Adam Stern

Date of Award


Level of Access Assigned by Author

Campus-Only Thesis

Degree Name

Master of Science (MS)


Resource Economics and Policy


Jonathan D. Rubin

Second Committee Member

Per Garder

Third Committee Member

Kathleen P. Bell


The U.S. Federal Highway Administration (FHWA) estimates that 673,000 people are injured and 7,400 people are killed each year from weather-related crashes. Researchers at the University of Maine estimate an average annual economic cost of $1.5 billion from crash-related expenses for that state alone. Additionally, in direct costs, Maine spent $98 million in 2008-2009 on winter road maintenance efforts to improve safety and mobility (MCSCPP, 2010). In order to efficiently allocate winter road maintenance resources, understanding the relationship between road safety and varying levels of adverse winter weather is necessary for managers and policy makers. No study has quantifiably analyzed this relationship for Maine. This thesis uses daily crash and weather observations to analyze the relationship between varying levels of temperature and snowfall, and crash frequency in the state of Maine. In addition, snowstorm salt data for the Bangor area is used to measure the effectiveness of winter road maintenance (WRM) across road priority levels. Weather and exposure variables are structured to account for the non-linear relationships with daily crash totals. Crash data found in police reports from 1989 to 2008 are used and multiple statistical models are tested. Using the Akaike Information Criterion (AIC) values to measure model goodness of fit, Generalized Negative Binomial (GNB) regression estimates proved a better fit than Standard Negative Binomial (NB), Variance Weighted Least Squares (VWLS) and Ordinary Least Squares (OLS) regression models. GNB results indicate that crash frequency increases dramatically with colder temperature and with daily snowfall between one and five inches. Specifically, daily crash totals are shown to increase by over 120% under such conditions compared to non-winter days. Using FHWA estimates for the cost per crash, the results show that days with temperatures well below freezing and snowfall greater than an inch cost the state approximately $1.5 million each in additional crash costs. Results also show a significant safety benefit on roads that are given top WRM priority.