Date of Award
Master of Science in Systems Engineering
Mechanical, Industrial and Systems Engineering
Studies pertaining to election systems have historically identified paper-based voting systems as better performing than most voting system alternatives. Despite this, there is a lack of literature exploring the use of paper ballots from an in-depth perspective. This study investigates different metrics of ballot length (i.e., words, questions, selections, pages, sheets, and bilingual) and how they impact voting errors (i.e., machine-based errors, human-machine interaction errors, and ballot marking errors) during the 2018 Midterm election in Rhode Island. Logistic regression models are developed to measure the relationship between ballot length and voting errors while controlling for municipal and precinct level demographics. The findings indicate that areas with longer ballots and urban areas significantly increase the odds of encountering voting errors. Among the most contributing measures are the number ballot pages, the number of local questions, and the number of candidate selections allowed on a ballot. These factors significantly increased the odds of experiencing voting errors, in some cases as high as 160%. The statewide impact of these errors is presented and opportunities for future work are shared.
Bernardo, Nicholas D., "QUANTIFICATION OF VOTING ERROR: THE 2018 RHODE ISLAND MIDTERMS" (2019). Open Access Master's Theses. Paper 1524.