Collective heat-transition voting: An agent-based model integrating bounded confidence, social pressure, and information diffusion

The transition to neighborhood-scale district heating in the Netherlands frequently stalls due to the challenge of achieving a collective agreement among individual homeowners. To explore the micro-dynamics of this collective action problem, this paper introduces an agent-based model that integrates three key drivers of collective choice: heterogeneous decision styles (deliberative, social-comparison, repetition, imitation), networked social pressure governed by bounded confidence and minority influence, and dynamic information diffusion with spatiotemporal decay. A novel three-layer architecture maps the progression from continuous opinion exchange to voting intention and a final binary vote. Scenario experiments reveal the system's fragility and key leverage points: the benchmark scenario, which targets community leaders with early positive information, achieves an 88 % yes-vote share, while early endorsement by cross-cutting brokers who link different sub-communities achieves a 97 % yes-vote share and high agreement (0.866 global consensus; 1 indicates unanimity). Conversely, a single, early negative narrative can trigger a rapid collapse of support from over 90 % households within 30 simulation rounds. Late-stage positive messages proved most efficient at tipping fence-sitters, achieving 95 % yes votes with smaller aggregate opinion shifts than early messaging. A comparative analysis across small-world, scale-free, and random networks demonstrates that these mechanisms are robust. These findings indicate that information timing, targeting, and valence serve as critical non-economic levers in community energy transition, and yield actionable heuristics for policy: target credible brokers such as local energy coaches with data-rich information early, use visible social proof just before a decision point, and actively neutralize negative narratives to prevent cascading resistance.