We model an organization by a preference relation over uncertain acts satisfying a
set of axioms capturing a legitimate decision making process. We show that our set of
behavioral conditions is equivalent to a legitimate representation of preferences: there
exist two nested sets of priors in which the smaller one represents the set of priors from
individuals who occupy higher positions within the organization, while the larger one
takes into account also probabilistic beliefs from people of lower positions. A legitimate
decision that supports an act f against an act g includes authorization, which occurs
when the smaller set of priors unanimously deems the act f better than the act g, a
la Bewley (2002), and endorsement, which occurs when the larger set of priors deems
the act f better than the act g according to a maxmin rule a la Gilboa and Schmeidler
(1989).
