There is much interest these days in what we call “convenience voting” methods — vote-by-mail, early voting, and other approaches to absentee voting. Of these various forms of “convenience voting”, the ones that have seen the most attention from social scientists are vote-by-mail techniques, as a number of states (especially California, Oregon and Washington) have had a relatively lengthy experience with voting-by-mail in the past few decades.
Of these procedures, the one that social scientists have studied the least is early voting. The conceptual distinction here between early and absentee voting is that the former is done in person, typically in a location that is under the direct administration of local election officials. “Absentee” voting (and generally voting-by-mail) is done by a voter in some location outside the control of local election officials, and where the voter’s physical presence is not required to cast the ballot.
To date, there have been few published academic studies of early voting, primarily because few jurisdictions have used early voting until very recently. The academic expert in the area of early voting is Robert Stein, of Rice University, who has published two seminal studies on early voting in 1997 and 1998; Stein’s testimony before the Carter-Baker commission on June 30, 2005 discussed some new research that he is doing on the use of early and election-day voting centers in Larimer County, Colorado (he has a working paper on this research that we received recently!).
Paul Gronke, whose blog “Early Vote” we link to, has been undertaking some very important new research on early and absentee voting, and is presenting some of this new and innovative work at the APSA conference next week. In a paper co-authored with Benjamin Bishin, Daniel Stevens, and Eva Galanes-Rosenbaum, titled “Early Voting in Florida, 2004”, they present some analysis of an exit poll of voters in the 2004 presidential election in Miami-Dade County, Florida.
They summarize their results in the “Discussion” section of the paper:
“Older voters, Cuban Americans, and women voted earlier, ceteris paribus. Political information showed a curvilinear relationship with early voting, with respondents in the middle range of information showing the highest rates of early voting. Finally, party affiliation and party contacts did not operate as we expected, although reported contact by non-partisan groups did increase early voting rates.
“Finally and perhaps most importantly, we found that voters who trusted local government, and those who had confidence in the integrity of their ballot, were more likely to avail themselves of the opportunity to vote early.”
These are intriguing results. One of the important debates about convenience voting reforms is the extent to which they increase turnout, or whether they simply make it easier for otherwise high propensity voters to participate. The Gronke et al. data can’t solve this debate, but there are hints in this data that indicate the story is more complicated than either side in that debate might realize. On one hand, that older voters are more likely to cast early ballots seems to indicate that the early voting option in Miami-Dade County in this election might have simply made it easier for this high propensity group to vote. But also note that in their analysis income and education had no statistically significant effect on the likelihood of early voting, while other variables that might be seen as predictors of the propensity to vote (news exposure and strength of partisanship) had very weak and statistically insignificant effects in their analysis. On the other hand, we do not see strong evidence supporting the idea that early voting stimulates turnout by low propensity voters, as their results do not show that racial minorities (here Blacks and Haitians) were more likely to vote early, nor were younger voters more likely to vote early.
Another important debate about early voting is whether or not early voters are voting on the basis of different information than election day voters. Some have raised serious questions about this concern, noting that when late-breaking news or campaign events arise, those who have already voted may miss out on that information. Gronke et al. do not address this question in their paper, but it appears that their data might allow some testing of this question. They do have in their survey instrument measures of issue opinions, political knowledge and news exposure. Their analysis does show that the latter two variables do not significantly predict whether one is an early voter or not. However, I wonder if they could look whether early voters and election day voters have different issue preferences, and whether “early” early voters have different information, issue opinions, or knowledge than “late” early voters. Their survey responses span the period October 22, 2004 through November 2, 2004, so they might be able to exploit the temporal nature of their data in future work.
Last, the only major caveat regarding this paper is that despite the title (“Early Voting in Florida, 2004”) the exit poll survey used here are really only for voters in Miami-Dade County. That clearly limits our abillity to generalize the results of this analysis to even other counties in Florida, not to mention the rest of the United States. (By way of disclosure, I’ve produced an analysis of absentee voting in Los Angeles County with Thad Hall and Betsy Sinclair [“Whose Absentee Votes Are Counted: The Variety and Use of Absentee Ballots in California”], which has “California” in the title, and so I am well aware of the problems of generalizing from one county to a state).
In conclusion, this is very good work. We need to know much more about early voting, ranging from studies like Gronke et al.’s to comparative studies across jurisdictions. There is much work to be done!