Violence and authoritarianism are present factors in most societies, as well as prejudice and discrimination. Social variables, such as race, profession, and place of residence, are important factors that interfere with the portrait of each member of society and may override the truth and how judicial laws are applied to them. The main goal of this work is to apply machine learning techniques to predict the verdict given by the population to the accused at a crime scene. The database was created with the social information collected from 1102 participants from 21 different states of Brazil, in the year 2020. The data used are part of the “Brasil 2020 Decide” survey, which has assertions about social conditions and authoritarian behavior related to economic conditions, color and race. In the described crime scenario, the SVM (Support Vector Machine) algorithms achieved an average accuracy of 68% in predicting the sentence for the accused.