Inference under imprecise probability assessments

Abstract Under partial knowledge, the use of the precise probability might be misleading. However, it is possible to process imprecise assessments, such as comparative previsions or grades of previsions. Rules for checking their coherence with the theoretical model and for making inference are given. Sometimes the derived conclusions might provide us with a complete answer for a given problem. In any case, technical tools can measure the imprecision of the answer and reveal if the analysis is thorough enough.