Charles Stewart III has posted a very important paper (forthcoming in the Journal of Law and Politics).  I heard him present this paper a few weeks ago in Charlottesville, and the data are fascinating and key to seeing how we might “fix that:”

This article provides an empirical grounding into the patterns of long lines, focusing on the 2012 presidential election and utilizing a unique public opinion survey. I show that two-thirds of voters in 2012 waited less than 10 minutes to vote, and that only 3% of voters waited longer than an hour. I show that there was considerable variation in line length, as a function of geography and race. Consistent with news reports, Florida’s voters waited the longest to vote in 2012, nearly 40 minutes on average, while Vermont’s voters waited less than two minutes. Urban voters waited longer than rural voters, early voters waited longer than Election Day voters, and African American and Hispanic voters waited longer than whites.

I also show that lines were nothing new in 2012. The states whose residents waited the longest to vote in 2012 also waited the longest in 2008. This fact cautions against blaming long wait times in the most recent presidential election primarily on factors specific to 2012.