Understanding and Predicting Human Behaviour in Maritime Emergencies

Maritime disasters can result in devastating loss of life. In order to mitigate such loss, it is necessary to optimise the efficacy of evacuation procedures. Previous research on maritime disasters and the behaviour observed during evacuation is limited. This is mainly due to issues involving the collection of sufficient reliable data. Human behaviour is a complex phenomenon variously affected by experience, emotions, interactions and environment. These factors are further modulated by the life-threatening nature of uniquely evolving disasters. Currently, computational evacuation models fail to consider passengers as sentient, psychological agents. Some have shown success in the prediction of actions undertaken.

This thesis describes three interlinked studies intended to crystallise a methodological approach for effective future research into disaster evacuations. The first study was a replication of a previous study of the Costa Concordia disaster. Behavioural Sequence Analysis was conducted on data collected from a new sample of survivors. Statistically significant correlations were found which indicated the reliability of the method of data collection. These data were then broken down by gender, age, companions, and experience in order to detect intra-cohort differences. Analysis of routes and transitions in decomposition diagrams detected differences between sub-cohorts. However, existing psychological literature was unable to offer consistent, persuasive explanations for certain detected phenomena.

The second study involved the recreation of the first study using the ‘Talk-Through’ method. This method is potentially valuable for the creation of reliable imagined data, which mirrors real-life experiences. The third study involved an in-depth comparison of the data extracted by each method. This filled a specifically noted gap in knowledge concerning the ecological validity of the ‘Talk-Through’ method. It was anticipated that the results of the two methods would be comparable. Indeed, correlation analysis provided evidence for the validity of the talk-through method with respect to the number of acts and transitions reported in each condition.

In order to accelerate evacuation research, refinements to data-collection and analysis were proposed. By taking control of the storytelling, more controlled and comparable data may be produced which focus on choices made at each phase during a complete sequence of evacuation. This would create data more appropriate for computer modelling and more capable of quantitatively evaluating choices. Understanding motivations is critical to the communication of effective and authoritative instruction. By redirecting psychological research towards this goal, persuasive evidence may be produced to guide and inform the implementation and execution of emergency evacuation procedures.