Announcing the 2019 New Initiatives Grant Recipients

Simple. Innovative research in election science can:

• make election administration more efficient
• provide empirical evidence for policy decisions
• improve voters’ confidence

…and that’s just the start of it. Creative approaches to the scientific study of elections and election reform can turn old questions into catalysts for new ideas— or use brand new questions to open up unique or unexplored areas for research.

When the MIT Election Lab started in 2016, we saw that support for this kind of research could be difficult to secure, especially for emerging scholars or those outside of the more traditional political science discipline. So in 2017, (with the generous support of the William & Flora Hewlett Foundation,) we started a program for New Initiatives in Election Science.

Each year, we’ve invited researchers from around the US each year to apply for funding under this program, asking them to pitch a project that proposes an innovative or interesting approach to the scientific study of elections and election reform. In doing so, we’ve been able to support all sorts of research focused on election management, no matter what discipline it examined the issue from: computer science, psychology, public administration, or something else entirely. We’re also proud to award a significant percentage of these grants to projects headed by Ph.D. candidates and junior faculty members, supporting emerging scholars in the space.

Perhaps we’re biased, but we think the research the program is able to support is laying important groundwork for the conduct of elections in the United States. New research — and a widening field of scholars engaged in it — benefits all of us who consider ourselves geeks of the election variety.

Each time the application period rolls around, it’s both exciting and nerve-wracking to read through the fascinating research proposals we receive. All the proposals are read and discussed by a panel of our advisors, who take a number of things into account as they weigh their recommendations for funding, from proposal’s design and the applicant’s qualifications, to the significance of the research project and its relationship to our own goals at the Election Lab.

Past grants have covered topics from voters’ perceptions and political trust to primary election laws, voter list maintenance, and precinct boundaries. We’re thrilled to be announcing the 2019 recipients of our New Initiatives in Election Science grants! Read on below for summaries of their planned research, or check out the full list of recipients from all years on our website.

Voter Detection of Anomalies on BMD Ballots

Michael D. Byrne and Philip Kortum (Rice University)

As a result of the Help America Vote Act, many jurisdictions in the U.S. began using “direct recording electronic” systems (DREs). These systems were more accessible than a paper ballot, but opaque on how votes were recorded and plagued by security problems. To alleviate fears that the machines were cheating, some had printers attached which produced a paper tape known as a “voter-verified paper audit trail” — but these in turn were difficult for administrators to manage and did little to alleviate security concerns. In response, voting system vendors have taken a new approach, called a “ballot marking device” (BMD). A BMD is a computer interface to a ballot, which provides the potential usability and accessibility advantages of DREs, but still produces a paper ballot. The paper ballot is what voters submit to the ballot box and the record that is counted.

A compromised computer can still produce a ballot that does not match the selections made by the voter. However, it cannot do so invisibly, since voters can examine the BMD-produced ballot before they cast it. The issue is that it is not clear that voters can or will do so, or how accurate they are when they do. Our proposed research is straightforward: we will recruit a sample as representative as we can, have them vote in a mock election on a BMD which will be programmed to generate a paper ballot that does not match their selections, and measure the rate at which they detect the anomalous ballots.

The Maine Event? The Gap Between Expectations and Reality in Election Reform

Jesse T. Clark (MIT)

Ranked-Choice Voting (RCV) has been the subject of increasing focus by the media and political actors, especially those seeking to reform American elections. Reformers claim that RCV will bring about more civil campaigns, increase voter turnout, and increase voter confidence. While these claims would make RCV highly desirable to election officials if true, analyzing the actual impact of RCV has been greatly hindered in two major ways. This study seeks to overcome these obstacles in two ways.

First, I will to leverage the first-ever statewide implementation of RCV in Maine to understand the impact of RCV on a variety of outcomes, including voter confidence, campaign civility, voter participation, and voter choice. Second, I will implement a large-scale nationally representative study to understand perceptions and beliefs around RCV and other types of election reform. This dataset will then be used as a baseline by which researchers will be able to measure future implementations of RCV, which have been proposed in various states in the past year.

Finding Families Affected by Felon Disfranchisement

Laurel Eckhouse (University of Denver), Allison Harris (Penn State), Hannah Walker (Rutgers), and Ariel White (MIT)

Many people in the United States are convicted of felony crimes each year, and in some states these convictions render them ineligible to vote. Though available evidence suggests that people facing felon disenfranchisement would not often vote even if they were allowed to, some have argued that we should look beyond individual effects to consider how felon disenfranchisement laws affect the families and communities of people who are subject to them.

We have very little information, even descriptive, about the political participation of people who are affected by felon disenfranchisement laws “secondhand.” Do people vote less often when their spouse becomes ineligible to vote? Might they go back to voting if their spouse were reinstated? And are there interventions that could reincorporate both re-enfranchised people, and perhaps their families, into political life? We propose to use administrative data to begin answering these questions, and hope that the project undertaken this year will also lay the groundwork for larger-scale observational and experimental approaches to answering them.

Do Post-Election Audits Increase Confidence in Elections?

Christian R. Grose and Nathan K. Micatka (University of Southern California)

Post-election audits are conducted in more than half of the U.S. One reason to conduct post-election audits is to validate that the votes were counted correctly. Proponents of post-election audits say that audits yield increased public confidence in the election , but this has not been theorized nor tested empirically by scholars. Our project will conduct research to examine whether the confidence (1) of voters and (2) of local election administrators increases when jurisdictions have been subject to a random post-election audit.

Why does studying the effect of post-election audits on voter and election administrator confidence matter? Confidence from both voters and local election administrators in the electoral process is central to a well-functioning democracy. In the absence of voter confidence, democratic systems can break down; if local election administrators have low confidence or satisfaction in their state election administration, then this is normatively concerning. If post-election audits increase confidence in election outcomes, their widespread adoption may be helpful not only for assessing the accuracy of the Election Day vote count but also for increasing confidence in elections for voters and local election administrators.

Using Data Science to Create Election Science

Martha Kropf, Jason Windett, and Samira Shaikh (University of North Carolina at Charlotte)

The partisan selection of election officials has been controversial for many years, including through the 2018 midterms. Allegations of misconduct against chief election officers should be concerning, not least because administrators are asked to create and codify administrative rules, policies, and otherwise fill in and carry out the laws created by state legislators. Even with regulations guarding against bad practice, sometimes administrative decisions go completely under the radar. Likewise, state court decisions are relatively invisible as well, when statutes do not address a particular situation, or there is disagreement on how to interpret a law. Local or state actions such as a court ruling on when a polling place might stay open late, or how to handle a natural disaster at or near the location of an election, are indispensable part of running a complex intergovernmental structure — yet often seem insignificant next to national, high-profile decisions.

Scholarship indicates that there may be some very real implications for democracy where these institutions are concerned. Our aim is to construct a publicly accessible database consisting of all election-related administrative rules and court decisions (state and federal) along with related statutes. We expect to examine whether political actors may be utilizing these relatively under-cover institutions to further partisan electoral goals. Furthermore, using data science to will also allow us to create a searchable database (for scholarly and public use) and to open the data to an interdisciplinary group of scholars to expand the types of institutions we are able to factor into analyses of election administration and reforms as well as political behavior.

Early Voting and the Implications of Non-Simultaneity for Political Discussion

Todd Makse (Florida International University)

As early voting has become a staple of American elections, empirical studies of its effects have zeroed in on questions related primarily to voter turnout; fewer have drilled down on the ways in which early voting affects elections themselves. Specifically, early voting creates a division in the electorate — and in discussion networks — between voters who have already cast a ballot (“confirmed”) and those who have some probability of voting between a given point in time and Election Day (“potential” voters). This distinction may not be apparent in all networks, but when the distinction is present, it may serve to create dynamics that echo important findings from the literature.

I argue that impacts of early voting’s violation of the simultaneity principle can be fruitfully explored by focusing on two key aspects of political discussion: credibility and persuasion. The presence of confirmed voters in social networks alters network contours in two meaningful ways. First, confirmed voters have acted on their preferences in an immutable fashion, meaning that their expressed preferences (e.g. for one candidate over another) may be perceived as something more than “cheap talk.” Second, confirmed voters are no longer persuadable (in the sense that they cannot alter their vote) but are still capable of persuading others, creating an asymmetry in discussion networks. This project will use a two-stage experimental design that simulates crucial aspects of the decision-making and social interactions that occur in an early voting context.

Media Myths and Public Perception of Polling Place Wait Times

Christopher Mann (Skidmore College) and Kathleen Searles (Louisiana State University)

News media coverage on Election Day often focuses on wait times at polling places. News outlets are incentivized to cover long wait times, as they make for more dramatic television than well-run and efficient polling places — despite long waits being rare. Coverage of long wait times at the polls potentially deters voting participation due to public perception of long wait times, reduces confidence in election administration, and undermines trust in government institutions. News outlets have changed the way they cover exit polls and early returns after research suggested such calls affect voter turnout, and yet we know little about the potential deleterious effects of wait time coverage.

This two-part research project will first conduct a content analysis of television news on Election Day to characterize voting coverage, and then a randomized experiment to assess the impact of this coverage on public opinion and voting behavior. We believe this study will be the first systematic examination of news media coverage of polling place lines on Election Day.

Happy Birthday: You Get to Vote!

Ellen Seljan (Lewis & Clark College) and Paul Gronke (Reed College)

Automatic Voter Registration (AVR) systems register to vote all eligible individuals who conduct a transaction with proscribed government agencies, most commonly the Department of Motor Vehicles (DMVs). Sixteen states plus the District of Columbia have authorized some form of this policy since 2015, with several more states considering adoption. Notably, this is more than the number of states adopting “active” motor voter laws in the 1980s and early 1990s, a trend that sparked national legislation to be passed in 1993. Yet to date, there is no peer-reviewed analysis of the effect of AVR on voter turnout.

While it is clear that AVR has been a tremendous success in terms of enlarging and diversifying the voter rolls, it is more challenging to assess its impact on voter turnout. Causal inference is impaired by the absence of a true counterfactual; it is impossible to know what turnout in subsequent elections would have been in the absence of AVR. Both temporal analyses of voter turnout before and after implementation, and comparative analyses of voter turnout relative to non-AVR states, is subject to this important limitation. Our analysis will use a different approach to achieve a more precise, causal estimate of the effect of AVR.

Does Malapportionment and Voter Suppression Cause Disproportionality and Inflate Electoral Bias?

A district-level analysis of US House elections (1870–2018)
Bernard Tamas (Valdosta State University)

The goal of this project is to gather and organize clean, accurate district-level election result data from the US House of Representatives from 1870 to 2018 — which I will use to estimate the relationship between voter under-representation (including voter suppression), and disproportionality or electoral bias within US House elections.

This project expands upon my main finding in my most recent article, in which I presented a new measure of electoral bias, called DPIx. More specifically, DPIx measures asymmetrical disproportionality, or when election rules inflate the number of seats one party or set of parties wins in relation to another party or set of parties. This project is the next step in that research, aimed at determining what factors other than gerrymandering cause spikes in asymmetrical disproportionality. Specifically, for this project, I hypothesize that in single-member plurality (SMP) electoral systems, any type of voter underrepresentation, whether in the form of malapportionment, voter suppression, or even simply sagging turnout, can increase disproportionality against the party that represents the underrepresented group, especially if that underrepresentation is in any way geographically concentrated.

A New Database of Local Legislative and Executive Elections

Chris Warshaw (George Washington University)

For decades, the study of local politics in the United States focused primarily on case studies of individual cities. There were few large-scale studies published in mainstream political science journals, despite the fact that there are nearly 90,000 local governments in the country. The vast majority of elected officials serve at the local level; local governments employ over ten million workers and collect nearly a quarter of the nation’s total tax revenue. There is also more institutional variation at the local level than the state or national levels, which gives scholars an opportunity to use local politics as a laboratory on the effects of institutions in American politics.

In recent years there has been a surge in the study of representation and elections in local politics. However, the lack of a systematic database of local election results continues to be a barrier for more progress in the literature on local politics. In this project, I plan to assemble a comprehensive database of local legislative (e.g., city councils, county commissions, and school boards) and executive (e.g., mayors) election results in medium and large localities over the past three decades.

How Do Voters React to Problems When Casting Their Ballot?

Logan Woods (University of Michigan)

In the 2018 midterm election, students at the University of Illinois campus voted at unexpectedly high levels, leading to long wait times for students and unpleasant experiences for poll workers. Polling places on campus were reportedly understaffed, and the typical rooms used for voting were unavailable, forcing these voting stations into smaller rooms. There are both innocent and nefarious possible explanations for these problems (that can occur at any polling place) on Election Day. This project, part of my dissertation, attempts to improve our understanding of how voters react to these situations.

Although absentee voting and vote-by-mail are becoming more common, approximately 60 percent of voters still cast a ballot in person on Election Day. For voters casting a ballot in person, a number of factors can affect their experience — they may face a long wait, issues with their registration, or a malfunctioning voting machine. Even if a voter does not personally experience these problems, they may hear about it from friends, family, news reports, or social media. In this research, I propose that two possible causes voters might assign to polling place problems are 1) well-meaning incompetence by election officials, as well as 2) intentional malfeasance on the part of those election officials. These two possible explanations of Election Day problems may produce different voter reactions, making voters more or less likely to cast a complete ballot or participate in future elections, and affecting their feelings about politics in general. I plan to test these possibilities with a survey experiment describing polling place problems and varying the reported cause of those Election Day problems.

Please join us in congratulating the 2019 New Initiatives Grant recipients! We can’t wait to see where their research leads them.