The Machine Works: Why Turnout Buying is More Effective Than it Appears

Turnout buying is a mainstay of machine politics. Despite strong theory that selective incentives should spur turnout, meta-analyses of empirical studies show no effect, thus making machine politics seem irrational and unsustainable. I argue that the apparent failure of turnout buying is an artefact of common measurement decisions in experimental and observational research that lump together turnout buying, abstention buying, and vote-choice buying. Data generated using these compound measures include countervailing and null effects that drive estimates of the effects of each strategy toward zero. I show that machines have incentives to diversify their strategies enough to make compound measures substantially underestimate the impact of turnout buying. I propose simple alternative measurement approaches and show how they perform using new survey data and a constituency-level analysis of machine strategy in Mexico. Findings close the gap between theory and facts and reaffirm the rationality of machine politics.