Author ORCID Identifier

https://orcid.org/0000-0002-3469-6283

Year of Publication

2020

Degree Name

Doctor of Philosophy (PhD)

Document Type

Doctoral Dissertation

College

Engineering

Department

Civil Engineering

First Advisor

Dr. Nikiforos Stamatiadis

Abstract

In the U.S., road traffic crashes are a leading cause of death. Crash data from the state of Kentucky shows that the per capita crash rates and crash-related fatalities were higher than the national average for over a decade. In effort to explain why the U.S. Southeast experiences higher crash rates than other regions of the country, previous research has argued the region’s unique socioeconomic provide a compelling explanation. Taking this observation as a starting point, this study examines the relationship between highway safety and socioeconomic characteristics using an extensive crash dataset from Kentucky.

The primary goal of this research is to define the at-risk group of drivers based on the socioeconomic and demographic attributes of the zip codes in which drivers reside. This study utilizes crashes that occurred in Kentucky during the period 2013-2016. The quasi-induced exposure technique used assumes that the not-at-fault drivers represent the total population in question and the crash rate measure of exposure is developed in terms of the relative accident involvement ratio (RAIR), which is the ratio of the percentage of at-fault drivers to the percentage of not-at-fault drivers from the same subgroup. With fault status, dichotomous in nature, being the response variable, binary logistic regression is used, which is beneficial when the effects of more than one explanatory variable are examined. The final prediction model estimates the probability of the fault status of the driver based on multiple independent variables.

Logistic regression models are developed to predict the occurrence of single- and two-unit crashes based on socioeconomic variables. The models for single- and two-unit crashes are quite similar to each other. The results indicate that variables such as driver age-group and gender, rurality, poverty level, average conviction, and driver population density of the area are associated with a driver’s likelihood to be involved in a crash. Educational attainment is observed to have an impact only on single-unit crash occurrence. Finally, it is concluded that younger and older drivers residing in zip codes with low socioeconomic conditions have a higher likelihood of causing a crash for both single- and two-unit crashes: agreeing with prior research findings and maintaining the typical U-shape curve of crash involvement. Males have higher at-risk probability in their younger ages than females, while females perform better at their young ages when compared to males. The findings of this research thus identify at-risk groups of drivers who are most likely to be involved in crashes, and potential safety measures are recommended to control the risk of these targeted groups.

Digital Object Identifier (DOI)

https://doi.org/10.13023/etd.2020.152

Share

COinS