Voter coordination dominates bandwagon as the consequential channel

🟢 The coordination theory (Cox's M+1 rule) explains the heterogeneity pattern better than the bandwagon theory.

Draws on: the coordination-shift finding — peak β at rank 2 in small munis, rank 3 in runoff-eligible munis. The position-of-peak prediction is the cleanest design-level discriminator the project supports, and bandwagon predicts the same peak position in both muni-size classes whereas coordination predicts exactly the shift observed.

Strength: 🟢 because the prediction is qualitative (which rank peaks?) and the observed direction matches uniquely. SEs on the runoff rank-3 estimate are wide, so the magnitude isn't pinned down, but the direction-of-shift test passes cleanly.

Sources.

Own analysis: AN-004
Cross-refs: H6 coordination-peak; H7 bandwagon-peak; theory.md § Polls as coordination devices; theory.md § Polls as bandwagon triggers

Cited analyses

AN-004 yellow causal

Peak β shifts from rank 2 in small munis (+10.14) to rank 3 in runoff-eligible munis (+9.43). Direction matches Cox's M+1 rule (M+1=2 vs M+1=3); bandwagon would predict rank 2 in both. The cleanest design-level discriminator the project supports with current data.