From the Editors
From the Editors
How Can Risk Analyses Be Made More Useful?
Risk assessments have differed widely in their influence and practical value for improving risk management decisions. Some seem to provide little more than a patina of arithmetic and rational-looking justification for decisions that have already been made. Others, such as the WASH-1400 Reactor Safety Study, revolutionize how important risks are understood, quantified, and discussed. They provide conceptual frameworks and repositories for data and knowledge that continue to inform risk assessments and risk management decisions long after the original study is over.
Goble and Bier propose that risk analysts can deliberately cultivate such high-impact analyses. They propose creating and using “living” risk assessments to dramatically increase the value, reuse, and expansion and updating of risk assessment information. Such living risk assessments would enable the data, knowledge, analysis, and thinking embodied in a risk assessment to serve as a foundation for repeated updates and new applications and decision support as needed, multiplying the return on initial investments in risk assessments. The editors of Risk Analysis: An International Journal believe that this proposal has great potential for increasing the practical value of our field and its work products. We welcome further perspectives and suggestions for how best to implement living risk assessments, and for how to use journals — including their facilities for maintaining and indexing text, online supplements, supporting data and meta-data, and citation links — to help make such living risk assessments as valuable as possible. Making major risk assessments and their supporting data, models, and calculations fully available to readers could become a valuable new role for Risk Analysis and for authors who publish results of such assessments, helping to increase their reproducibility, credibility, and reuse.
Making Transportation Safer
Two articles in this issue consider how to use risk modeling to make transportation safer. Greenberg et al. grapple with how to make risk modeling of passenger rail security and safety more genuinely useful to users and sponsors with sharply limited time and resources. They develop three simulation models — of rail passenger traffic flows between stations in New York and New Jersey; dispersion of a plume of toxic gas from a slow-moving freight train; and regional economic impacts of rail-related disruptions — to help assess the regional economic consequences of a chlorine release and a bridge failure. Simulating such events, described as medium to low probability events with medium to high consequences, engages the practical interest of busy planners and senior managers and provides a way to focus attention and creative discussions on understanding and addressing potential problems that might otherwise be difficult to discuss constructively. Siddiqui and Verma consider how best to route intercontinental oil tankers to minimize estimated expected harm from oil spills, using a novel methodology to estimate accident probabilities per unit distance on different links from coarse global data. They find that the shortest routes do not necessarily minimize expected damage, and present a constructive procedure for finding lower-risk routes.
Risk Perception and Communication
Perko et al. study how effectively people receive (i.e., pay attention to, understand, and remember) precrisis information about emergency preparedness, using data from a large-scale Belgian opinion survey that dealt with reception of information about distribution of iodine tablets to people living near nuclear facilities. They report that people with little confidence in authorities pay less attention to such information, and that people who are knowledgable about radiation risks tend to receive the precrisis information more effectively than those without such specialized knowledge. Perhaps more surprisingly, those who perceive radiation risks as being relatively high also tend to be less attentive to information about protective actions. The authors hypothesize that relatively cognitive (System 2) systematic predictors, such as education and specific knowledge, most affect reception of precrisis information; while heuristic (System 1) factors such as fear and confidence in authorities may more affect acceptance or rejection of the received information.
Tseng et al. report results of a telephone survey in Taiwan examining self-reported psychopathologies and perceptions of health risks from electromagnetic fields (EMFs) and other environmental sources. More than half of all respondents believed that power lines and mobile phone base stations had a large effect on human health (almost certainly a mistaken belief), and more than 10% believed that they personally suffered adverse health effects from EMF. People with psychiatric conditions such as anxiety, depression, distress, and somatization are more likely to self-report sensitivity to EMFs and to be concerned about, and attribute health complaints to, mobile phone base. Women, married people, and people with higher levels of education are more likely to perceive environmental risks (such as EMF risks) as being high. The study shows that having a catastrophic illness, psychopathology, and self-reported sensitivity to EMF are also significant predictors of perceived risks from both EMF and non-EMF environmental sources in Taiwan.
Risk and Behavior
Past research on risk-taking by criminal offenders has shown that prisoners are deterred more by the likelihood of a penalty than by its severity, while non-offenders are deterred more by severity than by likelihood. Rolison et al. apply cumulative prospect theory (CPT) to compare the willingness of prisoners and ex-prisoners to accept financial risks with varying probabilities and magnitudes of gains and losses. They find that recently released ex-prisoners (within 4 months following their release) are more likely than those still incarcerated to take risks that involve potential losses as well as potential gains, in part because of reduced sensitivity to the magnitude, rather than to the likelihood, of losses. For example, 42% of ex-prisoners, compared to only 18% of prisoners, accept a 50:50 chance of gaining or losing 100 pounds sterling. In the United States and Britain, more than half of ex-prisoners are reconvicted within two years of release, due to commission of new offenses that involved risks of loss of freedom as well as chances for gains, such as satisfying current impulses. This research emphasizes the importance of studying risk-taking behaviors by offenders both outside and inside of the prison system. It adds new weight to the possibility that effective deterrence depends more on high likelihood of being caught than on severe consequences (e.g., longer sentences) if caught.
Yamane et al. study losses in property values caused by the Fukushima–Daiichi accident. They find that property values depreciated quickly and monotonically with higher levels of radioactive contamination, although the marginal economic value of proximity to the plant did not change in any obvious way.
Risk Matrices Revisited
Risk matrices are among the most widely used tools in enterprise risk management and many other applications, yet they have important theoretical and practical limitations. Reniers and Sorensen consider how to build on the types of historical frequency and severity data that are sometimes displayed in risk matrices, and that are available for most common work-related accidents (e.g., statistics on occupational slips and falls, small fires, etc.). Their goal is to compare groups or bundles of risks, and to select among alternative risk-reducing opportunities while taking into account not only the data in a risk matrix, but also information on costs and benefits of different combinations of control efforts, budget constraints, and precedence constraints and other interdependencies among risk management activities (such as installation and training for a new safety device). They express such interdependencies as constraints in a knapsack problem (an operations research optimization model for selecting a subset of costly risk-reducing activities to maximize the risk reduction achieved for a given budget). Considering the costs and benefits of risk-reducing activities, as well as interdependencies among them, can add substantial value to the data and insights obtained from risk matrices alone.
Ball and Watt emphasize the practical limitations of trying to use risk matrices to organize risk information and make reliable recommendations without such additional modeling and analysis. Differences among raters in interpreting what qualitative terms such as “high,” “medium,” or “low” might mean, or in focusing on some potential consequences more than others, contribute to ambiguity and unreliability in the qualitative ratings used in populating many risk matrices. Ball and Watt suggest that the apparent simplicity of risk matrices is deceptive and potentially dangerous, as the attempt to reduce multidimensional issues to two-dimensional ones may discourage deeper and more productive thinking about risk management opportunities and trade-offs. They conclude that, in the specific context of leisure activities (and possibly in other contexts where risk assessment is conducted in a public setting), risk matrices can be highly unreliable, ambiguous, and misleading, potentially leading to unjustified recommendations that impose real opportunity costs on those who follow them.
Tony Cox and Karen Lowrie
Note from Rick Reiss, Risk Analysis Editorial Board member and former Managing Editor:
This issue features two articles from Asian authors on risks of bovine spongiform encephalopathy (BSE) associated with beef consumption. The articles arose out of a meeting in Taiwan during 2010 where I attended the annual meeting of the Taiwan chapter of SRA while I was President of SRA International. I was struck by the presentations on BSE and the wide concern about BSE risks among the public in Asian countries with U.S. imported beef in particular. I thought this topic would make a good contribution to Risk Analysis and would be of interest to people in the United States, Asia and elsewhere. Jun Sekizawa from Japan provides a perspective on risk perception of BSE between East Asian countries and Western countries, and concludes that socioeconomic backgrounds significantly influence public understanding of the risks. While Japan has an institutional framework for food safety, risk communication is still immature. In Korea and Taiwan, political factors and Internet information appear to play a larger role. Chu-Chih Chen and co-authors present a mathematical model based on extreme value statistics for variant Creutzfeldt-Jakob disease (vCJD) and develop an exposure threshold hypothesis. They conclude that this hypothesis can be used to explain the observed vCJD cases in the United Kingdom.