Abstract
INTRODUCTION: Mass gathering activities are often accompanied by safety risks especially overcrowding, since the characteristics of risk factors may differ under various organizational patterns. However, this issue has not been sufficiently studied, which could lead to cognitive biases in understanding the risks associated with various mass gathering activities, thereby affecting the effectiveness of preventive measures. METHODS: This study investigates the risk factors of human stampede in mass gathering activities, with a particular focus on analyzing the influence of different organizational patterns on these risks and their implications for safety management. By combining Grounded Theory and Iterative Self-Organizing Clustering Algorithm (ISODATA), 209 overcrowding cases were coded to create a dataset containing risk factors. RESULT: The organization coefficient was proposed to characterize the degree of risk in different clusters, and the clustering results revealed three types of organizational patterns including organized( O31 =1.9174), applied( O82 =2.9831), and spontaneous( O73 =4.4327). DISCUSSION: The results indicate significant differences between the three organizational patterns in terms of aggregation causes, layout characteristics, and risk levels, which directly impacts the safety of mass gathering activities. Furthermore, similarities are observed in terms of triggering behaviors, knowledge, awareness, and management across the four categories. This study provides theoretical evidence for risk prevention and safety management of human stampede in mass gathering.