An application of the multicandidate calculus of voting to the 1972 and 1976 German federal elections

Two important assumptions underlying the Riker-McKelvey-Ordeshook, expected-utility-based, multicandidate calculus of voting are that citizens decide whether or not and for whom to vote in a single-member district (SMD), plurality-winner electoral system and that they ignore aggregate outcomes, that is, how citizens in other districts may vote. Aranson modifies this calculus by investigating citizens' decisions in proportional-representation (PR) systems and/or situations in which aggregate outcomes require parties to form parliamentary coalitions to govern. This study operationalizes and applies the probability and utility components of this modified calculus to data on the 1972 and 1976 West German federal elections, which combine features of SMD and PR systems and of which coalition politics is an integral part. Expected-utility calculations used to predict reported voting decisions achieve reasonably good levels of aggregate accuracy. Further analysis reveals that gradations in respondents' expected utilities, as well as in their utilities and surrogate measures of the subjective probabilities of electoral outcomes, relate significantly to reported voting decisions. Also, variables that model coalition considerations are commonly the most significant.