| This research presents model to identify on-network and off-network factors influencing high pedestrian crash locations. The models are based on statistical analyses of crash data. These are illustrated using pedestrian crash data for the Las Vegas Metropolitan area, in Nevada, USA. Crash data were collected for a five year period (1996--2000) and analyzed. These pedestrian crashes are classified into intersection crashes and non intersection crashes based on distances from the reference street. A Geographic Information System (GIS) software program was used to develop crash density maps to help select high risk intersection locations and non intersection locations. Descriptive analysis technique (e.g., frequencies, rates, and proportions) are then applied to identify to screen the variables that are most likely related to these pedestrian risk locations. On-roadway and off-roadway variables were considered in this step. Next, principal component analysis and binary logistic regression analysis are the statistical methods used. The analyses indicate that intersections prone to higher pedestrian risk generally had the following predictors: areas: low income households, significant elderly population exposure, undivided roadways, commercial land use areas with six lane roadways, high concentration of African American population with poor light conditions and six lane roadway, and divided roadways. Likewise, pedestrians at non-intersection locations are more prone to risk close to residential and commercial areas, six lanes undivided roadways, low income residential area with dark or dusk light roadway conditions. The methods developed in this research are applicable to other urban settings. The results provide critical inputs to identifying on-roadway and off-roadway factors affecting pedestrian safety and to develop strategies to address the same. |
