One problem to which we must address ourselves before constructing any deductive model of voting behaviour is why people vote at all. On the face of it at least, this seems to be a stumbling block, given the theoretical problem that, compared with the costs of voting, the benefits to be expected are very small; and given the empirical problem that large numbers of people turn out to vote when they cannot possibly hope to influence the result, while others stay at home on the few occasions when a close result is expected. The most common attempted solution amounts to saying that people like voting, or at least that they feel guilty about not voting. This solution can be traced from the early work of Downs to the more recent and rigorous analyses of Riker and Ordeshook. Motives such as citizen duty, the satisfaction of belonging to a democracy, the satisfaction of supporting a particular party, and so on, are ascribed to voters. Each voter is then assumed to conduct a ‘calculus of voting’ in which he or she weighs up these benefits against any inconvenience involved in going to the polling booth. For many the net effect will be positive and voting will therefore be rational.