The discovery of the effects of risky shift focusses the interest on the study of decision processes. Nevertheless the researches carried on until now have always shown confusion between the theoretical and experimental analysis of these processes and the risk-related behavior. In the first part of the article the authors try to clear this confusion as well as to define the conditions of a systematic study of the mechanisms of decision leading either to an averaging or to a polarization of individual views. The authors insist on the necessity to compare groups rather than individuals to groups. To start with, one hypothesis concerning the effects of group organization on the degree of polarization of views is proposed. The experiment described in the second part of the article confirms the hypothesis: A group in which individuals have the possibility to communicate with each other and to interact directly take more extreme decisions than a group deprived of this possibility. It is presumptuous and inexact to state that groups take more extreme risk than individuals whereas it is right to say that certain groups take risks when circumstances are favorable.