This article is a contribution to the debate between ‘empirical’ and ‘classical’ theories of democracy. It draws attention to a hitherto neglected aspect of that debate, namely the historical process by which a word like ‘democracy’ gains its commendatory overtones. To call a state a democracy was not always to praise it; the argument here is that an understanding of how this came about can clarify some of the issues involved in considering whether or not states are properly to be called democracies. Although the methods used derive from linguistic philosophy, the purpose is to direct attention towards the values and aspirations of historical agents using the term, rather than to a purely conceptual analysis of it.

Within the last seven years there have been three important, but largely unnoticed, changes affecting the respective powers of the National Executive Committee (NEC) and the Constituency Labour Parties (CLPs). Assessment of these changes is not facilitated by the fact that they are central to an inner party dispute between left-wing and right-wing factions. The political rhetoric which has accompanied them is not a reliable indication of their true import and nature. This paper aims to describe and analyse the three changes and to assess them in terms of their impact on constitutional structures and their likely consequences for factional struggle – but not to make any value-judgements upon them or upon the inter-factional disputes.

This paper describes an empirical procedure for analysing the difference between the proportion of votes a major party attracts and the proportion of seats it subsequently wins in a plurality election. This difference will be referred to as the electoral bias with respect to that particular party at the election being analysed:

B = S — V, (1)

where S is the seat proportion and V the vote proportion gained by the party. This bias is the basic concept underlying much of the debate about electoral reform but curiously it has been an under-researched topic in political science literature. Apart from a small continuing debate over the cubic law of seat proportions and some analyses of Australian and New Zealand elections most attention in election studies has been concerned with determining the proportion of votes rather than the subsequent proportion of seats.

Arms races have been the subject of many and various explanations, but of few formal theories which manage to combine the virtues of plausibility and parsimony. Perhaps the best known is Richardson's, to which interesting modifications have been proposed by Smoker, Saaty and most recently Hamblin et al.2 Common to all these models is the assumption that arms races can be explained by an action-reaction process between the two nations involved.

A recent unusual event in the First Division of the English (Association) Football League provides an interesting example of how a game that is normally zero-sum can be converted into a non-zero-sum game. The political implications are, I hope, obvious. If politics consists of zero-sum games such as Diplomacy, Marx's conception of class conflict, or some non-Marxists' conceptions of ethnic, language, or boundary disputes, then no long-run co-operation between the players is possible and war is a seemingly inevitable continuation of policy by other means. If it consists of non-zero-sum games such as Prisoners' Dilemma, most non-Marxists' conceptions of class conflict, or some Marxists' conceptions of ethnic, language, or boundary disputes, then there is scope for co-operation as well as for competition. The present example, originally developed as an undergraduate teaching aid, may be of some general interest.

In a recent article, Patrick Dunleavy argues powerfully for an independent effect of ‘consumption locations’ on the political process in general, and voting patterns in particular, in Britain (‘The Urban Basis of Political Alignment: Social Class, Domestic Property Ownership and State Intervention in Consumption Processes’, this Journal, IX (1979), 409–44). Through an analysis of the housing and transport markets, Dunleavy suggests that people involved in ‘collective’ modes of consumption (such as council tenants and public transport users) are as a result of their own distinctive interests more likely to incline to the left than people involved in more ‘individual’ modes of consumption (such as home-owners and car-owners). Dunleavy suggests further that since consumption locations are at least partially independent of occupational class, the spread of home-ownership and car-ownership in the post-war period may help to account for the declining electoral influence of occupational class.

Politicians and political scientists have long known that there is a slight tendency among voters to prefer candidates whose names appear at the top of the ballot compared with lower-placed candidates, and hence that ceteris paribus the former have a somewhat better chance of being elected than the latter. When the candidates' names appear on the ballot in alphabetical order, this positional voting bias is usually called ‘alphabetic voting’. It is of special importance and interest in preferential voting systems where the voters may indicate their first, second, third, etc., preferences among a list of candidates, as in the single-transferable vote elections to the Irish Dáil, or where they must do so, as in the alternative-vote elections to the Australian House of Representatives.

Studies of the domestic sources of Israel's foreign policy have tended to treat Israel's parties as homogeneous blocs of opinion. Israel's political parties are implicitly arrayed in such analyses on a left-to-right continuum. This begins on the left with the New Communist List (RAKAH) and extends through the Israel Communist Party (MAKI), MOKED, Haolam Hazeh Koah Hadash, the Civil Rights Movement, SHELI, the United Workers Party (MAPAM), the Labour Party (formerly Ahdut Haavoda), Israel's Workers List (RAFI) and the Workers of Israel Party (MAPAI), the Independent Liberals, the Democratic Movement for Change, the National Religious Party, the Likud (composed of the state list, the Liberal Party, Laam, the Free Centre, the Greater Israel Movement, Shlomzion and Herut) and finally ends on the right with the ultra-orthodox Agudath Israel (AGI) and the Workers of Agudath Israel (PAGI). The left-to-centre part of this continuum is presented as moderate in varying degrees on questions relating to Israel's most cardinal external question, namely the terms of accommodation with the Arabs. The right-to-centre part of this same continuum is associated with hawkish views, which also vary in their degree of militancy.

Martin Harrop makes a large number of disparate observations connected with some aspects of my paper. In the space available I can do no more than take up those which seem particularly significant or contentious. Four points struck me in this way.

In their article – ‘Public Welfare Provision, Scandinavia, and the Sheer Futility of the Sociological Approach to Polities’ – Castles and McKinlay claim to have demonstrated that

two factors contribute to and explain the particularly high levels of public welfare provision in Scandinavian countries [compared to other advanced, industrial democratic states]. The absence of a large right-wing party establishes the initial necessary condition that removes the primary barrier to the development of welfare provision. The historic strength and unity of the working-class movement, reflected in the dominance of social democracy, is the factor which is then responsible for the development of public welfare to its current high level.

In an ingenious piece of aggregate data analysis Stan Taylor finds a significant relationship between the percentage of the coloured population in a constituency, and the percentage voting for the National Front amongst whites, in the two general elections of 1974 (‘The Incidence of Coloured Populations and Support for the National Front’, this Journal, IX (1979), 250–5). He tests two alternative models of this relationship. The first is a linear model which postulates a white ‘backlash’ of increasing magnitude with further concentrations of coloured people; the second is a curvilinear model which postulates maximum National Front support in areas of moderate concentrations of coloured people, the argument being that people in this situation may feel most threatened by the possible future influx of coloured people. It is an elegant exercise in model building and testing, which finds the non-linear model consistent with the data for the February 1974 election, and the linear model consistent with the data for the October 1974 election.

On one level, it is something of a pleasure when one's colleagues take sufficient interest in a paper to write a rejoinder. Furthermore, Roweth et al. are to be complimented on writing their rejoinder in a systematic manner. It seems only reasonable therefore, that we should respond to each of their charges, which they label our ‘shortcomings’, by following their format.

In their article, ‘Some Correlates of Regime Support in Canada’, (this Journal, VIII (1978), 199–216), Allan Kornberg, Harold D. Clarke and Lawrence LeDuc set out to describe ‘the distribution and foundations of public support for the political regime in Canada’ (p. 199). Using the 1974 National Election Study as their data base and regression analysis as their primary technique, they attempt to identify the correlates of regime support and come to grips with the problem of causal sequence.

In a disingenuous piece of aggregate data analysis Paul Whiteley suggests that, because he found that the correlation between two variables in a local election held at one point of time in one city was spurious when a third was controlled for, this implies that the relationships between two variables, one calculated on a different basis, in two general elections which took place three years earlier and covered a wide area of England, will also be spurious when an additional variable is introduced into the analysis. It is an inelegant analysis in model building and testing, in so far as it cannot explain the findings it purports to explain. Unfortunately it is also a textbook case of making causal inferences from variables associated with each other for reasons that lie outside the scope of the model.

In their commentary on our paper, Atkinson, Coleman and Lewis, although acknowledging that ‘studies of political support are particularly appropriate in Canada’, are very critical of our effort for two principal reasons. Firstly, they argue that we have presented no compelling evidence that citizens are unable or unwilling to distinguish sharply between regime and the party in power. On the contrary, they note that Shanks and Citrin have concluded that ordinary American citizens routinely distinguish between regime and authorities. Secondly, they claim that rather than measuring regime support, the dependent variable is simply measuring partisan orientations toward the governing Liberal party and its leaders (e.g. ‘evaluation of incumbent authorities’ (p. 403); ‘the party in power and not the regime’ (p. 403); ‘the incumbent Liberal authorities’ (p. 404); and ‘the dependent variable is measuring feelings toward a Liberal party government and not the regime’ (p. 409)).

I heartily approve of research based upon poll data, in which I include specifically the contribution by Paul Whiteley, ‘Electoral Forecasting from Poll Data: The British Case’ (this Journal, IX (1979), 219–36). I note with pleasure the fact that the author has relied heavily upon Gallup data for the testing of his time-series model. There is one specific objection that I have to the detail of the contents of this paper. I refer to the sentence bridging pages 231–2, in which he states: ‘the large inaccuracy of the 1951 forecast is not, in fact, due to the model so much as Gallup data, which were extremely inaccurate in that year’. There is a footnote which states that this was inferred from the residuals of the forecasting model. The error quoted is 8·3 per cent. This does refer to the last pre-campaign poll, admittedly. The last campaign poll, the only one that could be compared with the result itself, shows, according to Gallup records, a deviation of 2·2 per cent. The bland assertion of Mr Whiteley that our results were inaccurate in that year is totally unsupported by any evidence of a practical kind. Instead we have the statement that the fit of the model was rather bad on this occasion. Remembering that public opinion, including support for a specific political party, is known to change radically between elections, between local elections and by-elections, and so on (I confine myself here to actual elections – I do not depend upon poll data to substantiate my case) and knowing that among the causes of this are political events, and events of a social and economic nature which affect the mood and attitude of the electorate, it does not surprise me that a purely mathematical model does not necessarily fit from time to time. Indeed elsewhere in the article the author mentions the possibility of shocks affecting public opinion and the degree to which these shocks are effected in the auto-regressive scheme that he has put forward.