Contributed Poster Presentations: Transportation Statistics Interest Group

Rapid advancements in crash modeling have yet to fully integrate multiple behavior-driven crash types and their severity outcomes within a single, scalable framework, specially in a way that captures the sequential nature of these behaviors where one risky action may amplify another. This study proposed an integrated multivariate econometric framework to jointly model crash frequency and severity outcomes for three major behaviorally driven crash types: alcohol-related, distraction-related, and aggressive-driving-related crashes. Specifically, using Oregon's 2022 census block group-level crash data, we propose an Integrated Multivariate Negative Binomial – Generalized Ordered Probit Fractional Split (IMNB–GOPFS) model to analyze these dimensions simultaneously while accounting for the sequential nature of the behavioral crash types. A comparison exercise in terms of model fit and predictive performance reveals the superior performance of the proposed framework over traditional non-integrated approaches, thus highlighting the existence and importance of capturing such interdependencies among behaviorally driven crashes. 

Growing questions about the role of gasoline-powered cars have intensified debate over the adoption of electric vehicles (EVs). Using merged state-level data on vehicle fuel types, economic indicators, and environmental outcomes from 2016 to 2023, we analyze the impact of EV adoption on gross domestic product (GDP) and carbon dioxide (CO2) emissions. While prior research finds a linear positive relationship between EV adoption and GDP, our analysis reveals a U-shaped effect. At low adoption levels, increasing EV adoption is associated with GDP declines, whereas at higher adoption levels, increases lead to GDP growth that offset initial losses. Unlike national-level studies, we showcase variation across U.S. states driven by differences in population and political alignment, with more populous states experiencing stronger positive economic effects. Environmentally, EV adoption is linked to lower CO2 emissions, although preliminary nonlinear results suggest that emissions may increase at very high adoption levels. This research highlights the trade-offs regarding increased EV adoption, underscoring the need to consider economic and environmental outcomes concurrently in policymaking. 

Americans spend approximately one-third of a year of their lifetime waiting at red lights. Current traffic light systems in many suburban areas operate on timers but ignore actual vehicle arrival patterns, causing frustrating consecutive red light stops.

This study proposes a low-cost probabilistic approach to optimize traffic light timing without requiring sensors or smart technology. We model vehicle arrival times at sequential lights by treating delayed start, acceleration, and cruising speed as normally distributed variables. Using these distributions and known distances between lights, we calculate arrival time distributions at each subsequent intersection. Green light timing is then optimized to maximize the probability that vehicles stopped at the previous red light encounter green at the next light.

This approach reduces consecutive red lights, a primary source of driver frustration and road rage, while requiring only one-time adjustments to existing timer-based systems at virtually no cost. Simulation results demonstrate meaningful improvements over baseline synchronized timing. 

Selecting the number of communities is a fundamental challenge in network clustering. The silhouette score offers an intuitive, model-free criterion that balances within-cluster cohesion and between-cluster separation. Albeit its widespread use in clustering analysis, its performance in network-based community detection remains insufficiently characterized. In this study, we comprehensively evaluate the performance of the silhouette score across unweighted, weighted, and fully connected networks, examining how network size, separation strength, and community size imbalance influence its performance. Simulation studies show that the silhouette score accurately identifies the true number of communities when clusters are well separated and balanced, but it tends to underestimate under strong imbalance or weak separation and to overestimate in sparse networks. Extending the evaluation to a real airline reachability network, we demonstrate that the silhouette-based clustering can recover geographically interpretable and market-oriented clusters. These findings provide empirical guidance for applying the silhouette score in network clustering and clarify when it is most reliable.