We model a decision-maker who is facing a sequence of experiments, and whose perception is that outcomes are influenced by two factors - one that is well understood and fixed across experiments, and the other that is poorly understood and thought to be unrelated across experiments (the "error term"). Consequently, there is incomplete confidence that experiments are identical. We argue that a Bayesian model cannot capture the above, but that belief function utility can. Our formal contribution is to generalize the de Finetti Theorem on exchangeability to a framework where beliefs are represented by belief functions. Moreover, this is done while extending the scope of the bridge provided by de Finetti between subjectivist and frequentist approaches. In particular, a model of updating is provided.
In order to set up a list of libraries that you have access to,
you must first login
or sign up.
Then set up a personal list of libraries from your profile page by
clicking on your user name at the top right of any screen.