In an important post, Hilzoy notes that an initial exit poll substantially overestimated African-American support for Prop 8. And once the outlier poll is eliminated, higher levels of African-American support are explained almost entirely by religion. It should be noted that something similar happened in 2004, when some answers to vague exit poll questions led a large number of people to assert that anti-same-sex marriage initiatives were a decisive factor in explaining Bush's victory. Such assertions turned out not to hold up very well under more careful scrutiny. Exit polls are useful in many ways, but it's important to be wary of pundits drawing broad conclusions from them, especially when they happen to dovetail with convenient pre-existing narratives. --Scott Lemieux
