Mixing, Matching, and Winning? The Electoral Effects of Fusion Voting in Connecticut and New York State Legislative Elections 2012-2020

Abstract

Fusion voting—the practice of two or more parties cross-endorsing the same candidate— operates in only four U.S. states, and its effects on state legislative elections have never been systematically tested. Using linear probability models on every Connecticut and New York general election from 2012 to 2020 across four legislative chambers, I find that fusion’s electoral relevance is real but operates through different mechanisms in each state. In Connecticut, minor party votes average 3.28% per fused race and are concentrated in competitive contests; each percentage point delivers a large marginal effect on the probability of winning, but the signal disappears in cycles with few close races. In New York, minor-party votes average 5.90% and are distributed across races of varying competitiveness; the per-vote effect is smaller but consistent across both chambers and nearly every cycle, even as individual minor parties rise and fall. Pooled interaction models confirm the asymmetry, showing Connecticut’s steeper but more fragile slope against New York’s shallower but more durable one. The implication for candidates is the same in both states: seek the cross-endorsement. The votes it delivers are positively associated with winning, and the cost of seeking them is low.