From Contestation to Camaraderie: Structural Similarity Dampens Toxic Discourse in Polarized Social Groups

Partisan animosity—negative thoughts, feelings, and behaviors towards an outgroup—has been on the rise around the world and has been linked to such societal threats as theweakening of democratic institutions and political violence. It is especially prevalent inonline social interactions and frequently results in group members not only disliking oneanother but even resorting to toxic discourse—language that is uncivil, intolerant, orthreatening toward others. We develop a novel theoretical account of toxic discourse’santecedents—one that spotlights individuals’ positions in the network structure of thegroup. We hypothesize that, in polarized online groups, the more two individuals arestructurally similar to one another, the less likely they will be to use toxic language intheir interpersonal communication. We further propose that the relationship betweenstructural similarity and toxicity will be moderated by group polarization: The more(less) polarized a group is, the more (less) structural similarity will tend to dampentoxic discourse between individuals. To evaluate these ideas, we draw on a rich dataset,encompassing more than 25 million comments made by over 1.7 million users in sixpolarized communities on Reddit. We build on recent advances in machine learning andnetwork analysis to derive an omnibus measure of interpersonal structural similarity.We validate our measure and then report results from cross-sectional analyses andtwo natural experiments that provide support for our theory. We discuss implicationsfor research on partisan animosity, group polarization, the measurement of structuralsimilarity, and the interplay of structure and culture.