The 'Truth Contagion' Effect in the US Political Online Debate

During the last decade, evidence shows that US politicians’ conception of honesty has undergone a distinct bifurcation, with authentic and sincere but evidence-free "belief-speaking'' becoming more prominent and more differentiated from evidence-based "fact speaking''. Here we examine the downstream consequences of those two ways of conceiving honesty by investigating how users engage with fact-speaking and belief-speaking texts from Democratic and Republican politicians on Twitter. We measured the conceptions of honesty of a sample of tweets and replies using computational text processing, and checked whether the conceptions of honesty in the original tweets aligned with those in their replies. We also measured polarizing language in the replies and examined how it related to the honesty components in the tweets. We found that the conceptions of honesty used in replies aligned with those of the original tweets, suggesting a "contagion'' effect such that politicians determine the tenor of a subsequent conversation involving the public. We also found that belief-speaking tweets were more likely to trigger polarized language in replies, whereas fact-speaking tweets had the opposite effect. Our analysis highlights the crucial role of political leaders in setting the tone of the conversation on social media and it illustrates how evidence-based communication could help decrease audience polarization in online political debate.