Developing Implicit Uncertainty Visualization Methods Motivated by Theories in Decision Science

Abstract

Agreement between public policy decision makers and geographic information systems and visualization researchers about the importance of uncertainty in decision support sits in contrast to a disconnect in approaches to incorporating uncertainty into decision support tools. This disconnect does not arise from how these two groups define uncertainty but instead occurs because they approach uncertainty from different problem perspectives (Miller et al. 2008; Pohl 2011). Public policy decision makers regularly contend with uncertainty based on how proposed policies will affect the future, resulting in a solutions-oriented approach that relates uncertainty of future conditions to policy outcomes. For researchers, uncertainty more often reflects unknowns in data values or modeling processes, such as the difference between a measured or predicted value and the actual value, resulting in a knowledge-production approach that relates uncertainty to the validity and legitimacy of methods, models, and data to produce knowledge. The research presented here contends that this gap between research and practice (Brown and Vari 1992; von Winterfeldt 2013) stems from these differing perspectives. To bridge this gap, we examine decision science theories to explain decision makers’ solutions-oriented approach to uncertainty. Decision science is concerned with understanding and improving how individuals or groups identify problems, make decisions, and learn from the outcomes. We then present a new methodology, implicit uncertainty visualization, that reflects how decision makers contend with uncertainty. Bridging this gap opens up opportunities to develop visualization methods and tools that help decision makers better deal with uncertainty in practice.