Dissertation

My three-paper dissertation focuses on how voters react when parties and elections fail at being tools for representation in two scenarios: uncontested elections, and problems voting on Election Day. The first two papers from my dissertation focus on a theory of withdrawn congressional coattails stemming from uncontested races: I propose that candidates down-ballot from uncontested races will suffer electorally from those uncontested races, and find evidence that this is the case.

The first paper of my dissertation evaluates whether candidates down-ballot from uncontested congressional elections suffer an electoral penalty when their party does not run a candidate. I use precinct-level election data from 2016, available from the MIT Election Science and Data Lab, to test my hypotheses, and find that state legislative candidates running down-ballot from a congressional race in which their party did not field a candidate suffer electoral consequences--they can expect anywhere from 20 to 150 fewer votes per precinct, on average depending on the race. Similar effects are found when using a regression discontinuity design, leveraging the top-two primary in California and Washington which can result in single-party general elections. These decreased vote totals are enough, in the aggregate, to change the outcome of close state legislative races. In other words, by not contesting congressional races parties are hurting their chances at winning down ballot races as well.

The second paper of my dissertation examines the individual-level mechanisms for the electoral penalty I show in the first chapter. First, I establish that some survey respondents do believe that political parties play an important role in the candidate nomination process. Second, I use an original survey experiment to examine how people react when their party does not compete. The responses to this experiment suggest that protest voting plays an important role in the electoral penalty.

The third paper of my dissertation evaluates how voters react in real time to problems they face when voting in person. Election administrators make every effort to ensure that voters have positive experiences when casting their ballot, but some voters inevitably encounter long lines, paperwork errors, or other problems while voting. How voters react to these experiences, though, might depend on what they believe the cause of those problems to be. For example, a voter might perceive a long wait to vote as the effect of something benign or positive, such as high turnout. However, another voter may believe that a long wait to vote is the symptom of something more nefarious, such as intentionally unhelpful poll workers. I hypothesize that Election Day problems that are perceived to be intentional malfeasance may make voters more likely to participate in politics, while problems attributed to benign causes may be associated with voters being less willing to participate. To test my hypotheses, I deploy a survey experiment describing Election Day problems on a nationally-representative survey of approximately 2000 respondents fielded in late 2019 and early 2020 in which respondents are presented with a description of Election Day problems. The source of those problems is described as either intentional malfeasance or innocent mistakes by election officials, and a control condition describing Election Day running smoothly. I find that under some circumstances, voters who report problems voting are more likely to cast complete ballots than those who do not. In addition, reading about problems on Election Day--regardless of the source of those problems--makes individuals more likely to vote in a future election, but it does not affect their willingness to participate in other ways. This research is generously supported by the MIT Election Data and Science Lab, and its funder the Madison Initiative of the William and Flora Hewlett Foundation.