Evaluating the impact of visualization of wildfire hazard upon decision-making under uncertainty

ABSTRACT

The paper investigates whether the methods chosen for representing uncertain geographic information aid or impair decision-making in the context of wildfire hazard. Through a series of three human subject experiments, utilizing 180 subjects and employing increasingly difficult tasks, this research evaluates the effect of five different visualizations and a text-based representation on decision-making under uncertainty. Our quantitative experiments focus specifically on the task of decision-making under uncertainty, rather than the task of reading levels of uncertainty from the map. To guard against the potential for generosity and risk seeking in decision-making under uncertainty, the experimental design uses performance-based incentives. The experiments showed that the choice of representation makes little difference to performance in cases where subjects are allowed the time and focus to consider their decisions. However, with the increasing difficulty of time pressure, subjects performed best using a spectral color hue-based representation, rather than more carefully designed cartographic representations. Text-based and simplified boundary encodings were among the worst performers. The results have implications for the performance of decision-making under uncertainty using static maps, especially in the stressful environments surrounding an emergency.

KEYWORDS:

• Uncertainty
• visualization
• decision-making
• wildfire hazard
• evaluation

Acknowledgements

This research project has been approved by the University of Melbourne Faculty Human Ethics Advisory Committee, number 1238459.2. We gratefully acknowledge Amy Corman for her invaluable assistance with the human subject experiments, Alexander Klippel for fruitful discussions and generous advice on the original experimental design, Christoph Kinkeldey for his valuable feedback on earlier versions of the paper, and Ingrid Burfurd for the original idea of using performance-based incentives in the evaluation of the different visualizations. Finally, we are grateful for the thoughtful and constructive comments of the anonymous reviewers.

Additional information

Funding

We gratefully acknowledge the financial support of the Australian Research Council (ARC) including the Discovery Projects Scheme [project DP120100072] ‘From environmental monitoring to management: Extracting knowledge about environmental events from sensor data’ and ARC Discovery Early Career Research Award [DE140101014] (Wilkening). This research is supported by funding from the Bushfire Cooperative Research Centre.