Resources

This paper compares in-person versus absentee voting, finding that voters randomly assigned to in-person voting reported significantly higher levels of voter confidence than those assigned to absentee voting.

In this paper, authors generate voter wait-time estimates using an indifference-zone generalized binary search method to optimize and determine resource allocation, such as electronic poll books and voting machines, to reduce wait times.

In this paper, authors test whether voters can detect malicious manipulation of ballot-marking devices, finding low detection rates and showing that signage and poll worker prompts can modestly improve verification rates.

In this MS thesis, Bernardo investigates how ballot-length metrics (words, questions, selections, pages, sheets, bilingual status) affect voting errors during the 2018 Rhode Island midterm election. He uses logistic regression models that control for municipal- and precinct-level demographics to analyze machine-based, human-machine interaction, and ballot-marking errors. Bernardo finds that longer ballots and urban precincts significantly increase the odds of voting errors, with implications for ballot design and jurisdiction-level oversight.

In this MS thesis, Houghton develops a methodology to estimate voter arrival rates at polling stations using electronic poll book transaction logs. It includes service time observations collected through time studies during the 2018 Rhode Island midterm election across seven precincts. The study applies a Hidden Markov Model to infer voter arrival patterns from the check-in records. Finds that e-pollbook logs offer a scalable, less labor-intensive alternative to manual observation for estimating arrival rates.

This paper analyzes how vote centers influence voter turnout in various election types in Texas, showing that their effect depends on the election context and voter demographics.

This paper finds that large numbers of voters do not perceive their ballots as secret and harbor doubts about the institution's ability to keep them private, with perceptions varying by voting method and polling place design.

This research focuses on whether voters’ confidence is shaped by the racial or ethnic representation of poll workers and election staff.

In this paper, authors develop models to estimate voter service times from voting machine log data, providing election officials with a scalable approach to analyze and improve polling place operations.

This paper finds that Black and Latino voters generally wait longer in line to vote than white voters. These instances occur with longer wait times, usually observed in neighborhoods with higher minority populations compared to predominantly white neighborhoods.

This PhD dissertation examines U.S. election administration through three empirical studies. Three core contributions: (1) documentation of racial disparities in voter wait times across polling places, showing that Black and Latino voters wait significantly longer than white voters due to differential resource allocation; (2) analysis of how the partisan and demographic composition of jurisdictions predicts administrative resource levels; and (3) estimation of the downstream turnout consequences of long waits, showing that each additional hour of waiting reduces the probability of future voting. Chapters were subsequently published in Political Science Quarterly and Electoral Studies. Advisor: Gary King.

This paper finds that voters with disabilities face significant barriers to in-person voting, including inaccessible polling places and equipment, which contribute to lower turnout rates among this group.