September 23, 2013

Study: Early voting associated with lower turnout

Reformers hate it when this happens:  The country’s most widely adopted reform designed to make voting easier may lower the chances that an individual voter will go to the polls, according to a new study to be published in an upcoming issue of the American Journal of Political Science.

“The most popular reform—early voting—is actually associated with lower turnout when it is implemented by itself,” according to the University of Wisconsin team of political scientists who studied state voting patterns in the 2004 and 2008 presidential elections.  “This result upends the conventional view that anything that makes voting easier will raise turnout.”

Controlling for other factors that predict an individual’s probability of voting, these researchers found that early voting appears to “lower the likelihood of turnout by three to four percentage points” compared with the probability in 15 states that do not allow early voting or had not implemented other voting reforms.

These early-voting states included some of the largest and most politically important, among them Ohio, Florida, Texas and New Jersey.  (They did not include Oregon and Washington, two vote-by-mail states, among their early-voting states because “they have unusual mail-in-balloting rules.”)

In contrast, these researchers found that Election-Day registration in which citizens may register and vote on Election Day appears to have boosted the probability of voting by about three to four percentage points in states that have adopted this reform.

Their claims are based on a detailed analysis of vote history and demographic information from more than 150,000 individuals interviewed in November, 2004 or November, 2008 as part of the Census Bureau’s Current Population Survey.  The surveys included a question that asked if the individual had voted.

To build their model predicting individual turnout, they examined demographic factors such as level of education, marital status, age and gender. Their model also included a measure of campaign competitiveness in each voter’s state as well.

The researchers also divided states into four groups based on whether the states have adopted any of these reforms: early voting; Election-Day registration; same-day registration, which allows citizens to register and vote on the same day but in advance of Election Day; or none of these reforms. They further identified states that had adopted two or more of these changes, including Wisconsin and Iowa, which have implemented all three.

Using a statistical technique called regression analysis, the researchers estimated the impact of each factor—including voting reforms in force in his or her state—on an individual’s probability of having cast a ballot, all other factors held constant.

They found that only early voting when it is implemented by itself and Election-Day registration appeared to significantly affect turnout chances—early voting in a negative way and Election-Day registration positively.  In 2008, eighteen states allowed early voting but had not implemented other reforms, including Florida, Texas, New Jersey and Indiana.  Four were classified as Election-Day only, including Minnesota and New Hampshire.

Why should turnout go down when people are allowed to vote early—and why do more people cast ballots in states that permit people to register and vote on Election Day?

These researchers say it’s because early voting robs “Election Day of its stimulating effects,” reducing social pressure to vote and gives less reason for campaigns to motivate their supporters and get them to the polls.

Voters are less motivated to cast ballots because early voting has the effect of “dissipating the energy of Election Day over a longer period of time….[S]ocial pressure is less evident, guidance on how or where to vote is less handy, and the prospect of social interactions at the polls is decreased,” they wrote.

Early voting laws also seem to affect the campaign itself by reducing efforts on both sides to mobilize support. For example, they examined patterns of media advertising in early voting and no-reform states. “The volume of ads is lower in states with early voting and the ramp up of ads before Election Day is also less steep in these states,” they found.

They acknowledge the added convenience of early voting “but this effect is more than offset by a reduction in mobilization efforts, resulting in lower net turnout.”

In contrast, Election-Day registration “eliminates the need to register before the campaign reaches maximum intensity and focuses social and political activity on a single day.  Election Day is abuzz with discussion, media coverage, and last-minute contacts from parties and candidates, factors that can exert a mobilizing impact on a wider group of potential voters in (Election-Day registration) states,” they wrote.

Earlier studies on the impact of voting reforms have produced mixed results.  A 2007 study concluded that early voting had no effect on turnout in midterm or presidential elections between 1980 and 2004. Another research team found that early voting produced a short-lived increase in turnout that disappears by the second presidential election after the reform was implemented.  Studies in Washington and Oregon have found that voting-by-mail had a positive impact on participation, though other researchers estimated that turnout would decline in California if that state shifted to mail-in balloting.

Note:  The study was conducted by University of Wisconsin professors Barry C. Burden, David T. Canon, Kenneth R. Mayer and Donald P. Moynihan. The researchers are affiliated with the university’s Election Administration Project, which is supported by funding from the Pew Charitable Trusts, the principal funder of the Pew Research Center.

Category: Social Studies

Topics: Voting Issues, 2008 Election, 2004 Election

  1. Photo of Rich Morin

    is a senior editor focusing on social and demographic trends at Pew Research Center.


  1. Matt Wood3 years ago

    Wait, they used a survey of a same of people and not readily-available actual turnout numbers? Sounds like they were trying to steer their study a certain way…

    1. m3 years ago

      Actual turnout numbers (a) are at the aggregate level and generally don’t tell us which individuals voted, and (b) do not contain any of the demographic data on individuals necessary to develop a model of the factors that influence the decision to vote.

  2. david4 years ago

    I know many people that no longer cast their ballot because they feel neither party offers them anything but same old same old.

  3. fire14 years ago

    I agree with the Daniel Smith’s analysis of the shortcomings of the research. Even more because publishing of dubious research will be used by the traditional vote blockers as reason to eliminate early voting.

    The comparisons also do not seem to have included differential measurement of early weekend versus early weekday voting (low wage workers have difficulty getting time off to vote). How long qualifies as early voting (1 day or two weeks)? What happens in places where the polls are normally over run with backed up wait lines?

    My own polling place is deserted for the entire day (student oriented precinct and they don’t vote in any numbers at all). Yet, I choose to vote absentee or early at the downtown elections office, because it disrupts my day too much to have to go to where I live to vote instead of where I work and recreate. Voting on my own schedule close to my daytime activities are the clearest reasons why I want early voting to continue.

    1. m3 years ago

      The research is hardly “dubious.” It may not be perfect, and it’s findings may not be equally applicable to every kind of early voting law, but the study is well done.

  4. Daniel Smith4 years ago

    Unfortunately, this study by my friends at UW-Madison, which uses CPS data from 2004 and 2008, employs a flawed research design, as it lumps all early voting together, regardless of whether it’s done in person or by mail (absentee). Conceptually and in practice, these are not the same voting processes, they are not used by the same types of voters, and the two types of voting are not targeted by the same mobilization efforts. The authors really need to disaggregate early in-person voting from absentee (mail) voting. On the dependent variable side, the authors also should disaggregate voter turnout among racial and ethnic groups within states, to see if the supposed negative effects of “early voting” on turnout are differential. My research with Michael Herron (Dartmouth) shows quite clearly that early voting in FL has been a major mobilizer of African Americans. I look forward to the next iteration of this important line of research.

    1. m3 years ago

      Of course looking at voting among minorities in 2008 — as is done in the paper you linked to — is problematic in its own right if the goal is to draw generalizable conclusions. President Obama’s historic appearance on the ballot renders 2008 quite different from the typical election year in that respect.