Modeling Root Causes of Construction Site Fatalities Using Social Network Analysis

In spite of the continuing decrease in accident rates in the US Construction industry, accidents are one of the key factors constraining construction industry in terms of cost and time. Fatality rates in the US are more than other developed countries. Research into accident causation included accident occurrence mechanisms, different levels and types of accident causes as and best practices for accident prevention. These efforts, however, did not focus particularly on fatal accidents and did not prioritize the relationships amongst root causes for accidents. Accordingly, this research’s goal is to quantitatively analyze relationships between fatal accident root causes as well how they are tied to direct causes commonly quoted in fatality investigations using social network analysis (SNA). Social network analysis is a set of tools derived from mathematical graph theory and utilized in several knowledge areas including social sciences, natural sciences, engineering, construction management and safety. Accordingly, a three-step methodology is devised. First, 100 case files are analyzed for accident causation data. Second, an SNA model is built utilizing the relationships between root causes and broken down according to direct causes. Third, model is analyzed, and results are interpreted and validated to provide insights into fatal accident causation and contributing underlying factors. The analysis of all social networks yield that accident root causes interact very closely, and that only a few causes contribute most to fatality networks, a relationship which can be described by a power law. The model is capable of successfully identifying the most influential root causes for fatal accidents and their relationships. It identified “lack of job specific training” as the key cause for fatal accidents particularly for the direct causes “struck by” and “caught in between”. Moreover, SNA showed that this cause is most influential when combined with “absence of fall arrest system” and “lack of personal protection equipment”. Benefits of this research include providing a different approach to causation of fatal accidents that provides deeper more holistic insights into their root causes. Additionally, it provides a quantitative tool for prioritization of root causes which can guide implementation of Safety Management policies and practices.