The conventional wisdom of British psephology is that voters' party preferences are a function of their political attitudes: people choose the party whose policies are closest to their own political beliefs. This view has recently been contested by Rose and McAllister. Discriminant analyses have been applied to various sets of attitudinal data in the 1983 British Election Study file. They show that votes in 1983 could be predicted from attitudes for well over half of those who voted either Conservative or Labour: Alliance voting was less predictable, although in some analyses over half of those expected to vote Alliance did so. Rose and McAllister's thesis is thus rejected.