Consistency, not waste aversion, explains apparent sunk‐cost effects in politics

Do sunk costs shape how voters judge political leaders, or do reputational signals carried by current choices dominate? Across three preregistered experiments (total N = 1462) using vignettes about a national rail project, we orthogonally manipulated sunk costs (high vs. absent), leaders' decisions (continue vs. abandon), campaign commitments (pro, anti, none), and reminders of prior voting behavior. In Experiment 1 ( N = 346), Bayesian ANOVA showed that evaluations (vote likelihood, perceived honesty, competence) were reliably higher when leaders continued rather than abandoned a project, with no main effect of sunk costs and no interaction. Experiment 2 ( N = 719) replicated the null sunk‐cost effect and further found no effect of campaign commitments once the leader's current decision was known. Experiment 3 ( N = 397) used a structural equation model to test mediation: perceived commitment—but not perceived sunk costs—robustly predicted evaluations, and abandonment sharply reduced perceived commitment; prior‐vote cues were weak and did not mediate effects. Together, the studies indicate that people take continuation to signal commitment and view it positively, whereas abandonment stopping signals a breach of commitment and is penalized. Apparent sunk‐cost sensitivity in politics may often reflect a premium on consistency rather than waste aversion. We discuss implications for designing “face‐saving” off‐ramps and reframing evidence‐based reversals as fidelity to higher order commitments.