Identifying and characterizing ideologically homogeneous clusters on Twitter and Parler during the 2020 election

Area of Science:

• Social network analysis
• Computational social science
• Political communication

Background:

• The 2020 U.S. election cycle saw widespread doubt cast on electoral legitimacy via public statements and social media.
• These sentiments culminated in the January 6th Capitol insurrection.
• Understanding the online dynamics of movements like #StopTheSteal is crucial for analyzing political discourse.

Purpose of the Study:

• To analyze network-level and content-level data to understand the effectiveness of the #StopTheSteal movement online.
• To identify and compare ideologically homogeneous groups on mainstream (Twitter) and alternative (Parler) social networks.
• To investigate the linguistic characteristics differentiating political discourse.

Main Methods:

• Utilized Louvain clustering to identify ideologically homogeneous groups.
• Developed and applied a novel homogeneity metric for group analysis.
• Analyzed discussion data from Twitter and Parler.
• Performed content analysis on text characteristics of different groups.

Main Results:

• Ideologically homogeneous groups were found to be more effective in spreading messages compared to diverse groups.
• Right-leaning groups exhibited stylistic similarities to conspiracy theory texts, specifically using "worldbuilding" language.
• Significant differences were observed in the text characteristics between left- and right-leaning homogeneous groups.

Conclusions:

• Ideological homogeneity significantly amplifies message dissemination in online political discussions.
• The linguistic style of right-leaning discourse within the #StopTheSteal movement aligns with patterns found in conspiracy theories.
• Network and content analysis provide valuable insights into the spread of election-related misinformation and political mobilization.