Assessment of potential site selection methods for use in prioritizing safety improvements on Georgia roadways

With over 40,000 people continuing to die on US roads each year, the US government has heightened the awareness of critical safety issues with the passage of SAFETEA -- LU legislation in 2005. The plan requires each of the states to develop a Strategic Highway Safety Plan (SHSP) and incorporate data-driven approaches to prioritize and evaluate program outcomes; else federal funds will be redirected. Seeking to meet the new demands for data-driven approaches, many states are struggling to identify data collection/maintenance requirements for satisfying new approaches to highway safety analysis. Recent research has shown that selecting projects on the basis of crash frequencies and rates are misleading due to selection bias (such as greater emphasis on traffic volume and cash severity etc) and Regression-to-mean phenomena. There are several safety analysis techniques that are preferred over traditional rates and frequencies. These include level of service of safety, empirical bayes method using SafetyAnalyst software techniques. While all the above mentioned methods are macroscopic (giving a bigger picture of the complete road), microscopic analysis could be done using the Interactive Highway Safety Design Model (IHSDM). IHSDM is a set of software analysis tools developed by Federal Highway Administration (FHWA) to evaluate safety on two lane rural highways.;This research aims at assessing the usability, data requirements, data availability and expertise required by different techniques that are deemed appropriate for safety analysis in Georgia. To streamline and reduce the scope of work, Cobb County was chosen as the analysis county because it had been used in a prior development effort and was expected to have the best level of completion and accuracy in the state. The procedure of using the state-of-the-art analytical tools is considered as the most comprehensive safety analysis method. Cobb County data set will be used to test the applicability of the four analysis methods: crash frequency, crash rate, critical crash rate and level of service of safety (LOSS). The results from various ranking criteria (crash frequency, crash rate, critical crash rate and LOSS) will be compared to the actual available crash data and enhanced SafetyAnalyst data.;SafetyAnalyst uses the Safety Performance Functions generated for northern states and it calibrated to Georgia data. SPFs applicable to Georgia data (generated from Cobb County) are compared to the non-calibrated and calibrated SPFs used in SafetyAnalyst. Analysis of costs and potential benefit of using various network screening methods is carried out to weigh the capabilities and limitations of various ranking methods.