His conclusion? “While early in-person voting and no-excuse absentee voting in Ohio has reduced waiting times on Election Day, it has had no measurable impact on increasing voter turnout,” he wrote.
For those who spend much time looking at early voting, that's not a big surprise.
After the 2014 midterms, the Census Bureau released a report looking at congressional election turnout since 1978. Included in that report was analysis of the percentage of the vote each year that was cast before Election Day, either by absentee ballot or at early-polling stations. We can include data from Michael McDonald's U.S. Elections Project for 2016 to demonstrate the steady upward tick of early voting as a percentage of votes cast.
But as that increase has happened, turnout hasn't changed much. (Here again we use McDonald's numbers.) The number of votes cast has generally increased, of course, which is mostly a function of the increasing national population. The percentage of those who can vote and do, though, doesn't seem to be affected by early voting — at least nationally.
Again, this isn't entirely a surprise. Even before this election cycle, analysis of early voting suggested limited effects on turnout. In one 2013 paper, researchers reported an inverse relationship between early voting and turnout — meaning that as early voting expanded, turnout dropped. What actually increases turnout is same-day voter registration, they found, making it easier for people to register to vote and cast a ballot at the same time. (Between 2012 and 2016, Ohio eliminated its “golden week” of early voting, a period during which voters could both register and cast an early ballot.)
The value of early voting to candidates doesn't lie solely in increasing turnout, of course. Instead, it's valuable because it increases the amount of time that campaigns have to get their base of voters to the polls. For Democrats, that can be important: Younger voters and people of color tend to turn out less regularly. Such increases are at the margins, a few hundred or few thousands in a handful of races, the sort of blip that might not show up in national numbers. But it can be important in a close race with a campaign that's running a strong get-out-the-vote effort.
Which, before Election Day, it seemed as if Hillary Clinton's campaign was doing. The results in the Midwest (and my observations in Pennsylvania) suggest otherwise. Dawson notes that there was a shift in Ohio: "2016 turnout in counties won by the Democrat decreased by 1.7% over 2012 and increased by 0.5% in counties won by the Republican.”
Early voting doesn't appear to help turnout by itself, but it can be helpful to campaigns. If the campaigns take advantage.
Update: McDonald emailed to note that he thinks it's premature to fully dismiss a link between early voting and turnout. I asked him to elaborate on his thoughts and he did so.
"My take is that there is a positive relationship between early voting and turnout. The positive effects of early voting on turnout are not large, and the largest positive turnout effects are found in all-mail ballot elections or some form of permanent mail ballot status. These effects are most pronounced in non-presidential elections."But simply stating early voting doesn’t increase turnout misses important nuances in the ways by which elections are administered."For example, during the primary Maricopa County, Arizona encountered significant problems on Election Day when election officials miscalculated by thinking that more early voters meant fewer Election Day voters. The county scaled back Election Day polling locations and the result was long lines on Election Day, which likely reduced overall turnout. This situation was not caused by early voting, rather it was caused by poor decision making regarding Election Day voting."In another example, North Carolina experienced lower early voting participation in predominantly Democratic and African-American counties targeted by Republican electoral boards who reduced the number of early voting polling locations. These shenanigans were the culmination of changes to election laws that a federal judge found to be intentionally discriminatory and in violation of the Voting Rights Act. Selectively targeting geographically clustered groups with different election experiences can increase and decrease turnout among different groups, while overall turnout can remain the same."
I would certainly agree with the latter point without hesitation, as noted above.