Only a tiny minority of American academics are Republicans, a fact that is a continuing source of angst for much of the political right, as well as quite a few centrists. It’s generally assumed that this fact requires some explanation specific to the way in which universities work. The implicit assumption is that the group of those qualified and willing to take up academic jobs is roughly representative of the US population, and therefore contains roughly equal numbers of Democrats and Republicans. To state that submission is to see immediately what’s wrong with it. As a group, academics are obviously not typical of the US population. They have much more education and significantly higher incomes, though not as high as those of highly educated Americans in general. We know that these two characteristics work in opposite directions politically. Other things equal, more income is positively associated with Republican voting while more education is associated with lower support. So, a proper test of the idea that there is something special about academic voting patterns would begin with a multiple regression incorporating income and education as explanatory variables, then see if a dummy variable for academic employment was (statistically and quantitatively) different from zero.

But this is a blog post, so I’m not going to bother with all that hard work. Rather, I’ll point to this New York times article about the voting patterns of doctors. It includes a bivariate regression of voting patterns on income, with specialisations marked as observations It includes a bivariate regression of voting patterns on income for a sample of 30 000 doctors. This graph shows the resulting regression and plots the mean values for different specializations

[click to continue…]