Permutation-Based Testing of Topic Co-occurrence: A Network Analysis of Reddit Debates on DOGE, Tariffs, and the Big Beautiful Bill

Do policy discussions follow predictable patterns? When individuals raise policy concerns on online platforms, do they systematically pair these concerns with specific related arguments? While these questions are central to understanding online political discourse, we lack systematic methods to identify which topics genuinely co-occur versus appearing together by chance. To address this gap, we analyze Reddit discussions of three major policies from Trump's early second presidency: the Department of Government Efficiency, Trump's tariffs, and the Big Beautiful Bill. Our approach combines three key elements: dictionary-based identification of policy considerations (developed with LLM assistance), permutation-based statistical testing to identify significant co-occurrences, and network visualization to map thematic clusters. These topic co-occurrence networks reveal the underlying structure of policy discourse on Reddit and illuminate how online political discussions are organized around predictable argumentative frameworks.