Google and the Brazilian Elections
elections; Google; SERP; browsing history; search engines; Brazil
This thesis investigates the role of Google in the Brazilian elections of 2020, focusing on the mediation of political information in a subnational election. The study was conducted with a methodological approach that combines: quantitative analysis of 518,000 search results (SERP - Search Engine Results Page), collected between August and November 2020 in a sample of Brazilian municipalities; two rounds of longitudinal panel surveys with 2,771 individuals, representative of the Brazilian population (before and after the elections); and the collection of the search history of some of the respondents. This work answers for whom, how, and when Google is relevant in elections, maps the main sites displayed for different search terms, and, through a “quasi-experiment,” assesses the existence of interference by Google in search results at the start of the electoral period. The results indicate that voters with strong political preferences are the ones who most frequently conduct political-electoral searches, challenging the Search Engine Manipulation Effect hypothesis; show that Google is crucial for accessing information about candidates, especially in the last three days before the election; highlight that 10 sites accounted for 60% of the political-electoral search results during the period; and suggest signs of possible Google interference at the beginning of the electoral period.