Basing their approach on Homans' group theory, the authors evaluated 168 examples of doctor–patient conflict that occurred between 2000 and 2016. The empirical results show that group relationship, group scale and event duration act as catalysts, while media attention can be a conditional buffer in influencing the evolution of such conflict behaviour. The authors also found that group scale has a buffering effect when the event duration is longer, while the buffering effect of media attention is weaker when the event duration is shorter. The authors discuss how these findings contribute to research on social risk management.