How Do Witness Deaths Impact Learning from Accidents? A Case Study of 58 Investigation Reports from Fatal Accidents in China
26 Pages Posted: 21 May 2018
Date Written: July 7, 2016
Using 58 investigation reports of fatal accidents from the State Administration of Work Safety of China (SAWSC), we studied how witness deaths affected several key variables using a Spearman correlation analysis and an ordinal regression analysis with three functions. We found that survivor and behavior data (gained from recording systems) significantly affected the information gathered about accidents. The behavior data and the course of the accidents also influenced whether the causes were successfully determined. Then we propose a “death paradox” to show that the people closer to an accident’s core area are more likely to obtain key information, but are also more likely to be killed. We concluded that: (1) Surviving witnesses provide intuitive information about an accident, but are unsure of the cause(s); (2) Comprehensive techniques for recording information are vital to learning from an accident; and (3) The characteristics of a particular industry may influence the importance of a witness in an investigation.
Keywords: Accident analysis; accident investigation; death paradox; Internet of things (IoT) for safety; ordinal regression
Suggested Citation: Suggested Citation