Bayesian Computations Protect Sexually Violent Predator Evaluations from the Degrading Effects of Confirmatory Bias and Illusions of Certainty: A Reply to Doren and Levenson (2009)

Wollert (2007) published an article in Psychology, Public Policy, and Law on Poor Diagnostic Reliability, The Null-Bayes Logic Model, And Their Implications For Sexually Violent Predator Evaluations.” The preceding article by Doren and Levenson (2009) criticizes this article. The article at hand answers their criticisms and also considers the importance of SVP selection systems to SVP evaluations, examines factors that degrade the evidentiary value of SVP evaluations, and suggests procedures that might be followed to preserve their value. It is concluded that many unresolved issues that pertain to SVP evaluations might be clarified by the application of Bayesian analyses and that evaluators could avoid problems associated with confirmatory bias, such as illusions of certainty, by using Bayes’s Theorem (Bayes, 1764) to appraise the adequacy of their SVP selection systems.