Political Communication and Conspiracy Theory Sharing on Twitter

Abstract

Social media has become an influential channel for political communication, offering broad reach while enabling the proliferation of misinformation and conspiracy theories. These unchecked conspiracy narratives may allow manipulation by malign actors, posing dangers to democratic processes. Despite their intuitive appeal, little research has examined the strategic usage and timing of conspiracy theories in politicians’ social media communication compared to the spread of misinformation and fake news.This study provides an empirical analysis of how members of the U.S. Congress spread conspiracy theories on Twitter. Leveraging the Twitter Historical API, we collected a corpus of tweets from members of the US Congress between January 2012 and December 2022. We developed a classifier to identify conspiracy theory content within this political discourse. We also analyzed the linguistic characteristics, topics and patterns of conspiracy tweets. To assess classifier performance, we created ground truth data through human annotation in which experts labeled a sample of 2500 politicians’ tweets.Our findings shed light on several aspects, including the influence of prevailing political power dynamics on the propagation of conspiracy theories and higher user engagement. Moreover, we identified specific psycho-linguistic attributes within the tweets, characterized by the use of words related to power and causation, and outgroup language. Our results provide valuable insights into the motivations compelling influential figures to engage in the dissemination of conspiracy narratives in political discourse.

Keywords: Conspiracy Theory, Political Communication, Twitter, NLP, Psycho-linguistic characteristics