Since Nov. 8, 2016, Americans have been debating the integrity of President Trump’s election. But whether or not Russia illicitly influenced the voting, a newly released report from the Elections Performance Index shows that actual election administration improved from 2012 to 2016.
It usually doesn’t make the news when bureaucratic structures — like the ones that enable efficient access to the polls — function as they should. But this news is important. The near-constitutional crisis of 2000 revolved around routine issues of election administration — which ballots to include in the count and how to interpret ambiguous voter marks.
Let’s examine what we measured — and how we did it.
What is the EPI?
Election administration operates on what I call the Roseanne Roseannadanna principle: It’s always something. If it’s not a hanging chad, it’s long lines; if it’s not long lines, it’s thousands of people left off the voter lists. The trick in measuring election performance overall, therefore, is to spread the measurement strategy across the entire election process, rather than focusing on any one indicator.
That’s what the EPI does, measuring election performance over 17 indicators, such as absentee ballot administration, the use of provisional ballots and the presence of post-election audits. It measures how well states address the needs of voters with disabilities and overseas and military voters, as well as more well-known factors such as the turnout rate and the average wait time to vote.
The indicators themselves emphasize voter convenience; a few touch on security. Because these were set well before the 2016 election, few address cybersecurity concerns. But the EPI does includes relevant items, such as whether a state mandates post-election audits. Each indicator is placed on a 0-100 percent scale, indicating how close the state has come to achieving its best. When the results of the indicators are averaged together for a state, the EPI can document whether it has improved overall and how it compares to others.
The index was first released by the Pew Charitable Trusts in 2013. In 2017, Pew passed the EPI on to the MIT Election Data and Science Lab, which has just released an update based on 2016 data. The EPI project continues to be an active research partnership involving both election officials and academics.
Here’s what improved between 2012 and 2016
The map below shows how all the states fared in 2016.
Forty-four states and the District of Columbia had higher overall scores in 2016 than in 2012. Six states declined, but by tiny amounts, often because of random variability. You can see the measures in the map below.
Looking across all the indicators, three areas of election performance saw notable improvements:
- Shorter wait times to vote. In 2012, long lines at the polls were the big election administration issue. According to responses to the Survey of the Performance of American Elections, the average wait time in 2012 for all in-person voters was 14 minutes; that went up to 45 minutes for Florida voters. In 2016, the national average fell to 11 minutes; Florida’s average collapsed to 5.5 minutes.
- Increased use of post-election audits. Auditing election outcomes is important to show that computer tabulation systems functioned properly. In 2008, 23 states required post-election audits. By 2016, 35 did. Colorado has even implemented statewide risk-limiting audits, which give high assurance that an election was counted correctly by efficient ballot sampling.
- Growing online resources. A decade ago, voters would have had a hard time finding basic election-related information online. That has changed. Two-thirds of states allowed voters to register online in 2016. In 2008, few states provided any personalized voter information online — such as where to vote and how to verify your registration. By 2016, every state provided some information online.
Of course, the EPI could be improved — and will be
Any effort to assess states’ performance in a complicated policy area using a common scale can be criticized. The EPI is no exception. A common critique is that the EPI uses a very narrow set of criteria to focus on election administration, omitting other factors affecting election quality, such as the robustness of political debate, the role of campaign money or the bias of legislative districts.
All that may be true — but the importance of high-quality election management hit the spotlight again this past week, with the razor-thin margins in Kansas’s Republican gubernatorial primary and the special election in Ohio’s 12th Congressional District. These elections will hinge on how these states handle provisional and absentee ballots. The EPI reports that both Kansas and Ohio tend to issue and reject many more provisional ballots than average. If either race requires a recount, adjudicating provisional ballots will probably be a major point of conflict.
In the future, the EPI will measure more security items — although observing increased election security is more difficult than observing increased voter convenience.
The EPI was developed when most voters cast ballots in their neighborhoods on Election Day. No longer. In 2016, 40 percent of ballots were cast before Election Day, up from 30 percent in 2008. As this number grows, the EPI will need to assess the many new models of convenience voting.
Elections in America are getting better, but they are not perfect. The latest EPI shows that we can use objective metrics to chart any policy change aimed at improving voting, and that it’s not as difficult as we thought.
Charles Stewart III (@cstewartiii) is the Kenan Sahin Distinguished Professor of Political Science at MIT and the director of the MIT Election Data and Science Lab. Before that, he was active in the leadership of the advisory committee that worked with the Pew Charitable Trusts to develop the EPI.