More on turnout

One problem with the turnout graph I posted earlier is that it uses the 2004 turnout rate as the baseline. Of course, some states had atypically high turnout that year because of hotly contested gubernatorial and senate races. So I've created the scatterplot below, which measures each state's turnout over the average turnout in the past three presidential elections.Once again, the biggest turnout spikes occurred in southern states. The trendline suggests that the biggest turnout increases came, unsurprisingly, in the battleground states. There's somewhat less of a turnout spike in the more liberal states and much less of a turnout spike in the more conservative states.

A regression analysis, incidentally, shows that the turnout increase correlates positively with the percent of the state that is African American and negatively with McCain's share of the vote. Both these variables are statistically significant at the .01 level.