Graduation Semester and Year
Spring 2026
Language
English
Document Type
Dissertation
Degree Name
Doctor of Philosophy in Urban Planning and Public Policy
Department
Urban and Public Affairs
First Advisor
Jianling Li
Abstract
Traffic crashes remain a major transportation, and policy challenge in the United States, with substantial human and economic costs. This dissertation examines injury crash risk through a multi-level framework that integrates human behavior, roadway characteristics, environmental conditions, demographic factors, and legislative context. The study is organized into three empirical parts that move from crash-level analysis to regional road-segment analysis, and finally to comparative policy analysis.
Part A investigates crash-level injury severity and culpability factors in Tarrant County, Texas. Using crash data from 2016 to 2022 and Bayesian logistic regression, the analysis shows that higher speed limits, traffic exposure, roadbed width, tailgating, violations, intoxication, and technical or parking-related issues are associated with higher odds of injury crashes, while wider shoulders and additional lanes reduce injury risk. The findings also reveal important demographic and interaction effects, particularly involving age, race/ethnicity, and inattention under higher-speed roadway conditions.
Part B examines injury crash incidence at the road-segment level across the North Central Texas Council of Governments region. Domain-level principal component analysis (PCA) is used to derive composite factors, which are then modeled using mixed-effects negative binomial regression. The results indicate that traffic exposure is the strongest predictor of injury crash incidence, followed by substance use and road type. Significant county-level heterogeneity further suggests that regional crash risk is shaped not only by segment characteristics but also by broader contextual and institutional conditions.
Part C extends the analysis to a comparative state context by examining marijuana-related injury crashes in Texas and Colorado. Using descriptive statistics, Welch’s t-tests, and zero-inflated negative binomial regression, the study finds that marijuana-related exposure, alcohol, other drugs, driver behavior, roadway context, lighting, and weather all contribute to injury crash frequency. However, after controlling these factors, Colorado experiences fewer injury crashes than Texas, suggesting that legalization alone does not determine traffic safety outcomes and that broader policy, enforcement, and institutional environments matter.
Collectively, the dissertation demonstrates that injury crash risk is best understood as a multidimensional and multi-scalar phenomenon. It contributes theoretically by advancing an integrated view of crash culpability, methodologically by combining crash-level, road-segment-level, and state-level comparative policy analysis, and practically by offering evidence to support international Vision Zero project, USDOT’s Safe System Approach (SSA), and more context-sensitive transportation safety planning.
Keywords
traffic safety, injury crashes, crash severity, culpability, Bayesian logistic regression, mixed-effects negative binomial regression, zero-inflated negative binomial regression, marijuana policy, Vision Zero, Safe System Approach.
Disciplines
Urban, Community and Regional Planning
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
Recommended Citation
Suma, Jessica T., "Understanding Culpability Factors in Traffic Crashes: Analyzing Causes and Policy Impacts Across the NCTCOG Region and a Comparative State-Level Evaluation" (2026). Planning Dissertations. 2.
https://mavmatrix.uta.edu/planning_dissertations2/2