Michael Lewis' essay-review of Louis Menand's Marketplace of Ideas in the February, 2010 Commentary (not available online) points to a 2007 survey of the political opinions of America's university professors:
The percentage of social-science faculty members at elite colleges and universities who voted for George Bush in 2004 was--statistically speaking--zero percent. Likewise among their colleagues in the humanities: zero percent.
Lewis comments: "This is the sort of results that usually sends worried statisticians scurrying back to their data to see what went wrong." The proble, I fear, ain't the quality of the survey.