BY JIM KANE
In recent months, some controversy has risen concerning new laws and policies that have changed the rules of when and how voters can participate in elections.
Most notably, requirements for photo identification and the reduction in the number of in-person early voting days have caused some critics to raise the specter of voter suppression by Republican legislators.
The prevailing wisdom suggests that higher turnout benefits Democratic candidates since voters least likely to vote come from the lower socio-economic levels (blue color, the poor and recent immigrants). If that’s true, the changes could harm Democrats in a close election.
But the real question, however, should be whether these voting reforms actually increase or reduce turnout to begin with?
Or perhaps, they have no effect on turnout at all.
For example, if in-person early voting was shown to reduce turnout, then reducing the number of available days could possibly help Democratic candidates.
I know what most of you are thinking at this point. Kane has had one too many glasses of wine at lunch today.
As a confirmed and unrepentant skeptic, I decided to see what academic studies have to say about the issue.
I found two recent articles that were right on point.
Both studies use the same data from the Census Bureau’s Current Population Surveys (CPS) containing some 60,000 to 90,000 interviews in each of three presidential elections (2000, 2004 and 2008). But each study uses a different methodology to come to the same conclusions.
Reforms Don’t Help Turnout
The first study was published in the academic journal State Politics & Policy Quarterly (March 2011) written by Roger Loracca and John Klemanski of Oakland University (Rochester, MI). Using a cost benefit model that gauges the burdens that state election and voting laws place on voters, these researchers measured the effects of early voting laws which they define as Permanent No-Excuse Absentee Voting, Nonpermanent No-Excuse Voting, and Early In-Person Voting. In addition, they also included states that allowed Election Day Registration or required Voter Identification. With these independent and key controlling variables (age, education, gender, race, competition, etc.), they measured the calculated the effect on whether a person voted or not (the dependent variable).
Since this is a probability model (Logistic Regression), the model’s independent variables measure the probability of an individual voting in the election cycle, controlling for other variables. As they expected, Permanent No-Excuse Absentee Voting, Nonpermanent No-Excuse Voting and Election Day Voting all increased the probability of voting (statistically significant). As for Voter Identification, however, the results were mixed at best.
Among some groups, the Voter ID laws actually increased the probability of voting. But in most cases it had no impact at all.
The most surprising result, however, was the effect of Early In-Person Voting on turnout. Not only did this reform not increase the probability of voting but also it significantly reduced the probability of voting across all groups.
Now before we discuss why, let’s look at a different approach to the same question.
In another study published on the Social Science Research Network and sponsored by the Pew Charitable Trust entitled “Election Laws, Mobilization, and Turnout: The Unanticipated Consequences of Election Reform” (2012), four University of Wisconsin-Madison political scientists looked at the same issues of early voting reforms and their effect on turnout.
These researchers used two different measurement techniques to test the effect of early voting. The first measures the aggregate changes in turnout at the county level using a regression model (OLS). This model found (controlling for other key variables) that early voting reforms, absentee and in-person voting, reduced turnout. In contrast, same day registration and Election Day registration increased turnout. This model, however, does not differentiate between absentee voting and early in-person voting.
In a separate individual-level probability model, the results were much the same with early voting reforms (early in-person voting separate from mail absentee voting) reducing the likelihood of voting while same day and Election Day registration increasing turnout.
Both models showed that voter ID laws had no effect on turnout at all.
As the authors write, these findings are counterintuitive.
These researchers suggest that early voting reforms, by spreading the election over several days or weeks, reduces the “stimulation effect” of election day, and, consequently, both nonvoters and marginal voters don’t respond by voting.
Secondly, early voters are likely voters.
These reforms make voting more convenient for them, but they would have voted anyway.
Finally, with 30% of the electorate voting early, the authors present evidence that presidential campaign advertising decreases as we move closer to Election Day thus further lessening interest by marginal voters.
In the Bahamas, Election Day is as much a social event as a political one. I know because I travel there frequently. Businesses close down and people party all day. There is no early voting; it’s all done on one day.
And what is the turnout?
Bahamian elections average over a 90% turnout rate among registered voters. In 1972, turnout was over a 100%, which led to reforms to prevent people from voting more than once. What a good problem to have.
(One of South Florida’s premier pollsters and political strategists, James G. Kane got his start in campaigns while still in a Fort Lauderdale elementary school. He is a frequent commentator in the national media and currently teaches graduate seminars at the University of Florida in survey research.)