Date of Award

May 2017

Degree Type


Degree Name

Master of Science



First Advisor

Xiao Qin

Committee Members

Robert J. Schneider, Jie Yu



The present study aims at analyzing drivers yielding behaviors to pedestrians’ right of way, who are attempting to cross at uncontrolled crosswalks. Three types of variables were identified to be collected for this research including characteristics of the locations as well as demographic, and behavioral characteristics of pedestrians and drivers. The behavioral characteristics of drivers and pedestrians is recorded only when a pedestrian arrives at the crosswalks trying to cross and a vehicle is approaching the intersection; so, the driver makes a decision whether or not yield to the pedestrian waiting to cross. Some behavioral characteristics of pedestrians include the pedestrian’s assertiveness, standing location and waiting time at the crosswalk to find a gap in traffic to be able to cross. The demographic characteristics also include age, gender, race. Some location specific variables include the presence of marked crosswalks, pedestrian crossing sign, near side bus stop, right turn lane, whether or not the location has had pedestrian-vehicle crash, type of land use surrounding the un-signalized intersections, crossing distance, AADT, the distance of last car parked from the intersection, the distance difference between the downstream and upstream signalized intersection to the un-signalized intersection, and the last location specific variable is the distance of uncontrolled intersection from the Atwater park locates in eastside of the city nearby the Lake Michigan. After identifying the variables and instructing the data collection process, the location studies were investigated. Twenty un-signalized intersections were selected that specific characteristics were similar among them to maintain consistency across all locations. Ten different uncontrolled intersections are selected as study locations, which each has had at least two pedestrian crashes in 2010 to 2014, and the other ten are selected as comparison locations, which none of them has had any crashes history in the same period of time.

To analyze the collected data, five different models are proposed using logistic regression and random effect models. Ultimately, the preferred model that has a better goodness of fit is selected. This model well displays that what variables are most statistically significant with the driver yielding behavior. Based on the final model, each variable may have a positive or negative impact on the driver yielding behavior. The variables that cause drivers yield to the pedestrians at crosswalks include the assertiveness of pedestrians to cross, standing in the street, and the pedestrians’ race with the ethnicity of white as well as the second crosswalk marked, nearside bus stop, and the distance of uncontrolled intersection from the Lake Michigan. Some other independent variables that cause drivers not yield to the pedestrian at crosswalks are the type of land use (commercial area), having a crash history, AADT, crossing distance, and the distance difference between the downstream and upstream signalized intersection to the un-signalized intersection. Note that many professionals cited the importance of land use (proximity to commercial districts, downtown,.etc) on driving yielding behavior because of its relationship with pedestrian volumes. This study does not include a variable representing pedestrian volumes, so that could be explored in future studies.

To better illustrate the effect of the variables on the likelihood of the driver yielding, the elasticity analysis was conducted. So, depends on the type of data, they were categorized into continuous and categorical variables. The elasticity from the continuous variable represents that 1% change in crossing distance variable reduces the driver yielding by 15.469%. For categorical variables, the sensitivity of the driver yielding variable is made by pseudo –elasticity. It represents that the existence of the near side bus stop at uncontrolled intersections increases the probability of drivers yielding by 0.54% while the existence of crash history reduces the probability of drivers yielding by 0.82%. It means that drivers still not tend to yield to pedestrians at crashes locations.

Eventually, to improve the drivers yielding behaviors at uncontrolled intersections, five E approaches including engineering, enforcement, education, encouragement and evaluation are recommended. The engineering treatments with the minimum cost have a capability of being implemented in a short period of time. Simultaneously, a designed program for applying the law enforcement and for increasing people’s awareness and education in a longer run is anticipated to have a significant impact on improving the drivers yielding behaviors to pedestrians’ right of way at crosswalks. At the end of the program, through evaluation and comparison of the before and after implementation of the engineering, enforcement, education and encouragement strategies, we can determine if the desired result have been met.

As part of the focus on enhancing traffic safety and reducing fatal crashes at the assigned locations, High Visibility Enforcement pilot program is also recommended. HVE combines highly visible and proactive law-enforcement strategies to target the violated drivers not yielding to the pedestrian right of way at crosswalks. It offers law enforcement agencies a proven alternative for preventing many of the unsafe driving practices that passenger and drivers engage in on roads. By targeting passenger and drivers, they raise everyone’s awareness of the joint responsibility that we all have to drive carefully and share the road safely.