Robust inference in probability under vague information.
Vague information can be represented as comparison of previsions or comparison of probabilities, and a robust analysis can be done, in order to make inference about some quantity of interest and to measure the imprecision of the answers. In particular, in some decision problems the answer can be unique.