Laver is undoubtedly right in suggesting (a) that ideas of rationality are more of a hindrance than a help in explaining electors' behaviour, but (b) that a rational-choice theory of party competition can still be superimposed on an a-rational explanation of voting choice. Since he develops this position through a critique of Robertson's ‘wide’ definition of voter rationality, it is only fair to point out that Robertson's ascription of modified office-seeking to politicians forms the best developed rational – choice theory of party competition. The hypothesis that politicians seek votes by widening their issue-appeals when they think the election is competitive, but stress partisan appeals otherwise, has been validated for British and United States' elections from 1920 to 1974 – a more general and more rigorous check than has been applied to any competing hypothesis. Moreover it has already been incorporated with an a-rational account of voting behaviour in the form Laver advocates. The key elements of this synthesis are:

(1) The location of electors and parties in a party-defined space, i.e. a space where individuals and groups are located in terms of their closeness to the election alternatives (party choices and non-voting). This is analogous to, but quite definitely not, the space produced by party identification, to which there are too many measurement objections.

(2) The division of influences upon voting into predispositions (loyalties, social group traditions, etc.) and cues (current issues, candidates, etc.). Predispositions are associated with electoral stability, i.e. limited and slow movement in the party-defined space. Cues have the potential to produce rapid change and widespread movement of electors, often associated with the appearance of new parties.

(3) Modified office-seeking by politicians. In elections which they expect to be competitive there is thus a premium on the introduction of new cues, while in other elections old partisan appeals will be made to existing predispositions.