Society For Risk Analysis Annual Meeting 2017

Session Schedule & Abstracts


* Disclaimer: All presentations represent the views of the authors, and not the organizations that support their research. Please apply the standard disclaimer that any opinions, findings, and conclusions or recommendations in abstracts, posters, and presentations at the meeting are those of the authors and do not necessarily reflect the views of any other organization or agency. Meeting attendees and authors should be aware that this disclaimer is intended to apply to all abstracts contained in this document. Authors who wish to emphasize this disclaimer should do so in their presentation or poster. In an effort to make the abstracts as concise as possible and easy for meeting participants to read, the abstracts have been formatted such that they exclude references to papers, affiliations, and/or funding sources. Authors who wish to provide attendees with this information should do so in their presentation or poster.

Common abbreviations

M3-E
Symposium: Game Theory, Decision Analysis for Homeland Security and Disaster Management

Room: Salon E   1:30 pm–3:00 pm

Chair(s): Bairong Wang   bairongw@buffalo.edu

Sponsored by Security and Defense and Decision Analysis and Risk Specialty Groups

This proposed symposium will present some latest update on game theory and decision analysis applications to homeland security and defense.



M3-E.1  1:30 pm  A Signal Detection Model and Analysis of Risk-Based Threat Assessment. John RS*; University of Southern California   richardj@usc.edu

Abstract: Increasingly Big Data analytics are being utilized to assess the terrorism threat of individuals. Such “risk based” approaches are designed to bin individuals in order to apply differential levels of scrutiny based on individual characteristics related to terrorism threat. Decades of previous research on violence risk assessment has demonstrated the limitations in how accurately future violent behavior can be predicted from individual characteristics and past behavior. In many respects, the terrorism case is a greater challenge due to the presence of low base rates, the lack of highly diagnostic indicators, and the ability of an adaptive adversary to learn how to attenuate their predicted threat level. We formulate this generic problem in a Signal Detection framework, and conduct sensitivity analysis to determine conditions in which such threat assessments are useful. Results depend on the prior probability that a randomly selected individual is a terrorist, the likelihood function associated with the predicted threat level, and the relative cost of false-positives and false negatives. Further sensitivity analyses explores the impact of attenuated diagnosticity resulting from an adaptive adversary capable of learning the characteristics required for a “low-risk profile” and either acquiring a low-risk profile, or recruiting an accomplice with a low-risk profile. Risk-based approaches to threat assessment are not effective for assessing terrorism threat due to low base-rates and adaptive adversaries capable to learning the predictive algorithm and appearing as low-risk individuals.

M3-E.2  1:50 pm  The Hurricane Decision Simulator and Its Impact on Decision Making. MacKenzie CA*, Regnier E, Hetherington S, Prisacari A; Iowa State University   camacken@iastate.edu

Abstract: The Hurricane Decision Simulator is a web-based simulation program to help personnel at the U.S. Marine Reserve Forces in New Orleans gain experience with hurricane-preparation decisions given forecast uncertainty. It provides simulated hurricanes with probabilities of whether a hurricane will strike New Orleans. The user is provided with forecasted probabilities of hurricane-force winds and decides whether or not to take preparatory actions. The simulation ends when a hurricane either hits or misses New Orleans. A study was conducted to assess how training on the simulation impacts people's decisions. The study suggests that people training on the simulation are more likely to wait to evacuate.

M3-E.3  2:10 pm  Estimating effectiveness of investment, optimal resource allocation, and predictive risk analytics for fire protection. Madasseri Payyappalli V*, Behrendt A, Zhuang J; University at Buffalo, The State University of New York   vineetma@buffalo.edu

Abstract: Fire-related hazards and incidents are an everyday phenomenon, and the estimated total cost of fire was $329 billion in 2011. We leverage the large amount of data available from sources such as National Fire Protection Association (NFPA) and National Fire Incident Response System (NFIRS), to create data-driven empirical and theoretical models for estimating the effectiveness of investment, for formulating optimal resource allocation strategies, and for efficiently assessing and mitigating risk in the context of fire protection. Our results show that fire losses have decreased exponentially in investment with high R2 values (~0.8), and also show potential under- and overspending in fire protection. We provide analytical closed-form solutions for effectiveness of investment as well as initial vulnerability, and derive insights. The results from the optimal resource allocation problem emphasize the need for considering the trade-off between equity and efficiency in resource allocation at state, county, and fire-district levels. In addition, a case study of federal fire grant allocation is used to validate and show the utility of the optimal resource allocation model. We use advanced data analytics and machine learning techniques, to work with data on fire incidents, fire department resources, socio-economic factors, and weather conditions, for dynamic fire risk assessment and resource scheduling. The core component of the analytics model is a generalized hazard function represented as a function of threat, vulnerability, and consequence. Finally, this work describes how the optimal resource allocation can be merged with advanced data analytics model to create a decision support system for real-time fire risk visualization as well as dynamic decision. The research will be of use to policymakers and analysts in fire protection and safety, and will ultimately help in mitigating economic costs and saving civilian and firefighter lives.

M3-E.4  2:30 pm  Rumor Response, Debunking Response, and Decision Makings of Misinformed Twitter Users during Disasters. Wang B*, Zhuang J; University at Buffalo, SUNY   bairongw@buffalo.edu

Abstract: The rapid spread of rumors occurring on social media is a critical problem that poses a great risk to emergency situation navigation, especially during disasters. Many research questions, such as how misinformed users judge potential rumors or how they respond to them are crucial issues for crisis communication, but have not been extensively studied. This paper fills this gap by originally documenting and studying Twitter users' rumor and debunking response behaviors during disasters, such as during Hurricane Sandy in 2012 and the Boston Marathon bombings in 2013. To this end, two rumors from each disaster and their related tweets are documented for analysis. Users who were misinformed and involved in the rumor topic by posting tweet(s), respond to a rumor by: (1) spreading (≥ 86%); (2) confirming (≤ 9%); and (3) doubting (≤ 10%). For confirmation-seeking (confirming) tweets, results show that not all of these tweets will get accurate replies. There were some confirming tweets from three cases were replied with rumor descendant tweets. Fortunately, more than 50% of these confirming tweets with rumor replies were also replied with accurate rumor debunking information at the same time. Confirming tweets are replied significantly more frequently than those non-confirming tweets from three rumor cases (P≤ 0.1). Rumor debunking analysis results indicate that if the given rumor spreading users were debunked, they would respond by: (1) deleting rumor tweet(s) (≤ 10%); (2) clarifying rumor information with a new tweet (≤ 19%); or (3) neither deleting nor clarifying (≥ 78%). We conclude that Twitter users would perform poorly in rumor detection and rush to spread rumors. The majority of users who spread rumors would not take further action on their Twitter accounts to fix their rumor spreading behaviors. Twitter users who post new tweets for rumor confirmation will face the risk of being rumored. But fortunately, more than half of them will eventually be debunked with accurate replies.



[back to schedule]