Snacks, meals, or buffets? (Dis)Aggregating social influence across social environments

Abstract

Much survey-based research on social influence in politics has measured disagreement within a single social environment, most often individuals’ core networks of closest ties. Yet individuals are simultaneously embedded in multiple, overlapping social environments that may push them in the same or opposing political directions. We use original survey data measuring political disagreement across four distinct social environments—core networks (closest ties), periphery networks (weaker ties), affiliation networks (voluntary organizations), and social contexts (geographic community)—to ask a foundational but underexplored question. Should these environments be studied separately or in aggregate? We first describe the landscape of disagreement across environments, documenting meaningful variation in both levels and correlates of disagreement exposure. We then ask whether disagreement across environments reflects a common underlying dimension or operates through distinct mechanisms, finding that a single latent factor accounts for approximately 85% of joint variability, consistent with but not conclusive evidence for aggregation. We assess what is gained and what is lost under each approach. Disaggregated models reveal that environments differ in their associations with partisan and ideological strength, while an aggregated “global disagreement” measure reveals that relying solely on core networks underestimates the relationship between the broader social environment and political attitudes. Our findings indicate that both approaches capture substantively distinct patterns, and that the appropriate level of aggregation depends on the inferential goal.