Masters Thesis Defense - Benzaman
- Monday, April 10, 2017 at 2:00pm
- Wilson Hall, Room 1-148 - view map
AN ANALYSIS OF ROAD TRAFFIC FACTORS AND ROAD SAFETY STRATEGIES THAT PREDICT ROAD FATALITIES OVER TIME ACROSS FIFTY STATES IN USA
Strategizing to decrease the state wise road fatalities is an important issue in road safety research in United States. But obtaining information on a variety of variables on economic, socio-cultural, demographic and political aspects at state level are a difficult task. The public databases sometimes do not provide full information on the variable or the factor due to missing data. If these variables are neglected from the analysis because of missing data points a lot of valuable information are lost in the process. Therefore, an approach to handle missing data has been considered as one of the many objectives of this study. In order to impute the missing data, multiple imputation method was chosen. After the data imputation, identifying the significant variable associated with road fatalities in 50 states was the secondary objective. This was done by linear regression model which revealed that the top reasons for road fatalities are drunk driving, distracted driving and number of unemployed people. During the process of data collection, it was observed that data sources did not offer necessary information on road safety culture, behavior, norms, attitude and beliefs which can be related with road fatality. To infer culture and understand the effects this study has offered two solutions. Firstly, analyzing the residuals from random effects two-way panel regression model and generating performance indicator of inferred road safety culture. From in the value of the indicators it was clear which state was the safest twenty years ago and which state is the safest now. Through the change in the value of the indicator, the progress in terms of safety culture were also measured. Secondly, using people’s political views on the democratic party and the republican party as a proxy for the culture resulted in a significant increase in the goodness of fit statistic in linear regression model. In terms strategizing road fatalities across the states, this thesis provides prediction models, significant factors, and performance indicators of safety culture which can be used in state level policy making and developing road safety strategy.