Risk Analysis, Va!. 13, No.6, 1993 Incorporating Structural Models into Research on the Social Amplification of Risk: Implications for Theory Construction and Decision Making William J. Burns,t,2 Paul Slovic,3 Roger E. Kasperson,4 Jeanne X. Kasperson,4,5 Ortwin Renn, 6 and Srinivas Emani4 Received August 24, 1992; revised August 9, 1993 A comprehensive approach to managing risk must draw on both the descriptive insights of the behavioral sciences and the prescriptive clarity of the management sciences. On the descriptive side, this study develops structural models to explain how the impact upon society of an accident or other unfortunate event is influenced by the physical consequences of the event, perceived risk, media coverage, and public response. Our findings indicate that the media and public response play crucial roles in determining the impact of an unfortunate event. Public response appears to be deter- mined by perceptions that the event was caused by managerial incompetence and is a signal of future risk. On the prescriptive side, we briefly discuss how these findings based upon structural models can be incorporated into a decision-analytic procedure known as an influence diagram. Risk perception; social amplification; impact analysis; structural models; influence diagram. 1. INTRODUCTION Risk research in the social sciences has developed along two important but distinct paths. Behavioral sci- entists have sought to describe how people and institu- tions perceive and respond to risk, while management scientists have attempted to develop methods that pre- scribe appropriate actions for managers to take. Despite significant progress in these separate areas, there is still no comprehensive approach that integrates our under- standing of how a society experiences risk with formal methods of decision-making (e.g., cost-benefit analysis and decision analysis). Managers continue, for example, to be surprised and dismayed when "minor events," as 1 Department of Marketing, PHBA, University of Iowa, Iowa City, Iowa 52242. 2 To whom all correspondence should be addressed. 3 Decision Research, 1201 Oak St., Eugene, Oregon 9740l. 4 Clark University, Center for Technology, Environment, and Devel- opment (CENTED), 950 Main Street, Worcester, Massachusetts 01610. 5 Brown University, Alan Shawn Feinstein World Hunger Program. 6 Center of Technology Assessment, Nobelstrasse 15, D- 70, Stuttgart, Germany. 611 assessed by experts, spark great alarm and subsequent societal disruption. A comprehensive approach to managing risk must draw upon the descriptive insights of the behavioral sci- ences and the prescriptive strengths of the management sciences. Formal analytic methods, for example, typi- cally have assessed the impact of accidents or other ad- verse events in terms of expected loss of life or property damage. The contribution of these methods has been to provide risk managers with explicit guidance, using cri- teria that are well defined. A nice illustration of these methods can be seen in Keeney, (1) who develops a utility model for evaluating potential fatalities and associated uncertainties of occurrence. The effect of fatalities on the public is evaluated in terms of the direct personal impacts of suffering and economic hardship and the in- direct societal impacts of possible political and economic turmoil. Keeney's disaggregation of impacts into per- sonal and societal is insightful, and his approach leads to clear recommendations for action. However, focusing on fatalities or related criteria may not adequately antici- pate public concerns as perceptions of threat and social response appear less a matter of physical outcomes than of attitudes, social influences, and cultural identity. (2) 0272·4332/93/1200·0611$07.00/1 © 1993 Society for Risk Analysis