According to the Massachusetts Department of Transportation’s (MassDOT) report on Vulnerable Road User Safety Assessment, 41% of all fatal pedestrian crashes  in Massachusetts between 2016 and 2020 occurred around a transit bus stop. This study aims to analyze pedestrian and driver behaviors  to identify key factors contributing to pedestrian crashes near bus stops through advanced data processing and engineering methods.  

            Data was collected at 20 selected bus stops across Massachusetts using  MioVision units.  MioVision units are a sophisticated camera-based data collection devices specifically designed for roadside deployment. An advanced video analysis rubric was developed, and researchers were assigned to independently code and record pedestrian behaviors. Given the subjective nature of human interpretation and potential for error, a logical coding protocol was designed and implemented to statistically measure their agreement and resolve human entry related errors. Additional refinements were made to the protocol based on pilot tests to ensure high reliability. Furthermore, a custom-designed weighted probability model was developed to process biased data by performing an inner join of entries from different researchers. The validated video data from the weighted probability model was then geographically linked to specific bus stops and integrated with infrastructure, traffic, and crash datasets.  

          These methods provided an immediate practical value by expanding traditional methods utilized for pedestrian safety research that may be subjective in nature. Through integrating behavioral insights with infrastructure analysis and crash prevalence, the study establishes new standards for enhancing pedestrian safety near transit.