Estimação de ideologia via dados do Twitter : interpretação para o caso brasileiro e aplicação para movimentos de renovação política

Abstract

In political science, various tools have been utilised to build spatial models in order to quantify the ideological positions of political actors, the most common of them being roll-call data analysis, codification of political texts and the application of surveys to political elites, experts or the electorate in general. While these different methods have strengths and weaknesses that turn each of them complementary to others, they all share a common shortcoming: most of the time, the results can estimate only the ideal points of either political parties or individual politicians who have been elected into office at least once. In this sense, social platforms stand out as a new data source with the potential to connect elites to other groups, like debutting candidates, social movements, pressure groups or common users. This dissertation is divided in three parts. In the first of them, the existing methodologies are revised and compared to each other, and the correlations between each pair of sets are analysed. Chapter 2 presents a method for ideal points estimation using Twitter data, with the goal of explaining the determinants of the values obtained by it. The accounts of legislators from the Brazilian Chamber of Deputies in 2015 and 2022 are utilized, and profiles followed by at least 30 of them being fed into the model. The resulting models suggest a political space influenced by ideology, pragmatic alignment to government and structural polarization related to the competition for the Executive. Hypotheses are tested, with Twitter ideal points serving as dependent variables and presidential saliency of political parties, the ideology estimated by other means and rate of adhesion to government at the floor independent variables. Lastly, the third chapter aims, in first place, to estimate the position of the Political Renewal Movements created from the second half of the last decade, which, although having their own identities, need to launch candidates through the existing political parties implies an ambiguous relation of cooperation and conflict between their leaders and those of the established parties. Thus, the second objetive of the last chapter is to test the significancy of some factors determining the party choice of members of renewal movements, among them their ideological proximity to the various political parties. To infer ideological position of users without a political mandate, the list of profiles followed by members of these movements are projected onto the model developed at chapter 2, what is followed by the test of hypotheses about the ideology of individual candidates, the movements which they belong to and the parties to which these new electoral competitors choose to affiliate.