EAC Survey Sheds Light on Election Administration
U.S. elections are administered at the local level, not by the federal government. Election results are reported by local election jurisdictions to their state counterparts, who certify the winners for their states. But there is no systematic, national collection of information that underpins the performance of our nation’s election administration, from how voters are registered to how they cast their ballots.
Until now, that is. Operating under the auspices of the Help America Vote Act, the U.S. Election Assistance Commission conducted the Election Day Survey, the first attempt by the federal government to systematically collect voting and election administration data from local election jurisdictions across the country. We were contracted by the EAC to analyze the Election Day Survey. Some of our findings and recommendations are presented here.
While the Election Day Survey holds promise to inform us about voters’ election experiences and the administration of elections, the survey also reveals continuing challenges.
We found that in the fall 2004 election, the “drop-off” — that is, the difference between the total ballots cast and the votes for a particular office — was the smallest in a post-World War II presidential election. While some people consciously abstain from a race on the ballot when they vote, the low residual vote suggests that voters were more conscientious about correctly casting their ballot and were aided by new electronic technology that warns a voter of potential errors with their ballot.
We found that of the at least 1.9 million provisional ballots cast, 1.2 million were counted. Provisional ballots provided a vehicle for many people to cast a ballot who might otherwise have been disfranchised.
We found benefits to centralized administration of registration rolls. HAVA mandates the adoption of statewide voter registration databases, rather than the administration of voter registration solely on the local level. In those states that have implemented statewide voter registration databases, provisional balloting rates were lower and more absentee ballots sent to voters were returned. These benefits not only help voters, they also reduce the cost of administering elections.
While these are encouraging signs of improved election administration, we also found continuing challenges to election administration in the United States.
The higher turnout rate in the 2004 election is encouraging, but we continue to observe — as has been observed for many elections — lower levels of turnout among persons living in areas that are poor and have lower rates of educational attainment.
In predominantly Hispanic jurisdictions, we found higher rates of inactive voters on the registration rolls, higher rates for provisional balloting and higher rates of unreturned absentee ballots. Language barriers may play a role in these findings, and we urge increased attention to the language provisions of HAVA and the Voting Rights Act.
The survey revealed that states and local jurisdictions use various definitions for such seemingly fundamental concepts as what constitutes a poll worker or a polling place, the difference between an absentee ballot and an early ballot and even how many registered voters live in a jurisdiction. Many local jurisdictions did not respond to all survey items, often because they do not routinely track the requested information.
In the absence of common definitions and full survey responses, we cannot compare local jurisdictions or states, or draw completely valid conclusions about much of the information found in the survey. For these reasons, “state rankings” that might reveal best practices or identify worst offenders are largely meaningless.
In the coming months, many organizations will analyze the Election Day Survey data and make claims based on it. We want to strongly encourage these organizations to be careful in their analysis. Things that may appear as evidence of fraud, such as a jurisdiction having more absentee ballots counted than returned, often stem from how local election administrators interpreted the questions asked of them — for instance, in this case, not counting an absentee ballot delivered to a polling place as a returned ballot. And this example just scratches the surface. It is not an understatement to say that we have written a novel-length report full of such idiosyncrasies.
Among our recommendations are to improve the questionnaire design and to develop a survey instrument that will encourage jurisdictions to respond. Resolving data collection and definition issues now is important as the EAC contemplates new voluntary certification standards for voting equipment and as states upgrade and centralize their management of voter registration databases. Developing technology that captures and records vital election statistics will greatly help future data collection efforts, which should in turn improve election administration and, ultimately, democracy in America.
Michael McDonald is a visiting fellow at the Brookings Institution and an assistant professor at George Mason University. Kimball Brace is the president of Election Data Services, a company specializing in redistricting, election administration, and analysis of census and political data.