Study Analyzes Social Media Behavior Trends During Election Period

The monitoring of the 2024 São Paulo elections, carried out by the Center for Violence Studies (NEV), in partnership with the team of researchers from the Institute of Mathematical and Computer Sciences (ICMC), both from the University of São Paulo, investigates the topics that generated the greatest engagement on social media. The project, entitled Crime, Insecurity and Legitimacy: A Transdisciplinary Approach , aims to understand the interactions between public security, human rights and public opinion, offering an analysis of the role of social media in the electoral dispute.

According to Pablo Romero Almada, a postdoctoral researcher at NEV, the project aims to monitor discourse on social media and its influence on the legitimacy of democratic institutions, especially during elections. “This project tends to analyze the relationships of democratic legitimacy and public opinion. Since we are analyzing the elections, we seek to analyze them to understand how these elements of legitimacy of institutions appear at this time,” he explained. This analysis seeks to understand how the most debated topics affect voters’ perception of public safety and other rights.

Regarding the methodology used in the research, he comments: “We have a base, a data collection technique, which seeks to bring together content from various social networks, at least the main ones, and also to search for content from candidates for Mayor of São Paulo, candidates for city councilors”. The methodology brings a novelty in segmenting the candidates: “Looking for some candidates who come from public security, candidates who self-declare as black or brown, and also some candidates who are between 18 and 29 years old”. In the second report,  the research also included the analysis of some fact-checking agencies, in addition to what happened on TV Cultura, where there were episodes of violence between candidates.

Social media

According to Almada, the use of social media has shown a great diversity of practices, depending on the platform. Video networks, such as Instagram and TikTok, have a much greater power of engagement than Facebook and Twitter in recent elections. The researcher pointed out that, often, political content is mixed with other types of posts, which can confuse voters. In addition, the role of algorithms is highlighted, since the content seen is not always directly produced by the candidates, but rather by parallel profiles, such as cutting channels.

Among the most recurrent themes, the team’s first report indicated that candidates from public security backgrounds tend to mobilize discourses focused on surveillance policies, such as the use of body cameras, while young candidates tend to raise ideological discussions; racial profiling, on the other hand, often deals with the construction of the idea of ​​the periphery. In the second report, the expert states: “In addition to these themes, we also added what the fact-checking agencies were analyzing. This was quite interesting, because we were also able to see how they ended up playing an important role in the investigation of themes related to the September 7th mobilizations. In general, this second report also brings something new to the topic, which is precisely this question of how profiles grow so quickly.”

Another interesting point raised by the study is the speed at which certain profiles can grow on social media, driven by algorithms. “This ends up confusing users and creating complex and turbulent scenarios so that the user, when accessing social media, has direct access to information,” adds Almada.

“Social networks end up enhancing certain types of content, which end up spreading, and we don’t know what criteria these algorithms end up using. What we do know, in the end, is that mastering these techniques and metrics allows for a certain imbalance in relation to content delivery. It is an interdisciplinary debate that allows us to conduct this analysis, a debate between social sciences, political science, and data sciences complementing each other, precisely to try to observe this dynamic,” he concludes.