Date of Award

2019

Degree Type

Thesis

Degree Name

Master of Science in Systems Engineering

Department

Mechanical, Industrial and Systems Engineering

First Advisor

Gretchen Macht

Abstract

Election officials at the state and local levels begin operations and logistical planning several months before election day. Current research efforts focus on creating basic systems to facilitate the collection and synthesis of polling place data by election administrators and poll workers. Practically all the current methods involve the manual collection of this data, and then some aggregated form is utilized in decision-making processes. The research contributes to the voting-systems literature in two ways. First, it broadens the scope of knowledge about check-in processing time variation both within and between precincts. Secondly, it proposes a methodology for using the EPB transaction logs to estimate arrival rates using a Hidden Markov Model.

Check-In processing time observations are collected through time studies during the 2018 Midterm elections at seven precincts throughout Rhode Island. An analysis of check-in observations revealed that processing times are reasonably similar both between and within precincts. Check-In observations are then used to model a stochastic process time distribution for four precincts in Providence, Rhode Island. The process models are combined with electronic poll book transaction logs to simulate voter arrival times. The count of simulated arrivals over discrete 15minute intervals are used to populate an observation sequence. Multiple observation sequences are used to compute parameter estimates for a Discrete-time Poisson Hidden Markov Model (dt-PHMM).

Available for download on Monday, August 16, 2021

Share

COinS