At least in this case:
From “Education, politics and opinions about climate change evidence for interaction effects” (PDF):
Abstract U.S. public opinion regarding climate change has become increasingly polarized in recent years, as partisan think tanks and others worked to recast an originally scientific topic into a political wedge issue. Nominally “scientific” arguments against taking anthropogenic climate change seriously have been publicized to reach informed but ideologically receptive audiences. Reflecting the success of such arguments, polls have noted that concern about climate change increased with edu- cation among Democrats, but decreased with education among Republicans. These observations lead to the hypothesis that there exist interaction (non-additive) effects between education or knowledge and political orientation, net of other background factors, in predicting public concern about climate change. Two regional telephone surveys, conducted in New Hampshire (n=541) and Michigan (n=1,008) in 2008, included identical climate-change questions that provide opportunities to test this hypothesis. Multivariate analysis of both surveys finds significant interactions. These empirical results fit with theoretical interpretations and several other recent studies. They suggest that the classically identified social bases of concern about the environment in general, and climate in particular, have shifted in recent years. Narrowcast media, including the many Web sites devoted to discrediting climate- change concerns, provide ideal conduits for channeling contrarian arguments to an audience predisposed to believe and electronically spread them further. Active- response Web sites by climate scientists could prove critical to counterbalancing contrarian arguments.
There seems to be something here that explains a lot about beliefs other than climate change: evolution and political subjects like health care and economic policy – pretty much any subject where an intellectually indefensible position is in fact defended by ideologically-driven snake oil outfits whose product is scientific-sounding doubt of some mainstream scientific consensus. See “The Merchants of Doubt”.
UPDATE: Here is some follow-up material on this issue, some of which shows that on the subject of evolution, there is still an enormous conservative/ liberal split (with reality favoring the liberals again), but education doesn’t make you dumber.
The Finch and Pea 
As we scientists like to point out to our fellow PhDs in the English department, not all post-grad education is the same. It would be useful to know how areas of educational focus are distributed at the higher end between party preference, how opinions on these topics vary in those fields, and whether there are multiple selective effects influencing this result (e.g., student imitating the teacher).
Also, given a total sample size ~1500, I wonder how large the “post-grad” bin actually is and the size of the error estimates.
The educational breakdown is maybe a little too fine-grained – I doubt that they can say with any confidence that there is an important difference between college and post-grad.
But you can break it down into simply college grads vs. non-college grads (as other studies have done), and the difference is pretty clear.