Science lab spotlight

Unraveling the intricacies of American elections

The MIT Election Data and Science Lab aims to improve the efficiency and security of democratic processes

On Nov. 7, 2000, the votes in Florida were, by the slightest margin, in favor of President George W. Bush. This narrow margin called for a recount of all the votes cast in Florida for the 2000 presidential election. Over the next few weeks, a number of oddities were discovered in the process — an unusual number of votes were cast for third-party candidates, which occurred as a result of a confusing ballot layout, and some African American citizens had been misidentified as felons, making them ineligible to vote. 

These issues of voter registration and the lack of security in the election process caught the attention of MIT Professor Charles Stewart, Kenan Sahin Distinguished Professor of Political Science and the Founding Director of the MIT Election Data and Science Lab (MEDSL). “The thing that I learned, as well as everybody else in America at the time,” said Stewart, “was that it was possible for you to be active and to vote, and for that vote not to count.”

Voter insecurity issues can take many forms, from cyber interference to a paper not being properly delivered, and Stewart commented that “in states that reject a lot of registrations, people should consider whatever form of registering as not being particularly secure compared to another state where they accept a very high fraction.” In fact, Stewart spoke of a related occurrence that happened at MIT almost one year ago. On National Voter Registration Day, a student group conducting a voter registration drive collected registration papers filled out by MIT students. However, a librarian later shredded these registrations accidentally. These MIT students were under the impression that they had been successfully registered to vote in the 2018 midterm elections, when in reality, that was not the case.  

In order to combat this national concern, Stewart and his lab are working with the Pew Charitable Trust to develop the Election Performance Index, or the EPI. The EPI was developed to encourage states to improve their voting practices by providing metrics on their performance using data on absentee ballots, voter turnout, and registration rejection rates. Despite all the negative media coverage of the voter experience and the efficacy of elections, the EPI has made a considerable impact on the way elections are run. “States certainly pay attention to the index whenever it’s revealed, and we can actually point to certain policy changes that have been made in the states as they’ve tried to get the ranking up,” Stewart reassured. “The evidence by and large shows that elections have actually gotten better and more professional so that in most states, if you’re a voter, you can be assured that the vote being cast is being counted and it’ll be convenient to you.”

Another goal of the MEDSL is to bring hard data, numbers, and science to the study of elections. As Stewart put it, he wants “to help people understand how elections are managed from a scientific, objective standpoint, and try to put their attitudes, partisanships, prejudices out the door when they think about election administration.” To this end, the MEDSL tries to regularly publish election returns of all precincts in the United States, which can be quite an undertaking due to the lack of uniformity in election reports. By utilizing data science techniques such as web data extraction and optical character recognition (OCR) on handwritten documents, Stewart and his team have been able to acquire datasets, publish them, and create publicly-available redistricting software based on the datasets, making it easier for citizens to be informed and get involved in the democratic process, specifically in the upcoming redistricting round during the 2020 census. In fact, Stewart added that “there are a number of states that mandate the participation of citizens whenever they draw districts. Part of it is to guard against gerrymandering… Citizens can also go into state legislative hearings or citizen commissions and say, ‘I’ve drawn a district that meets certain desirable criteria, and I want to offer it as evidence that [citizens] can draw good-looking districts.’” In the case where state officials try to develop plans hidden from the public eye, citizens can use this software to question, or potentially litigate, the administration if they believe anything is unjust or corrupt.

Ultimately, people across the globe look to MIT as a distinguished institution and center of knowledge, and the MEDSL strives to uphold this reputation. “Anything that’s a controversy, I want to be in the middle of, and provide the data, provide some insights, so that other academics can work in this area, and so that the press and public officials have access to the best research and researchers in these fields,” said Stewart. “It’s a role that MIT frequently plays in scientific questions.” Looking forward to the 2020 presidential election, with all the hyper-vigilance surrounding cybersecurity, the press and election officials are strengthening defenses to fend off potential cyber attacks. Stewart hypothesizes that “something will go wrong, but we don’t know what it is yet” — but when it does come, he is ready to tackle it head-on and keep the public well-informed with data.

Update 11/3/19: This article has been updated to correct a misspelling of Charles Stewart's name.