An experimental study of preferences for information.

Summary In most concrete choice situations, decision-makers cannot predict the outcome of their action with certainty. Decision theory generally distinguishes between the typical case of risk, where the probabilities of different events are objectively known, and the situation of ambiguity, where the informational context does not allow for such objective probabilities.However, agents often hold partial information in the form of sets of past observations, leading to situations that go beyond the traditional dichotomy between known and unknown probabilities.This dissertation provides an empirical analysis of decision behaviors under uncertainty when statistical information is available. The first two chapters focus on choice behavior in the presence of information, i.e., when different options are described by ex ante data sets. These chapters study the influence of the accuracy of information on the perception of ambiguity and on attitudes towards risk and ambiguity.In the third chapter, I consider the case in which agents are not informed and have the opportunity to acquire information. In the third chapter, I consider the case in which agents are uninformed and have the opportunity to acquire information. This chapter proposes an experiment whose objective is to study the way individuals evaluate additional information.