Degree Type


Date of Award


Degree Name

Master of Science


Civil, Construction, and Environmental Engineering


Civil Engineering

First Advisor

Peter T. Savolainen


Various studies demonstrated that the human factors, driver performance, and the interactions among humans and other elements in the transportation systems significantly contributed to the traffic safety and highway design. Therefore, it is critical to understand driver behaviors to reduce the likelihood of crashes and enhance the design of the highway system. The major objective of this study is to investigate driver behavior, particularly during crash and near-crash events, as well as during the preceding time intervals. Of specific interest is how drivers’ reaction times, deceleration rates, and speed selection vary under different roadway environments. The freeway non-crash and crash or near-crash events were obtained from the Second Strategic Highway Research Program Naturalistic Driving Study (SHRP 2 NDS) dataset and the associated Roadway Information Database (RID). Due to the unique features of the data, the random effect linear regression model with a participant-specific intercept term was utilized to perform the analyses.

The participants’ reaction times of crash/near-crash events were determined to have a mean value of 1.51 sec. and a standard deviation of 1.25 sec. The results of the analysis showed that reaction time varied based upon the type of crash/near-crash event, the gender of the driver, and whether the driver was distracted over the course of the driving event. The driver’s deceleration rates of crash/near-crash events were also calculated in the study. The mean and standard deviation of deceleration rates were about 9.53 ft/s2 (0.30 g) and 4.99 ft/s2 (0.15 g) respectively. The initial speed of braking, the grade of the roadway, and the type of incident presented significant influences on the deceleration rates of crash/near-crash events. Lastly, the mean and standard deviation of travel speed for non-crash and crash/near-crash events were investigated to explicitly understand the speed selection of drivers. On average, normal drivers showed higher driving speeds and less variability. Speed limits and traffic density had relatively consistent impacts on mean speed and speed variance under both baseline and crash/near-crash conditions. However, opposing effects of curves and work zones occurred on the standard deviation in travel speed between two groups. These effects suggested drivers were more likely to put themselves at risk for crashes by failing to reduce their speeds in response to these conditions. Other roadway and driver characteristics such as age, time of day, shoulder width, and weather conditions also somewhat showed influence on average speed and speed variance.

Copyright Owner

Qiuqi Cai



File Format


File Size

78 pages