Early Warning

Information Seeking and Use in Early Warning

Social Sciences and Humanities Research Council of Canada (SSHRC) 

Standard Research Grant (2009 - 2012)

Dr. Chun Wei Choo, University of Toronto (PI) [cw.choo@utoronto.ca]

Papers

Project Introduction

With growing populations and expanding infrastructures, our exposure to natural and man-made hazards has increased dramatically. Although disasters often appear to happen suddenly with little or no prior warning, in fact most disasters incubate over long gestation periods during which warning events accumulate. While these warning signals become painfully clear in hindsight, why is it so hard to detect, recognize, and act on these precursor conditions before they tailspin into tragedy? This study explores two possible reasons. First, in early warning it is difficult to discern signals about a real threat from random noise. Second, even when signals are detected, policymakers have to weigh the costs and consequences of possible false alarms and misses in deciding whether to activate a warning.

The objective of this research is to examine how the use of information affects the effectiveness of early warning systems. By effectiveness, we refer to the capacity of the system to detect and decide on the existence of a threat or hazard. There are two aspects to effectiveness: being able to see the evidence that is indicative of a threat; and making the decision, based on the weight of the evidence, to warn that the threat exists. In early warning information use is encumbered by cues and messages that are fallible, dispersed, and equivocal. Cues that are true indicators of a threat are invariably obscured in a cloud of events generated by chance. Apart from being able to see the threat, policymakers also face the difficult decision of whether to issue a warning based on the information received. Because the information is rarely complete or conclusive, such decisions have to consider the consequences of failing to warn, or giving a false warning.

There is a lack of research in early warning that focuses on the use of information in threat detection. This study draws on socio-cognitive theories of perception and judgment to analyze the two aspects of early warning mentioned earlier: detection accuracy (how well does perception correspond to reality?); and decision sensitivity (how much evidence is needed to activate warning?). Using Cognitive Continuum Theory, we explore how detection accuracy depends on the fit between the information needs profile of the threat and the information use environment of the warning system. Applying Signal Detection Theory, we investigate how decision sensitivity depends on the perception and balancing of the risks of misses and false alarms that is inherent in all early warning decision-making.

We propose a mixed methods multi-case study of 3-5 early warning systems, such as the Global Public Health Intelligence Network (GPHIN, Public Health Agency of Canada); Risk Assessment and Horizon Scanning System (RAHS, Singapore Government); and Global Information and Early Warning System (GIEWS, Food and Agriculture Organization). These systems target different types of threats and provide the variation in information and detection environments needed by the study. RAHS is notable for its use of innovative methods and techniques, while GIEWS is one of the older, more established systems. 

We believe that this study is one of a few to link information seeking and use to early warning effectiveness, using concepts and instruments derived from established models of social judgment and perception. We hope that a deeper understanding of information use in threat detection will enable the design of systems that can provide effective warning and enable the timely implementation of preventive measures.