Some thought on early voting trends

Early voting has been widely covered in the news recently, and of course we’ve blogged a fair bit about some of the trends. Paul’s the expert on early voting, and he’s been watching things closely.

Other than the big trend that we’ve all been following — what looks like record early voting turnout in many places — there has been more recent reporting and discussion about potential partisan skews in early voting patterns this year. In particular, there have been some media reports that Democrats might be gaining from early voting this year; for example, USA Today reported “Early voting a boost for Democrats.”

In fact, Paul was quoted in that story:

Democrats are voting early in greater numbers than their Republican counterparts in several closely contested states, reversing a pattern that favored the GOP in past elections.

The trend is evident in Ohio, North Carolina and Iowa, Nevada and New Mexico, state and county figures show. In Georgia, blacks are voting in greater numbers than they did in 2004.

The early voting trend is about even in Colorado. Republicans claim the edge among absentee voters in Florida, but Democrats are voting in far greater numbers at early voting polling places where voters lined up this week.

“This is like a mirror image of what we’ve seen in the past,” says Paul Gronke of the Early Voting Information Center at Reed College. “This cannot be good news for John McCain. It’s the 100-yard dash, and (Barack) Obama is already 20 yards ahead.”

This will be an important research question after the election. There is a “conventional wisdom” story that early voting and voting-by-mail has in the past favored Republicans; it will be very important to see what evidence there is that supports both that conventional wisdom, whether it has changed in 2008, and most importantly why it might have changed.

Another thought about the reports of record-breaking early voting trends, and the potential partisan skews: what effect might all of this have on election-night vote projections, and especially reporting based on exit polls? In recent years, the exit pollsters have tried to gauge the effects of early and absentee voting in elections by using telephone samples of early and absentee voters. It might be worth keeping in mind that in states with record levels of early voting — especially where that might have not been well-anticipated in advance of the election — the exit poll estimates might have a bit more uncertainty in them than the pollsters think. This will also be something to look at after the election; to what extent do widespread and unexpected levels of early and abstentee voting hamper election-night forecasting, and what can be done to improve those estimates?