to increase their votes from the middle class. For example, if one bases one's calculations on the class structure of 1970, the Danish Social Democrats would have been able to recruit and keep the largest possible number of voters if the share of workers had been 41 per cent instead of 72.5 per cent as was in fact the case; for Norway the comparable figures are 29 and 47 per cent, respectively, for Sweden 21 and 78 per cent. Concern for other values (class, ideology, relations with the trade union movement) has outweighed the desire for votes. The French, Finnish, and German Social Democrats had, prior to 1933, an even more difficult situation. The 'completely classless strategy' would have been the best one with regard to vote- maximization, with the 'class strategy' coming second and the 'limited classless strategy' last. The parties vacillated between these alternatives. Up until 1970, for example, the French Socialists seemed to choose almost the worst possible strategy. They tried to win middle- class votes without losing any workers but failed on both counts. The German left landed half-way between the next worst strategy, which in this case would have been an unqualified bid for the working-class vote, and the worst strategy, which was a half-hearted appeal to the middle class.
The authors formed their conclusions into a powerful polemic against Downs and the vote-maximization theorem. Parties cannot jump freely from one position to another in order to adopt the policy they believe the majority of voters accepts at any one time. To disregard such limitations on the parties' room to manœuvre is to commit the error of reducing the analysis of political science to 'empty formalism'. Downs and his followers have conjured up a strange world in which votes alone are worth anything, but in real life parties also struggle to achieve other goals. Equally 'absurd' is the conception of these theoreticians that the views held by the electorate are independent of the parties and something to which the politicians are obliged to accommodate themselves. On the contrary, it is the politicians who 'present the public with images of society, evoke collective identifications, instill political commitments' and thereby create opinion.25
The conclusion of this chapter is that the self-interest hypothesis cannot be sustained with regard to politicians either. The basis for this