Integrated Environmental Assessment and Management — Volume 15, Number 1—pp. 165–169 165 Book Reviews LET THE EVIDENCE SPEAK: USING BAYESIAN THINKING IN LAW, MEDICINE, ECOLOGY AND OTHER AREAS by Alan Jessop Let the Evidence Speak by Alan Jessop provides a clear and enjoyable discussion of how evidence and Bayesian reasoning go hand in hand. This book makes Bayes accessible by explaining why it is useful to know what Bayes’ Rule is, how it can be used to solve problems, both simple and complex, and how to communicate Bayesian reasoning to a broad audience. Bayes’ Rule is not just an equation for statisticians or data miners; it is “a way of thinking,” as Jessop states, that helps determine how the available evidence can revise previous beliefs in a hypothesis. This book explains how Bayesian thinking is useful for many different types of problems, from the everyday to the uncommon. Overall, the case studies and discussions provide a useful introduction on the advantages of Bayesian thinking and Bayes’ Rule when weighing evidence. The book is well written and well organized, making it an enjoyable read for both novices and advanced Bayesian practitioners. Jessop gently introduces the components of Bayes’ theorem in a manner that makes the reader forget she is learning about a mathematical formula. Rather than jumping into Bayes’ theorem as an overarching paradigm, Jessop initially focuses on 1 part of the equation that binds Bayesian and classical analysis—estimating and interpreting likelihoods. The chosen examples and descriptions in this and other sections are a strong draw and enhance understanding of the more advanced technical subjects. The book moves into Bayes’ theorem by introducing the assumption of equal base rate probabilities. Incorporating prior knowledge with Bayes’ theorem is elucidated through base rates or contextual probabilities with medical case studies (e.g., patient history). The discussion Integr Environ Assess Manag 2019:165–169 SETAC Book Reviews Editor Glenn Suter US Environmental Protection Agency 26 Martin Luther King Drive MC-A110 Cincinnati, OH 45268 USA T 1-513-569-7808 F 1-513-569-7475 suter.glenn@epamail.epa.gov includes cases in which base rates are wrongfully ignored and in which they are difficult to estimate. In the final sections, case studies from different fields are provided as applications of Bayesian analysis including archaeology (carbon dating), document analysis (authorship determination), and forestry. The emphasis is on the foundation provided in earlier chapters and its relevance to more complex problems, but advanced concepts are introduced, such as change point analysis, parameter estimation, and model testing. The final section provides an example around a deceptively simple psychological puzzle of how Bayesian analysis (and information theory) can still provide insights when problems are fuzzy and probabilities are difficult to estimate. Jessop skillfully introduces important and initially daunting concepts. For example, Shannon entropy is first introduced as a measure of surprise from receiving news that the author’s favorite English football team had won a difficult match. The first time the author hears the news, he has a high level of surprise, but when subsequently informed of the win, the surprise (and the information content) is low. This leads into a discussion of the potential “surprise” or information value of a medical test and how entropy can be used as a measure of uncertainty and information value of evidence from modifying prior probabilities by the certainty the evidence can provide. In another example, Jessop communicates the effects of dependent variables through descriptions of gross domestic product predictions. If 2 banks copy each other’s predictions (complete dependency), then there is no increase in evidence value from receiving both reports and the lack of evidence strength must be accommodated in an assessment. The discussion of independence and dependence issues with evidence in legal cases is also of high interest. One of the key contributions of this book is the authors’ development of a useful tool for assessing evidence called the Bayes grid. Bayes grid is applied throughout the book to calculate posterior probabilities. I can see Bayes grids being adapted for meetings or group settings, communicating evidence analysis, and instances DOI: 10.1002/ieam.4109 C 2018 SETAC 166 Integr Environ Assess Manag 15, 2019—G Suter, Editor in which simple and intuitive ways of calculating conditional probabilities are needed. There are no requirements for remembering which permutation of Bayes’ theorem is correct with a Bayes grid; only an understanding of how to build the tables and transform numbers with simple algebra. As stated in the book, Bayes grid will not work on all problems, especially ones involving complex model building or continuous parameter estimation, but it does help to understand and approximate these approaches. By the end of the book, the point is made—Bayesian analysis can be a powerful and usable aid for a diversity of important problems involving information or evidence. This is not to say the discussion of Bayesian analysis is not practical or balanced. In 1 of my favorite sections of the book, Jessop provides a nuanced discussion on the use of Bayes’ theorem in the criminal court system that incorporates multiple diverging perspectives. The potential benefits of a Bayesian analysis are first introduced through the actual case of Sally Clark, in which a wrongful use of statistics by an expert witness led to a conviction for murder. The errors that led to this conclusion and their correction are clearly illustrated through the application of a Bayes grid. But how and whether to implement a Bayesian analysis in legal settings is still an unresolved issue. The legal expert debates and Jessop’s practical perspectives on the use of Bayes’ Rule are important and could pertain to many jurisdictions and settings. Let the Evidence Speak does not provide the technical background or guidance for choosing and using software for running a Bayesian analysis for complex problems. There are important topics that are not covered in sufficient detail for a full perspective (such as parameter estimation, the frequentist perspective, or Bayesian networks). But I do not view this as a detraction. This book provides a valuable resource for understanding and communicating Bayesian reasoning to a broad audience. Moreover, the information in this book provides a requisite foundation for designing, comprehending, and critically evaluating a complex Bayesian analysis. I will recommend Let the Evidence Speak to colleagues who are interested in learning about how Bayesian thinking works, looking for a very useful tool for teaching or rapidly applying Bayes’ Rule, in need of a handy reference on the strong insights a Bayesian analysis can provide, or wanting a technical book that is just plain fun to read. Disclaimer—The views expressed are those of the author and do not necessarily represent the views or policies of the US Environmental Protection Agency. 2018. eBook. 224 pp. ISBN 978-3-319-71392-2. $29.99. Springer International Publishing AG, Cham, Switzerland. John Carriger US Environmental Protection Agency Integr Environ Assess Manag 2019:165–169 ANTICOAGULANT RODENTICIDES AND WILDLIFE edited by Nico W van den Brink, John E Elliott, Richard F Shore, and Barnett A Rattner Warfarin and related rodenticides are commonly recognized as posing significant risks to nontarget species. Their mechanisms of action are common to all vertebrates, and they are sufficiently persistent to cause effects on various nontarget species due to primary exposures in bait and secondary exposure from scavenging of animals killed or debilitated by primary exposures. However, their use continues because rodents are such important pests and because there are few good alternatives. This book is an excellent summary of the risks of anticoagulants to wildlife, edited and authored by experts. The 14 chapters do a good job of covering the subject, with chapter topics ranging from use and exposure, through toxicology, epidemiology, mitigation, regulation, and potential alternatives. Some topics are missing, but that is inevitable because the editors are limited by their author pool and available information. For example, there is a chapter on the diagnosis of effects in birds but not one for mammals, and a chapter on risk modeling for nontarget small mammals but not for secondarily poisoned birds and mammals. Environmental chemistry is given minimal attention because transport and fate play minimal roles in wildlife risks from these compounds. Given the long history and common use of anticoagulant rodenticides, the information gaps identified by the authors are remarkable. Little is known about relative sensitivities to these agents and, even though tissue residues are commonly monitored, the relation between residues and effects are poorly understood. The importance of this knowledge gap is highlighted by the statement that “58% (2414 out of 4187) of predators analyzed in world-wide monitoring studies have AR tissue residues.” Further, contrary to expectations, shrews and hedgehogs commonly bear residues. Although mitigation is discussed at length, alternatives are not. For example, gene drives are being discussed for eradication of rodents on islands. That technology could be more effective than anticoagulants, would avoid the poisoning of native wildlife, wileyonlinelibrary.com/journal/ieam C 2018 SETAC