One of the most employed explanations for vote choice is ideology. However, it competes with others that stress the importance of sociological characteristics, psychological mechanisms, or retrospective evaluation. This paper, rather than being the arbiter between them, assumes that either mechanism can be at work in the same electorate simultaneously. It examines which contexts make the left-right dimension more or less pertinent as a decision-making device. Its primary explanation is the number of dimensions that parties compete on, which empirically varies considerably across countries. I implement an agent-based model to develop hypotheses on how voters with incomplete information choose among parties in systems of varying complexity. I test these predictions combining manifesto data with surveys that allow me to operationalize how well the spatial model of party competition explains party preferences of individuals, thus augmenting earlier work which has mostly examined this aspect on the aggregate level.
