Eleições gerais 2014: fraude ou não?

Abstract

Frequent movements and discussions question the efficiency of the electoral calculation’s process and the second round’s results of Brazilian elections in 2014. Mostly when it is about the presidential election. This study intends to evaluate the reliability of the election results for presidency through a statistic method called the Newcomb-Benford law or the First Digit law, which delineates a probability for the digits of a sample. Taking into consideration the peculiarity of the distribution of digits of aggregate data, this law is wildly applied to detect evidences of fraud where there may be manipulation of numbers, such as in invoices, numbers presented in financial reports or, as proposed, election results. It was possible to conclude that the method is an effective support tool to auditing and controlling the adulteration. From the statistic analysis of the data, it was verified the suitability of the numbers of votes for the candidates Dilma Roussef and Aécio Neves and also for null votes. However, in relation to blank votes, the data did not adhere to the law.