MEASURING THE EFFECTS OF SYSTEM CONDITIONS ON U.S. ELECTIONS THROUGH SIMULATION

Elections are fundamental to the success of democracy in the United States (U.S.). For decades, researchers from various domains have investigated voting in the U.S. with many aims. One focus of investigation has been the act of in-person voting, sparked by instances of voter disenfranchisement and severe delays at polling locations. Despite many efforts to improve voting processes, some voters continue to face multi-hour long waits to cast a ballot. This research builds a robust simulation framework that is applied iteratively to assess several system conditions to quantify their exacerbation of voting delays. Once quantified, simulation results are analyzed through the novel application of several statistical methods to evaluate the effects of system conditions on the performance of voting processes. Throughout this work, the effects of COVID-19, polling location consolidation, and the allocation of ADA compliant voting equipment are investigated through the application of discrete-event simulation utilizing real election data. The results of each simulation set are analyzed through a range of methods including naturalistic case studies, the novel application of simulation optimization through metamodeling, and statistically robust comparisons of system performance to understand the relationship between system conditions and system performance (i.e., voter time-in-system, average wait time, and maximum wait time). The findings of this work indicate that system conditions directly impact system performance and exacerbate voting obstacles. Elections held during COVID-19 are found to experience significantly longer wait times due to sanitization processes required for viral spread mitigation. Assessing polling location consolidation strategies through simulation optimization, findings suggest that specific strategies for consolidating voters and equipment require unique resource allocations to accommodate the electorate. On resource allocation, hypothetical simulations indicate that current requirements on the allocation of ADA compliant voting equipment are likely ineffective and minimum requirements must be tailored to specific counties, precincts, or polling locations to enfranchise voters with disabilities. This work aims to demonstrate and quantify the effects of various system conditions on elections in the United States through the application of robust and data-informed methods. Current scholarship and guidance on election planning lack the incorporation of unique, temporary, or evolving conditions when investigating the voting process, leading to the limited applicability of research findings. The results of this dissertation’s analyses suggest that broad, “one-size-fits-all”, and “Rule-of-Thumb” election planning methods are often ineffective and lead to voter disenfranchisement through long wait times. Further, methods for assessing systems that consider unique conditions are exemplified. Beyond this work’s relevance to the literature, results suggest that applied engineering tools and methods can play a role in facilitating elections