Folks from states which favor Obama seem to be significantly more intelligent than those from states which favor McCain.
This is an unprofessionally-produced statistical analysis I have thrown together using data from online sources (current as of September 24). Here's how I did it:
I took the estimated average IQs from a table at http://sq.4mg.com/IQ-States.htm which computed intelligence from SAT scores by state (utilizing a table at http://sq.4mg.com/IQ-SATchart.htm) The averages are well above 100 because they are derived from college-bound SAT scores.
Then, from RealClear Politics http://www.realclearpolitics.com/epolls/maps/obama_vs_mccain/ I created the following variables:
1 - solid McCain 2 - leaning McCain 3 - toss up 4 - leaning Obama 5 - solid ObamaSo, Texas (solidly pro-McCain) is a 1, North Carolina is a 2, Florida (a toss up) is a 3, Oregon is a 4, and Massachusetts (solidly pro-Obama) is a 5.
The actual data table I created is at the end of this article.
A Spearman Rank Correlation produced a probability of 0.005, which is generally considered a highly significant correlation. (It means that the probability that the relationship found between estimated IQ and candidate preference was due to chance was 1 in 200.)
It strongly suggests that states with more intelligent citizenry (by virtue of being better educated) are more likely to favor Obama than McCain.
There's some problematic history to such claims. A few years ago, a series of "studies" purported to show that states with more intelligent inhabitants were much more likely to vote for Kerry than for Bush: for example, http://chrisevans3d.com/files/iq.htm Unfortunately, the estimated IQs were never found to be valid and were largely dismissed as fraudulent.
In 2006, conservative Steve Sailer, founder of the Human Biodiversity Institute and movie critic for The American Conservative, published new figures based upon the work of Michael McDaniel, a professor of business at my old alma mater, Old Dominion University. He computed an insignificant probability (-0.12) that there was a correlation between average state IQ and voting for Bush over Kerry, thus claiming to debunk the notion that intelligence and voting patterns were related. http://www.vdare.com/Sailer/061022_iq.htm
Why is my analysis different from Sailer's? I can't say. I'm not a statistician; I'm a clinical psychologist (very familiar with IQ tests in an assessment context), and while I have some basic knowledge of statistics (and used rather complex canonical models for my thesis and dissertation), I've done no such analyses in many years. I'm certain a competent statistician or research designer would find shortcomings in my design, and challenge my conclusion. I'm certain questions may be raised regarding the validity of candidate preferences from RealClear, the actual relationship between SAT scores and estimates of IQ, and the appropriateness of the Spearman statistic. But, for now, I'm just putting it out there to provoke some discussion, perhaps to compel others to do more serious work addressing the issue.
And asking. . . is anyone really surprised?