Abstract:
Pedestrian crashes and fatalities are a significant global concern, especially in developing countries where pedestrians face even more vulnerability. In developed countries, prior research has explored factors contributing to such collisions using statistical models based on reported crash data. Questionnaires have been employed to gauge pedestrian attitudes and behaviors towards safety. Some studies have even adopted systems approach and human factors methods to assess pedestrian safety risk factors from cognitive and systems perspectives. Nonetheless, research in urban roads of developing countries remains limited. Furthermore, these individual methods often narrow their focus on specific risk factors at particular levels and may overlook the intricate interplay among different factors across various levels. Therefore, there is a pressing need for an integrated framework to comprehensively identify potential risk factors, spanning from micro to macro levels. This thesis makes substantial contributions to pedestrian safety risk analysis, both quantitatively and empirically.
Thisresearchfocuses on four primary directions: (1) identifying an appropriate modeling framework and potential risk factors for pedestrian injury severity in developing countries, (2) assessing the attitudes and behaviors of pedestrians, including predictors, within the developing country context, (3) comprehending the decision-making mechanisms of pedestrians in urban road settings concerning crash potential, and (4) ascertaining systems-level risk factors for pedestrian accidents in developing countries. Risk factors identified in the current study are also compared to those established in other developed and developing countries.
Five years of crash data is analyzed employing three statistical models- Multinomial Logit, Ordered Logit and Partial Proportional Odds (PPO) model to identify potential risk factors behind pedestrian injury severity. Analysis shows that PPO model performed better than other two models in terms of identifying more risk factors. Regarding pedestrians’ attitudinal and behavioral aspects, data are collected through questionnaire survey and employed linear regression and structural equation modeling technique to explore predictors of risky behavior. Pedestrians’ verbal data and observational data are collected through ‘think aloud study’ and analyzed with perceptual cycle modeling and binary logistic model technique to identify the potential factors regarding risky road cross decision making. Five pedestrian crash case studies are used to identify systems level risk factors through employing Socio-technical systems based Accimap method.
The analysis uncovers a gamut of risk factors, encompassing attributes linked to the road environment, vehicles, and drivers (e.g., heavy vehicles, unfit vehicles, young and older pedestrians, etc.). It also delineates the factors influencing pedestrian risky behavior and the decision-making process for road crossings. The research spotlights how pedestrians' inattentiveness and deliberate rule violations contribute to riskier pedestrian behavior, with rule violations correlating with crash involvement. This aspect of pedestrian behavior is not explored previously in Bangladesh. The study elucidates that pedestrians’ risky road crossing behavior is influenced by both human psychology (e.g., mental templates of real-world situations) and road environmental features (e.g., vehicle speed, parked vehicles, etc.). These findings are unique in the context of developing countries regarding pedestrians’ road crossing decision making. Furthermore, the study posits that enhancing pedestrian safety necessitates addressing higher-level organizational activities and deficiencies within the transportation system (such as absence of road safety audit and safety facilities in work zones, inadequate planning, design, monitoring, lack of vision, research etc.) instead of singling out specific actors or their activities at lower levels (e.g., blaming drivers for over speeding and reckless driving). These findings are also new to pedestrian safety research, especially in the context of developing countries.
Most of the identified factors in this study are method-specific, with very few overlapping risk factors acknowledged by two or more employed methods. Consequently, the current study underscores the imperative need to amalgamate multiple methods and proposes an integrated data collection framework with adaptations to the conventional Accident Report Form, widely used for crash data collection and analysis. This is the major methodological contribution of the current research. The framework is also validated and tested for reliability. This integrated framework can streamline the identification of additional risk factors associated with pedestrian safety.
Moreover, the study signifies policy implications for elevating safety through targeted recommendations and interventions with the 6-Es- Engineering, Enforcement, Education, Economics, Emergency Response, and Enablement at various levels of the transportation system in developing countries such as Bangladesh.