Change We Can Believe In: Structural and Content Dynamics within Belief Networks

Scholars have used network analysis to explore the structural properties of mass ideology. This article incorporates two important, though ignored features in past research by investigating how time shapes the properties of belief networks for different populations of people-those who exhibit high and low levels of political knowledge. We find that (1) belief network density increases asymmetrically among people with high relative to low knowledge; (2) symbolic preferences are more central to belief networks irrespective of survey timing or population; in contrast, policy beliefs exhibit some increase in centrality over time among the politically knowledgeable; and (3) a belief's centrality is unrelated to the amount of change it explains in other beliefs. Troublingly, this latter finding presents problems for describing belief networks using the vernacular of Conversian belief systems-a disconnect that seems grounded in the mismatch between Converse's individual-level theory and network analysis' population-based properties.