Making decision models probabilistic
Andrew H. Briggs, Karl Claxton, Mark J. Sculpher · 2006
Abstract In this chapter, we describe how models can be made probabilistic in order to capture parameter uncertainty. In particular, we review in detail how analysts should choose distributions for parameters, arguing that the choice of distribution is far from arbitrary, and that there are in fact only a small number of candidate distributions for each type of parameter and that the method of estimation will usually determine the appropriate distributional choice. Before this review, however, consideration is given to the rationale for making decision models probabilistic, in particular the role of uncertainty in the decision process. There then follows a discussion of the different types of uncertainty, which focuses on the important distinction between variability, heterogeneity and uncertainty.