From words to votes: Decoding sentiment in Brazilian presidential candidates' tweets

Abstract

This study investigates the sentiment analysis of tweets from the four most influential presidential candidates in the 2022 Brazilian elections. Tweets from July 1 to October 30 were gathered using a custom Python script, and the sentiments expressed in these tweets were analyzed with the VADER sentiment analysis library. Two econometric models were estimated to study temporal shifts in sentiment and the effect of opponents' sentiment. Results showed strategic shifts in sentiment during different campaign phases. Lula da Silva, the election winner, showed significant increases in negative tweets during the second-round campaign and seemed to adjust his sentiment based on opponents' sentiments, suggesting the potential effectiveness of adaptable communication strategies. This study contributes to understanding political communication dynamics on social media and their potential impact on electoral outcomes.