From the Editors


  • Tony Cox,

  • Karen Lowrie


A recent article by Charles Vlek in the June issue of Risk Analysis critically appraised how well national risk assessments (NRAs) succeed in helping governments to allocate limited resources fairly and effectively to reduce risks that could harm their citizens. The article recommended several possible improvements in NRAs. This issue begins with a commentary by Pruyt et al. suggesting that Vlek's analysis mistakes some of the intent and emphasis of several components of the Dutch and British NRA process, including its use of multi-criteria analysis and risk diagrams and its tolerance of false-positive and false-negative risk scenarios. In a response, Vlek welcomes Pruyt et al.'s comments as a useful contribution to a needed debate, but maintains that false positives and false negatives do and should matter in making risk-informed resource allocation decisions; that decision theory and risk psychology highlight the pitfalls (e.g., interest bias, illusion of control, groupthink) of using expert risk evaluation, and the challenges of making useful comparisons across multidimensional risk scenarios; and that firmer grounding in risk analysis, decision theory, and safety psychology could improve the value of NRAs. The Editors welcome readers to contribute further insights and new research about how to make NRAs more valid as well as communicable.


Before undertaking a full quantitative risk assessment, it may be possible to use relatively quick and inexpensive screening questions to identify which possible hazards are most likely to be worth investigating or managing further. Four papers in this issue consider the development and applications of a Freshwater Fish Invasiveness Screening Kit (FISK). Gordon Copp, in his introduction to this special series of papers, explains that “the generally-accepted premise that weeds in other parts of the world have an increased chance of being weedy (i.e. invasive) in other areas with similar environmental conditions… provided a conceptual, semi-quantitative basis from which to develop similar screening tools for a range of aquatic species.” This insight led to various versions of FISK in the U.S., Europe, and Australia among other locations. The papers in this special series discuss experiences applying FISK in different countries and climatic zones. They provide new evidence that such screening tools can be valuable in identifying aquatic species with the greatest potential to be invasive in different areas.


The practical applicability of quantitative health risk assessment methods is advanced by developing dose-response models and uncertainty characterizations for an increasingly wide range of hazardous agents. Leleu et al. present a Bayesian analysis of the dose-response relation observed experimentally in mice for the prevalent airborne fungal pathogen Aspergillus fumigates. This paper should be of interest not only to practitioners specifically interested in dose-response relations for fungi, but also to risk analysts interested in uncertainty analysis. It illustrates a novel application of directed acyclic graph (DAG) modeling, together with exponential and beta-Poisson sub-models, to characterize uncertainty and inter-individual variability in exposure and dose-response models. Najita and Catalano advance dose-response modeling methodology by extending two-stage sampling for benchmark dose (BMD) estimation to developmental toxicity studies with multiple sources of prenatal exposure and multiple possible outcomes (e.g., both discrete and continuous indicators of malformation or other harm). They consider how to adaptively select subsets of malformed and healthy pups to increase precision of the BMD estimate in this highly multivariate context, treating unobserved values of predictors (due to limited sample sizes and sample designs) as missing data. Under the conditions that they studied via simulation, oversampling typically outperformed simple random sampling or under-sampling for small subset sizes, especially when malformations were relatively rare.


Three papers apply systems modeling to identify effective risk management interventions for reducing risks in complex systems. Delgado et al. critically review current import risk analysis (IRA) methods for exotic animal diseases, and propose an alternative, network-based approach for systematically identifying how different interventions at critical control points (CCPs) could reduce the likelihood of disease incursion. They illustrate the network modeling approach with an application to reducing the likelihood of U.K. pig herds being exposed to classical swine fever (CSF), using a combination of expert elicitation and sensitivity analyses to obtain needed inputs and to systematically identify both previously known and previously unidentified threats that must be managed to reduce CSF risks. Jones and Adida formulate a simple stochastic model of population mixing and influenza transmission, together with an explicit objective function (considering costs of interventions and of illnesses), to optimize selection of hygiene and social distancing interventions to reduce and mitigate risks of influenza pandemics. They find that the optimal mix of non-pharmaceutical interventions (NPIs) for influenza control is sensitive to compliance behavior as well as to probability and duration of infection – areas where additional research may have especially high value of information for improving model-based risk management decision recommendations. Wang et al. consider the physics of crowd flows and congestion, and apply queuing theory and simulation to gain insight into the effective design of evacuation paths (e.g., number of exit passageways) for safely evacuating crowds of different sizes and densities from stadiums or other crowded areas.


How do people's views on contentious environmental issues, such as cormorant management in the Great Lakes Basin, influence each other? Muter et al., study the interaction among expert and lay risk perceptions within the framework of the social network contagion theory of risk perception (SNCTRP). They find that contagion effects, in which perceptions spread among people via contacts with each other, are strongest when experts and lay people interact. Herrmann studies two related questions: How well can people estimate information technology (IT) risks (a challenging assessment task); and how do people react to each other's estimates? Among the key findings are that even older and more experienced estimators do not significantly out-perform younger and less-experienced estimators. Neither estimates risks very well. However, older and more experienced estimators are better able to use the estimates of others in updating their own initial estimates. Meyer et al. use a web-based simulation environment to study in detail how people gather and use information from various media and from neighbors about an oncoming hurricane, and how different types, sources, and formats of risk information affect their intentions to take protective actions. This investigation yields a number of striking insights, such as that showing the most likely path of a storm, instead of only an uncertainty cone, motivates higher levels of preparation, even among participants living far from the predicted center path.