educated liberals and educated conservatives have a lot of policy views in common. But when they meet, there's a lot of social distance; people don't feel like they can "be themselves."
That's an interesting story. But when I was watching the movie Rachel Getting Married, I started thinking: Which wedding would be more awkward? A wedding where the groom's and bride's families differ sharply in left-right ideology? Or a wedding where the groom's and bride's families differ sharply in education? That seems like a pretty good measure of social distance. And while I'm not sure, it seems like the mixed-education wedding would be more awkward than the mixed-ideology wedding.
What do you think?
* If you include both education and income in a regression, education seems to make people slightly more Democratic, and income seems to make people slightly more Republican. But the key word is slightly.