Communicating risk in disaster risk management systems – experimental evidence of the perceived usefulness of risk descriptions

Abstract

Disaster risk management (DRM) requires the collaboration of a variety of stakeholders working in different sectors. They depend on each other to share risk information, and effective collaboration requires efficient communication. Clearly, the communication of risk descriptions is a key issue for the success of DRM activities. This study investigated the communication of risk-related information between DRM actors, and how to present such information in order to improve its usefulness as a basis for decision-making. Two experiments were conducted to investigate the issue. The first related to the type of hazard, while the second examined the presentation of likelihood and consequences. We addressed two research questions: Does the presentation of a risk description influence its perceived usefulness for decision-making in a DRM system? If so, which format is perceived as most useful? Our findings indicated that the type of hazard did not affect perceived usefulness to any significant extent. However, the presentation of likelihood and consequences did have a significant effect. These results indicate that the presentation of risk information greatly influences stakeholders’ ability to make well-informed decisions. Specifically, quantitative scale and semi-quantitative ranking scale formats are perceived as most useful.

Keywords:

• disaster risk management (DRM)
• risk communication
• risk description
• perceived usefulness
• decision-making

Acknowledgements

The authors wish to thank the Swedish Civil Contingencies Agency (MSB) (project PRIVAD, No. 2010-2872) and the Swedish International Development Cooperation Agency (SIDA), whose funds are coordinated by the International Science Programme (ISP-Uppsala University) within the program ‘Research Capacity Building in Nature-Induced Disaster Mitigation in Central America’, for supporting the research on which this paper is based.

Notes

1. The power efficiency of the Wilcoxon signed-rank test is 95% of that of the t-test for small sample sizes (Siegel and Castellan 1988).

2. The minimum relevant difference between the two experimental conditions was set to one step on the Likert scale. The standard deviation of the difference between perceived usefulness was unknown; however, the standard deviation in a similar study was 1.5. We therefore assumed it to be the same in this study. Finally, α was assumed to be 0.05 and β, 0.2.

3. Delaney and Vargha (2002) suggest that an A₁₂ of .56 corresponds to a small effect size, .64 to a medium effect and .71 to a large effect.