Assessing evidence and uncertainty:informed decision making in drug approval and replication target selection

This dissertation examines the general question of ``when do we have enough evidence?''. Informed decision making is considered in two contexts. Part I zooms in on the use of statistical evidence in regulatory decision making. Specifically, the evaluation of statistical evidence in the context of drug approval by the FDA in the United States. The evidential standards for psychotropic and cancer drugs are discussed using Bayes Factors. The Bayesian re-analyses revealed a substantial degree of variability in the strength of statistical evidence supporting the efficacy of novel psychotropic and cancer drugs approved by the FDA in the pasttwo decades. Although heterogeneity is to be expected due to the uniqueness of each drug evaluation, it is concerning that some drugs were approved despite the lack of statistical evidence of their efficacy or even evidence of their ineffectiveness. Part II considers decision making in the context of replication target selection in (clinical) psychology. To ensure that individual study results are not chance findings , studies are increasingly being replicated. However, given limited resources, only a small number of studies can be replicated. To make best use of the available resources, replication target selection should be systematic, justified, and transparently communicated. This dissertation describes the evidence considered when selecting replication targets in practice and presents formal suggestions for how to select replication projects in clinical psychology and psychology in general