A Slate article reports that in surveys the proportion of people who say they voted for Obama over McCain does not match the results of the election.   Of course this panders to my inner economist.  I’m interested in how much of the difference can be attributed to outright lying versus self-deception.  An outright liar knows he is lying while credible self-deception involves some chance you actually believe the story you tell yourself.

It would be cool to have an experiment that distinguished between the two.  Maybe it’s already out there?