Reciprocal causation in a model of the vote: Replication and extension

Abstract

Page and Jones, and others, have argued that recursive models of the vote fail to take account of the inherent complexity of the decision processes involved and that estimates from these models are prone to errors, which result in understatement of the amount of policy voting. They propose a causal structure, in which reciprocal influences are modeled explicitly, and estimate its parameters with partially independent samples from two presidential elections. Their model, however, is restricted to presidential contests and omits economic issues. Similarly, the literature on economic voting has ignored the presence of potential reciprocal influences on economic policy attitudes. This paper develops a novel measure of economic issue preferences, modifies the earlier model to include both economic and social issues, and tests the model on data from an independent sample of voters in the 1978 congressional elections. Estimates from the elaborated model generally comport well with the results obtained by Page and Jones, and the findings show a higher degree of economic issue voting than most previous studies.