According to Geoff Nunberg at Language Log, all those recent studies showing a ridiculous skew among faculty in humanities toward liberalism and the Democratic and Green parties are distorted. Apparently the polling favored people in departments that are more skewed than others, e.g. English, Women's Studies, History and avoided fields like business, economics, law, physics, or engineering. Well, humanities faculties don't normally include the latter fields, so I wouldn't expect them to be inaccurate in describing humanities faculties by leaving those out, but if they focus on the most skewed humaties departments it's not quite as accurate as it could be.
Nunberg admits that the general conclusions of these studies are likely to be true, that there are more liberals than conservatives in academia. He makes a fair point that it's inconsistent to complain about bias in humanities departments while ignoring a similar skew in economics departments the other way (though I suspect it's not a deliberate ignorance but rather that people complaining about this just haven't thought of that consequence of their view). Nunberg concludes by pointing out that having a view doesn't amount to having a bias anyway. To some degree I agree, but when you've got groupthink going on it's very difficult to understand the opposing side. See Mark Bauerlein's excellent article in The Chronicle of Higher Education for more on that. I have a number of colleagues who don't even come close to understanding why someone might like President Bush.