Title: Generalized Zipf's Law in proportional voting processes

Abstract: Voting data from city-councillors, state and federal deputies elections are analyzed and considered as a response function of a social system with underlying dynamics leading to complex behavior. The voting results from the last two general Brazilian elections held in 1998 and 2000 are then used as representative data sets. We show that the voting distributions follow a generalized Zipf's Law which has been recently proposed within a nonextensive statistics framework. Moreover, the voting distribution for city-councillors is clearly distinct from those of state and federal deputies in the sense that the latter depicts a higher degree of nonextensivity. We relate this finding with the different degrees of complexity corresponding to local and non-local voting processes.