It is widely believed that there exists in the United States a phenomenon called ‘clientelism’: that most federal government departments in the USA do not really control the interests with whom they do business but rather are controlled by them; that the interests and the departments between them typically have great influence over the department secretaries appointed by the president; and that in consequence a department secretary is, in reality, more likely to act as the interests' spokesman, or as his department's, than as the spokesman of the president. Thus it is often asserted that ‘cabinet officers are a president's natural enemies’. My aim in this article is to show that this belief, as it relates to the relationship between department secretaries and presidents, is widely held, plausible, but mistaken.

The image of voting in the United States developed by political scientists over the last decade differs markedly from the perspective offered in that classic study of electoral behaviour in the 1950s, The American Voter. Whereas the authors of The American Voter painted a rather unflattering portrait of the way in which the voter of the 1950s made his electoral choice, contemporary research has begun to discover some unexpected virtues in the American electorate of the 1960s and early 1970s. Compared to his counterpart in the previous generation, today's voter seems to attach less significance to his party identification, and more importance to his perceptions of the parties' stands on issues with which he is concerned, in deciding which party to support in presidential elections. Indeed, it would perhaps be only a slight exaggeration to suggest that the notion of electoral choice has now become a realistic, and not merely a metaphorical, manner of speaking about the American voter.

The immodest purpose of this essay is to analyze the traditional arguments for political equality, and, with the help of a case study, to determine when these arguments apply and when they do not.

To live is to have problems. However the country as a whole fares, the individual has bills to pay, work to do, children to worry about – to mention only a few of the commonplace problems that people face in their daily lives. Commonplace or not, these are problems that people must wrestle with. They are immediate, inescapable, and serious, far more so for most than the ‘large’ issues facing the country. Students of voting have long suspected that such problems may influence political choices, but key questions remain unanswered. Which personal problems are taken to be political and which non-political? Do personal problems have an impact on voting behavior only if they are taken to be political? When and how do personal problems become translated into political choices? In this paper we shall address such questions as these.

Whips have always aroused curiosity. Their peculiar title, which suggests coercion and social control, derives from ‘Whipper-in’, a man who keeps headstrong hounds running with the pack. The packs, that is to say the parliamentary parties, are heterogeneous associations of independent and egocentric individuals, not at all the sort to shy away from dissent and rebellion. Between 1966 and 1970 the Labour Government faced backbench opposition on almost every major policy; the 1970–74 Conservative Government was actually defeated five times by its own supporters. Still, although it may well be true that ‘the golden age of ultra-disciplined parties is past’, when the time comes to march, and regardless of the issues involved, most Members seem ready to follow a lead: remarkably few bolt upon reaching the division lobbies. Even the recent ‘undisciplined’ record must appear to administrators everywhere as very successful social control indeed.

It is probably accurate to say that one of the major growth areas in political science over the last decade has been the analysis of public expenditures. Beginning with what now seems prehistory and the original Dawson and Robinson article, a growing number of books and articles has sought to explain the growth of public expenditures, especially of expenditures on the social services. Despite the numerous criticisms of these studies, they have added immensely to the discipline's knowledge of public policy processes, if for no other reason than that the critics themselves have been forced to engage in the types of research that they believed were really necessary to determine the causes, processes and consequences of expenditure decisions.

Analyses of elections, whether academic or journalistic, are permeated by terms drawn from physics. We discuss the inertia or volatility of an electorate, the momentum of a campaign, pressures and cross-pressures, party actions and reactions, the impact of short-term forces. Although imprecise, these borrowings convey genuine information about the characteristics of an election, and they would be difficult to replace.

K. I. Macdonald (this Journal, vi (1976), 380–2) asserts, first, that my claim that the concept of power is essentially contested is ‘technically mistaken’ and, second, that the mistake is ‘substantively pernicious’. Unfortunately, he does not succeed in showing what the mistake is – nor indeed what makes it ‘technical’, rather than just a mistake. He does not even try to show why this alleged mistake is pernicious – let alone substantively so.

An attempt to rationalize policy outcomes in voting bodies has been made by Rae and Taylor (this Journal, 1 (1971), 71–90). In that work the assumption is made that all voters' preferences are representable by ‘city block’ type utility functions. The principal result obtained in that case by the writers is that for an odd number of voters there is always a unique equilibrium point under simple majority rule, independent of the distribution of individual optimal points and independent of the dimension of the policy space. Unfortunately, the Rae and Taylor result is not correct. In particular their result holds for a one or two dimensional policy space but not in cases where the policy space has three or more dimensions. The proof given by Rae and Taylor is correct for n = 2 but is false for n ≥ 3.