ABSTRACT

This research extends a previously published risk analysis of mail voting in the United States (Scala et al., Risk Analysis, 2022) by applying a structured modeling framework to evaluate risks associated with in-person voting using Precinct Count Optical Scanners (PCOS), which are critical infrastructure within the Government Facilities sector. While PCOS systems are widely used across U.S. elections, (e.g. approximately 70% of Americans voted on a PCOS machine in 2024) a comprehensive risk profile of this technology has not yet been developed at scale. This study aims to systematically identify threats to PCOS-based voting, estimate their relative likelihood, and assess the implications for risk mitigation strategies at the local level.

We constructed an attack tree model encompassing the full temporal PCOS voting lifecycle — including setup, voting, and teardown phases — to capture structural and procedural vulnerabilities. A Delphi panel was convened to assess the relative likelihood of identified threats. We synthesized these expert judgments using utility theory to model system-level vulnerability. To support analysis at scale, we developed a custom software application, the Attack Tree Analysis Tool (AT-AT), which enables the visualization and probabilistic evaluation of over 70,000 unique threat scenarios. This modeling effort represents an updated, comprehensive threat catalog for PCOS systems, extending significantly beyond the last attack tree published by the U.S. Election Assistance Commission in 2009, and the first to extend the catalog into a risk assessment. To note, we are the first academic team to consider threats to elections systemically, and the first to extend a catalog of risk and threat to the relative likelihood of occurrence and impact.
Results from the model revealed the increased complexity and threat surface of in-person voting compared to mail voting. Through sensitivity analysis, we identify the threats that contribute most impactfully to overall system vulnerability — enabling election officials to prioritize high-impact mitigations. These findings have several operational implications: local jurisdictions can use this analysis to focus mitigation resources on the most critical threats, refine poll worker training to emphasize process points with elevated risk, and anticipate areas of potential insider threat during the administration of elections. This work offers a practical, evidence-based toolset for targeted, local-level risk reduction. Considering election security at the local level remains an under-developed area in the risk literature, and this study provides a foundation for future research in this area.

Future research will focus on quantifying the effectiveness of specific mitigation strategies, expanding this framework to support decision-making under resource constraints. Ultimately, this work aims to strengthen the resilience of in-person voting infrastructure by aligning risk modeling with actionable insights for election administrators.

Skylar Gayhart is a 2024 graduate who earned both her Bachelor’s and Master’s of Science in computer science with a focus on cybersecurity and data science. Her thesis focused on ballot-marking devices, specifically exploring their vulnerabilities and potential mitigation techniques. She has been with the lab since 2023 working on updated the EAC attack tree to analyze precinct count optical scanners (PCOS).

Katherine Tan, B.S., graduated from Towson University in 2025 with a Bachelor of Science in Computer Science. During her time with the Empowering Secure Elections Research Lab, she primarily contributed to the development of the Attack Tree Analysis Tool, a web application used to generate and analyze attack trees in support of the lab’s election security research. Her work focuses on developing web applications that support data analysis and management, including projects within the defense sector.

Alisa Martin, M.S., is a 2025 graduate of Towson University with a Master’s of Science in Supply Chain Management. She has extensive professional quality management experience in the Fisheries Division of the Ministry of Agriculture in St. Vincent and the Grenadines. She also holds a Bachelor’s of Science in Biology with Microbiology from the University of West Indies as well as a graduate certificate from the United Nations University in Iceland. Her elections research focuses on relative likelihood and identifying threats of most concern to states and localities.

Dr. Natalie M. Scala is a professor in the College of Business and Economics, a fellow of the Center for Interdisciplinary and Innovative Cybersecurity, and the director of accelerated programs at Towson University. She is also a faculty affiliate at the University of Maryland’s Applied Research Lab for Intelligence and Security. She earned Ph.D. and M.S. degrees in industrial engineering from the University of Pittsburgh. Her primary research is in decision analysis, with specialization in military and security issues, including risk in voting systems. Her work in elections security earned a University System of Maryland Board of Regents Award for Excellence in Public Service, the system’s highest faculty honor.