¹ Figures showing how shifts in support for the Labour and Conservative parties have become increasingly volatile not only at general elections, but in by-elections, local elections and the monthly opinion polls, are given in Crewe, Ivor, ‘Party Identification Theory and Political Change in Britain’, in Budge, Ian, Crewe, Ivor and Farlie, Dennis, eds., Party Identification and Beyond: Representations of Voting and Party Competition (London: Wiley, 1976), pp. 37–9.Google Scholar

² See Butler, David and Stokes, Donald, Political Change in Britain, 2nd edn. (London: Macmillan, 1975), pp. 181–90.Google Scholar The argument is that Labour is at an electoral disadvantage amongst those who entered the electorate before Labour became a major national party in the early 1920s (i.e. those aged over 50 in 1950) and should therefore have the electoral benefits from their death.

³ See, for example, Rose, Richard, The Problem of Party Government (London: Macmillan, 1975)- Chap. VGoogle Scholar; and the special Introduction to the third edition of Pulzer, Peter, Political Representation and Elections in Britain (London: Allen and Unwin, 1975).Google Scholar For press speculation on the future of the two-party system, see Beloff, Nora, Observer, 3 03 1974Google Scholar, and ‘A mandate for moderation’, leader in the Sunday Times, 3 03 1974.Google Scholar Close observers of British electoral trends had already detected a weakening of public commitment to the two main parties in the 1960s, but the stress tended to be on the downward drift of turnout and the growing volatility of their support rather than its absolute decline. See, for example, Butler, David and Stokes, Donald, Political Change in Britain, 1st edn. (London: Macmillan, 1969), pp. 121–2.Google Scholar It was also after 1974 that serious attention began to be turned to the desirability of a two-party system in Britain. See, in particular, Finer, S. E., ed., Adversary Politics and Electoral Reform (London: Anthony Wigram, 1975).Google Scholar

⁴ The number was also thirty-seven in 1945, but this included Members for the now defunct University seats.

⁵ By the conventional definition (the average of the change in the two parties' share of the vote) there was a 1·4 per cent swing to Labour; the two-party swing (the percentage change in the ratio of votes between the two parties) was 0·8 per cent to Labour. In both cases, the figure did not indicate an actual movement of that amount from Conservative to Labour, but simply reflected the fact that Labour's substantial loss of votes did not quite match that incurred by the Conservatives.

⁶ A further three minor party seats went to the Social Democratic and Labour party (Belfast West), the Social Democratic party (Lincoln) and an Independent Labour candidate (Blyth).

⁷ For more detailed accounts of the events leading up to the election, see King, Anthony, ‘The Election that Everyone Lost’, in Penniman, Howard R., Britain at the Polls: The Parliamentary Elections of 1974 (Washington, D.C.: American Enterprise Institute for Public Policy Research, 1975)Google Scholar; Butler, David and Kavanagh, Dennis, The British General Election of February 1974 (London: Macmillan, 1974), Chaps. 2 and 3CrossRefGoogle Scholar; and Fay, Stephen and Young, Hugo, The Fall of Heath (London: The Sunday Times, 1976).Google Scholar

⁸ The thirty-nine were made up of: thirteen Liberals, eleven SNP, ten Ulster United Unionists, three Plaid Cymru, one Social Democratic and Labour and one Independent (Fermanagh and South Tyrone).

⁹ See Key, V. O. Jr., ‘A Theory of Critical Elections’, Journal of Politics, XXI (1959), 198–210.Google Scholar One feature of Key's concept, however, is not appropriate to the British context: the mobilization of hitherto unpolitidzed social groups into regular voting.

¹⁰ Turnout in the U.K. rose from 72·0 to 78·1 per cent, the largest increase between two elections since 1945–50. If the figure is adjusted for the age of the register, the increase was from 75·1 to 79·1 per cent, still the sharpest since 1945–50. For the formula on which adjusted turnout is based, see Rose, Richard, Electoral Behavior: A Comparative Handbook (New York: Free Press, 1974), p. 494.Google Scholar It should be noted that V. O. Key probably identified ‘critical’ elections with more dramatic increases of turnout than that recorded in February 1974.

¹¹ ‘Accompanied by’ is used deliberately. Which trend caused the other, or which third factor caused both, received little attention and no analysis. Butler and Stokes described the weakening of the class alignment as ‘consistent with’ the growing volatility of support for the two parties (Political Change in Britain, p. 206).Google Scholar

¹² Butler, and Stokes, , Political Change in Britain, pp. 203–6.Google Scholar

¹³ An early view influenced by the Conservative election victories of the 1950s was that ‘affluent workers’ would turn to the Conservatives in increasing numbers. See, for example, Abrams, Mark and Rose, Richard, Must Labour Lose? (Harmondsworth, Middx.: Penguin Books, 1960).Google Scholar A later view was that the ‘affluent’ working class would not necessarily turn to the Conservatives but would make their Labour support less automatic. See Goldthorpe, John H. et al. , The Affluent Worker: Political Attitudes and Behaviour (Cambridge: Cambridge University Press, 1968).Google Scholar

¹⁴ For a more detailed account of the political and economic events of the early 1970s, see Butler, and Kavanagh, , The British General Election of February 1974, Chap. 2.Google Scholar

¹⁵ The gradient of the increase and the rapidity of the narrowing would be subject to considerable variation, of course, but so long as support for the two main parties rose with age, some increase and narrowing would take place.

¹⁶ And on the assumption that it increases – something that would have to be established independently of the age-cohort trends, i.e. by examination of the columns.

¹⁷ For a thorough discussion of the complexities involved in the attempt to distinguish age, period and cohort effects, see Mason, Karen O. et al. , ‘Some Methodological Issues in Cohort Analysis of Archival Data’, American Sociological Review, XXXVIII (1973), 242–58CrossRefGoogle Scholar, and Glenn, Norval D., ‘Some Cautions Concerning Statistical Attempts to Separate Age, Cohort and Period Effects’Google Scholar, paper presented to the 71st Annual Meeting of the American Political Science Association, San Francisco, September 1975; see also Abramson, Paul, ‘Generational Change and the Decline of Party Identification in America’, American Political Science Review, LXX (1976), 469–78.CrossRefGoogle Scholar

¹⁸ The wording was slightly varied in February and October 1974 to: ‘Would you call yourself a very strong Conservative (Liberal etc.), fairly strong or not very strong?’

¹⁹ For the sake of prose style we shall use the terms ‘party identification’ (‘party identifiers’) and ‘partisanship’ (‘partisans’) interchangeably. The terms partisan abandonment or partisan defection refer to a fall in the level or incidence of party identification; the term partisan weakening refers to a fall in its mean strength; and the term partisan decline denotes both in combination.

²⁰ Evidence in support of this argument for the Netherlands is presented in Thomassen, Jacques, ‘Party Identification as a Cross-National Concept: Its Meaning in the Netherlands’Google Scholar, in Budge, , Crewe, and Farlie, , eds., Party Identification and Beyond, pp. 63–80.Google Scholar See also the concluding remarks by Campbell, Angus and Valen, Henry in Campbell, Angus et al. , Elections and the Political Order (New York: Wiley, 1966), pp. 267–8.Google Scholar

²¹ We recognize that there are alternative ways of measuring long-term partisan predispositions, especially methods based on the social and demographic characteristics of voters, i.e. on attributes that are unambiguously antecedent to party choice in a way that party identification is not. See Budge, Ian and Farlie, Dennis, Voting and Party Competition: A Critique and Spatial Analysis of Existing Approaches Applied to Surveys from Ten Democracies (London: Wiley, forthcoming) especially Chaps, 1 and 2Google Scholar; also Budge, , Crewe, and Farlie, , eds., Party Identification and Beyond, pp. 13–17 and Chap. 6.Google Scholar

²² The meaning respondents attach to a concept, and to the questions designed to measure it, can, of course, change over time. From the validity testing we undertook it did not appear that our construct of party identification had changed meaning over the decade. In particular, the direction and magnitude of its relationship with a wide range of allied variables did not alter, e.g. the vote defection rates of groups defined by the direction and strength of their party identification remained more or less the same throughout the period. The meaning of party identification can also vary between electors, of course: for evidence that Conservatives are more prone to negative partisanship than Labour identifiers, see Crewe, Ivor, ‘Party Identification Theory and Political Change in Britain’Google Scholar, in Budge, , Crewe, and Farlie, , eds., Party Identification and Beyond, p. 52.Google Scholar

²³ The test reported in this paragraph is very similar to that conducted by Butler and Stokes over three earlier time points (1963–64–66). See Butler, David and Stokes, Donald, Political Change in Britain, 2nd edn. (London: Macmillan, 1974), p. 43.CrossRefGoogle Scholar They found, as we did, that the proportion with a stable party identification but variable voting record (8 per cent) was twice the proportion with a changing party identification but stable vote (4 per cent). But in contrast to our case they found that the dominant mode of instability was for both party identification and vote preference to change (13 per cent). However, Butler and Stokes do not give the proportion whose change of vote and party identification was in exactly the same direction-in our view, the crux of the matter.

²⁴ See Butler, and Stokes, , Political Change in Britain, 2nd edn., p. 45.Google Scholar

²⁵ The proportion of self-declared Independents amongst respondents of the University of Michigan SRC election surveys rose from 23 per cent in 1952 to 38 per cent in 1974. By 1974 they easily outnumbered Republican identifiers and were almost equal in number to Democratic identifiers. See Nie, Norman H., Verba, Sidney and Petrocik, John R., The Changing American Voter (Cambridge, Mass.: Harvard University Press, 1976), p. 49.Google Scholar

²⁶ Although this article is not primarily concerned with party comparisons it is worth noting that partisan strength was equally matched between the two parties at the beginning of the period but clearly in Labour's favour at the end.

²⁷ By the time of the Common Market referendum the proportion of ‘very strong’ identifiers in the two parties combined had slipped even further – from 31 per cent in October 1974 to 26 per cent in June 1975. Amongst Conservatives the figure stuck at 25 per cent; amongst Labour identifiers it fell from 35 to 26 per cent. Partisan strength tends to decline between general elections, and the absence of a clearcut party division at the referendum may have led to a further diminution of partisan strength, especially amongst supporters of the divided Labour party. But we might have expected Conservative partisanship to strengthen, given the near unity of the Conservative party in the campaign, the overwhelming ‘Yes’ vote of Conservative identifiers (90 per cent ‘Yes’ to 10 per cent ‘No’) and the explicit anti-Labour appeals made at the time by prominent Conservative politicians (on which see Butler, David and Kitzinger, Uwe, The 1975 Referendum (London: Macmillan, 1976), Chaps. 4 and 6).CrossRefGoogle Scholar Yet the proportion of ‘very strong’ Conservative identifiers was well below the 36 per cent recorded by Butler and Stokes in 1963 when the Conservative Government's fortunes were at an unusually low ebb and there was no election campaign to mobilize Conservative loyalties.

²⁸ Before the 1970 election, the voting age was lowered from 21 (effectively 21½) to 18, but on best estimates only 70 per cent of those aged 18–21 got onto the 1970 register (see Butler, and Stokes, , Political Change in Britain, 2nd edn., p. 240Google Scholar). Most of those not registering would have done so by 1974, thereby inflating the number of new electors to beyond that accountable by comings of age and immigration alone. The contribution of new electors to the decline of partisanship in February 1974 is dealt with later in the paper.

²⁹ The mean level of public satisfaction with the Labour government as registered in the monthly opinion polls fell from levels in the low to mid forties in 1964, 1965 and 1966 to 22·0 per cent (Gallup) and 24·8 per cent (NOP) in 1968 and 25·6 per cent (Gallup) and 26·3 per cent (NOP) in 1969. Levels of approval have not ebbed so low in the 1970s for either the Conservative or the present Labour Government. The years 1968 and 1969 also recorded the lowest levels of approval for the Prime Minister and the Leader of the Opposition (on which, see Rose, Richard, ‘Voters Show Their Scepticism of Politicians’, The Times, 9 04 1968Google Scholar). Individual membership of the Labour party also dwindled sharply in the late 1960s, on which see Forrester, Tom, The Labour Party and the Working Class (London: Heinemann, 1976), pp. 80–1.Google Scholar For evidence of the high incidence of public cynicism towards politicians in this period see, for example, the NOP bulletin for March 1968; for evidence that cynicism had subsided by 1972, see Crewe, Ivor, Two Cheers for Parliament: The Granada Survey of Public Attitudes to Parliament (London: Macmillan, forthcoming), Chap. 8.Google Scholar

³⁰ Trends are based on consecutive cross-sectional surveys, not a longitudinal survey. Interpretations of change for the period 1970–February 1974–October 1974 should therefore take into account the fall in the level of major party identification in February 1974. We would expect the remaining identifiers, all things being equal, tobe a little more ‘hardcore’ in their attitudes, and thus less dissatisfied or indifferent. An increase in recorded dissatisfaction/indifference in this period would underestimate the magnitude of the increase had nobody abandoned their major party identification after 1970; on the same assumptions stability would hide a small increase, and a small decrease would denote stability.

³¹ The wish to denationalize, however, did fluctuate a little, the proportions being 33 per cent in 1964, 38 per cent in 1966, 51 per cent in 1970 and 39 per cent in February 1974.

³² The distinctive principle of Conservative policy on the social services is not to cut expenditure but to make benefits selective rather than universal. However, since selective benefits would probably lead to savings in expenditure, we feel justified in counting opposition to further spending as a rough indicator of Conservative policy.

³³ The question on nationalization was not asked of the 1970 wave of the 1970–February 1974 panel. We do not examine the relationship between partisan decay and attitudes to the power of big business because the latter remained stable over the decade. The reader should therefore bear in mind that three of the four attitude items examined in Table 8 concern an aspect of trade union activities.

³⁴ The reader is reminded that the decline in support for Labour party policies amongst its identifiers was largely confined to the 1960s (except in the case of attitudes to strikers). Among the Labour identifiers of 1970 a minority did abandon their Labour policy position by February 1974; but a minority of similar size adopted a Labour policy position (as comparison of columns (1) and (2) in Table 8 makes clear). The most important feature of Table 8 is the large number of Labour identifiers who dissented from their party's principles at both elections-and the high rate of partisan weakening and abandonment amongst them.

³⁵ ‘Working class’ was defined as being a manual worker, or in a family whose head of household was a manual worker. Because membership of a trade union was part of the definition, the core group consisted predominantly of men under the age of 65. This adds rather than detracts to its ‘core’ quality.

³⁶ It is worth noting that amongst Labour identifiers as a whole we did not discover any significant variation in attitudinal change between ‘very’, ‘fairly’ and ‘not very strong’ identifiers.

³⁷ The number is three if the belief that big business has too much power (omitted from Table 9 because there was little change in this attitude over the decade) is included.

³⁸ The items are listed in Tables 12 and 13 in order of percentage difference for ease of reference. The items listed are drawn from a larger set of thirty-two items initially analysed, and reflect those yielding percentage differences of at least 13 per cent between those groups maximizing the discriminatory power of the independent variables. Discriminatory power was calculated as the ratio of between groups sum of squares to the total sum of squares of the dependent variable. This ratio was significant at the 5 per cent level in all cases listed in Tables 12 and 13.

³⁹ The extent to which a variable discriminated between those who weakened or abandoned their very strong identification and those who did not will also have depended a little on the number and ‘location’ of the variable's cutting points. A five-category variable is likely to contain at least one category which discriminates more strongly than either half of a dichotomous variable. Readers should therefore not place too faithful a reliance on the rank ordering of the variables in Tables 12 and 13. However, the fact that the categorizations of these variables ranged widely in number and location gives us confidence in the conclusions we have drawn in the text.

⁴⁰ Strictly speaking, with the length of time the same party identification has been held: see Campbell, Angus et al. , The American Voter, pp. 162–6Google Scholar; and Converse, Philip E., ‘Of Time and Partisan Stability’, Comparative Political Studies, II (1969), 139–71.CrossRefGoogle Scholar For the evidence on Britain, see Butler, and Stokes, , Political Change in Britain, 2nd edn., pp. 58–60.Google Scholar The French Fifth Republic is an important exception to the rule: for evidence that there it is the young who have the stronger and more stable partisanship see Cameron, D. M., ‘Stability and Change in Patterns of French Partisanship’, Public Opinion Quarterly, XVI (1972), 19–30CrossRefGoogle Scholar; also Inglehart, R. and Hochstein, A., ‘Alignment and Realignment in the Electorate in France and the United States’, Comparative Political Studies, V (1972), 343–72.CrossRefGoogle Scholar

⁴¹ An age-cohort or ‘generation’ is a group defined by its particular year(s) of birth. It therefore ages over time. Our table is constructed from a series of separate cross-sectional surveys, not a longitudinal panel survey, and therefore cannot trace the behaviour of the same respondents as they aged over time. It enables us to follow the progress of an age-cohort by providing data for an independent representative sample of the appropriate age category at successive elections.

⁴² We checked the alternative explanation that partisanship declined amongst the young because in this period Labour partisanship declined more than Conservative partisanship and the young tend to be Labour identifiers. We believe this to be at most only a small part of the explanation. For one thing, the partisan strength of Labour identifiers only declined by ·05 in the period 1964–70 (and that of Conservatives only rose by ·05); for another, separate examination of Conservative and Labour identifiers revealed similar age-related patterns in both.

⁴³ It is also worth noting how by 1974 partisanship no longer strengthened by leaps and bounds amongst electors between their mid-twenties and mid-forties. In 1964 partisan strength rose by ·25 between the age categories of 26–33 and 42–49; in 1966 by ·30; and in 1970 by ·33; but in both February and October 1974 by only ·13.

⁴⁴ Although the term ‘explanation’ is retained here, it is meant in a slightly looser sense than previously. The section that follows does not test hypotheses with the rigour that we have tried to apply in the preceding two sections, largely because the data available are inadequate to the task. We have therefore limited ourselves to the exploration of trends in the relationship between class and party, and to the suggestion rather than demonstration of the connection between these trends and those relating to partisanship over the same period.

⁴⁵ But because there are almost twice as many manual as non-manual workers, working-class Conservatives still outnumbered the Labour middle class.

⁴⁶ The manual/non-manual dichotomy might also be regarded as increasingly misleading, given the expanding overlap of status across the divide. The non-manual category grew by 5 per cent over the period, moreover, recruiting increasingly from the children (and wives) of working-class and thus mainly Labour families. We are satisfied, however, that the weakening of the class alignment and in particular the growth of the Labour middle classes cannot simply be attributed to the ‘proletarianization’ of the lower non-manual categories. Analysis not reported in this article suggests that the Conservative party lost proportionately more (although not necessarily numerically more) identifiers from the upper and middle echelons of the middle classes than from the ‘marginal’ middle class. A detailed analysis of the changing relationship between class and party is to be given in a following article.

⁴⁷ It might be argued that a Liberal identification amongst non-manual workers should count as class consonant since the Liberals are a ‘bourgeois’ party. On this adjustment the proportion of the middle classes with a class-consonant partisanship would have been (that for manual and non-manual combined is in parentheses) – 1964: 70 per cent (61 per cent); 1966: 71 per cent (63 per cent); 1970: 68 per cent (60 per cent); Feb. 1974: 68 per cent (62 per cent); Oct. 1974: 66 per cent (60 per cent). By this definition there was no discernable decline in class-consonant partisanship overall and only the mildest decline within the middle classes. And, of course, on this definition the level of class-consonant partisanship reaches a markedly higher level amongst non-manual as against manual workers.

⁴⁸ Of course, the proportion of the electorate with a Conservative or with a Labour partisanship is, in each case, less than 0·5. Knowledge of an elector's occupational status is, therefore, still superior to guessing as a predictor of his or her party identification.

⁴⁹ Butler, and Stokes, , Political Change in Britain, and edn., pp. 193–208.Google Scholar This is an enlarged revision of what was a short chapter section in the first edition.

⁵⁰ Nonetheless, the class alignment remained slightly stronger in the inter-war cohorts than in the rest of the electorate, even in October 1974. A further feature of interest in the cohort differences is the deviant pattern of the war-time and immediate post-war generations. Among those first entering the electorate between 1940 and 1955 the class alignment remained steady and relatively strong, despite its marked decline in the elactorate as a whole. We know from Butler and Stokes' analysis (see Political Change in Britain, 2nd edn., pp. 186, 202Google Scholar) that the class-party tie has been closer in this than any other living generation. But it should not be inferred that the experiences which shaped the class alignment in this generation were so deep-seated that the class alignment in this generation failed to weaken over the decade as a result of the ‘deviant’ behaviour of its middle-class component, especially between 1964 and 1970. Whereas the middle classes in the electorate as a whole shifted gradually away from the Conservatives and towards Labour over the decade, the middle classes in the 1940–55 generation moved in the opposite direction. It should be remembered that the Labour proportion of middle-class major-party suporters in the 1940–50 cohort was unusually high. It appears therefore that the effect of the 1964–70 Labour Government was to ‘break’ the exceptionally large middle-class Labour generation that emerged into the electorate during the Second World War and under the Attlee Government.

⁵¹ We have no proof that partisanship strengthened with age amongst the young generations of the past. Evidence is limited to the fact that at a single point in time partisanship hardens in successive age groups. But it is possible that this results, not from the age of older electors, but from the more partisan times in which they first entered the electorate. Butler, and Stokes, (Political Change in Britain, 2nd edn., pp. 58–60)Google Scholar, however, support the ‘ageing’ explanation on the grounds that it can be shown that what strengthens partisanship most is not the era in which an elector grew up but the duration of his party attachment.

⁵² See fn. 44.

⁵³ For evidence based on the Michigan SRC election studies since 1952 see Nie, , Verba, and Petrocik, , The Changing American Voter, p. 49Google Scholar; and Ascher, Herbert, Presidential Elections and American Politics (Homewood, III.: Dorsey, 1976), p. 70.Google Scholar For supporting evidence from AIPO and Gallup surveys see Ladd, Everett C. Jr, and Hadley, Charles, Transformations of the American Party System (New York: W. W. Norton, 1975), p. 292.Google Scholar In the United States, unlike Britain, the party identification question invites the respondent to describe himself as an Independent. It is unlikely, however, that this discrepancy in question wording explains the difference in the proportions of non-identifiers between the two countries. The concept of an ‘Independent’ is unfamiliar to British electors. Butler, and Stokes, (Political Change in Britain, 2nd edn., p. 44)Google Scholar draw attention to a British Gallup poll of August 1966 which found that, after explicit prompting, only 3 per cent of respondents called themselves ‘Independent’.

⁵⁴ Abramson, Paul R., ‘Generational Change and the Decline of Party Identification in America’, American Political Science Review, LXX (1976), pp. 469–78.CrossRefGoogle Scholar

⁵⁵ See Alt, James, Särlvik, Bo and Crewe, Ivor, ‘Partisanship and Policy Choice: Issue Preferences in the British Electorate, February 1974’, British Journal of Political Science, VI (1976), 273–90.CrossRefGoogle Scholar

⁵⁶ The best accounts of this are still to be found in Lazarsfeld, Paul F., Berelson, Bernard and Gaudet, Hazel, The People's Choice, 2nd edn. (New York: Columbia University Press, 1948)Google Scholar, and Berelson, Bernard R., Lazarsfeld, Paul F. and McPhee, William N., Voting (Chicago: University of Chicago Press, 1954).Google Scholar

⁵⁷ It is worth noting the greater frequency of landslide results in American presidential elections in this century than the last. See Ladd, and Handley, , Transformations of the American Party System, pp. 278–9.Google Scholar

⁵⁸ See Miller, William, ‘Nationalists Cut Across Class Division’, The Scotsman, 15 10 1975.Google Scholar

⁵⁹ See Miller, William, ‘The Connection Between SNP Voting and the Demand for Scottish Self-Government’, European Journal of Political Research (forthcoming).Google Scholar

⁶⁰ See Crewe, Ivor, Särlvik, Bo and Alt, James, ‘The Decline of the Two-Party System’Google Scholar, paper presented to the Annual Conference of the Political Studies Association, Oxford, 1975, Table VI. For detailed analysis of Liberal support see Alt, James, Crewe, Ivor and Särlvik, Bo, ‘Liberal Support in 1974’, Political Studies (forthcoming).Google Scholar

⁶¹ See Converse, Philip E. et al. , ‘Continuity and Change in American Politics: Parties and Issues in the 1968 Election’, American Political Science Review, LXIII (1969), 1083–1105, especially pp. 1101–5CrossRefGoogle Scholar; and Converse, Philip E. and Dupeux, Georges, ‘Politicization of the Electorate in France and United States’, Public Opinion Quarterly, XXVI (1962), 1–26.CrossRefGoogle Scholar

⁶² See, for example, Verba, Sidney and Nie, Norman H., Participation in America (New York: Harper and Row, 1972), Chap. 12Google Scholar; Milbrath, Lester W., Political Participation (Chicago: Rand, McNally, 1965), pp. 65–6Google Scholar; on the importance of partisan strength for turnout regularity in Britain, see Crewe, Ivor, Fox, Anthony and Alt, James, ‘Non-Voting in British General Elections 1966–October 1974’, in Crouch, Colin, ed., The British Political Sociology Yearbook, Vol. III (London: Croom Helm, 1977).Google Scholar

⁶³ Indeed, what little evidence exists suggests the contrary. A comparison of an NOP special survey on attitudes to politicians in February 1968 with identical questions in the Granada Survey of November 1972 (for a report on which, see Crewe, , Two Cheers for ParliamentGoogle Scholar) revealed a drop in levels of cynicism over the four years. It is easy to exaggerate levels of public satisfaction in the past. In 1944 the Gallup poll asked British electors whether they thought ‘British politicians are out merely for themselves, for their party or to do the best for the country’. The replies were: themselves, 36 per cent; party, 22 per cent; country, 36 per cent and don't know, 7 per cent. A similar question in the Granada Survey in 1972 produced a similar distribution of responses. Considering that 1944 was a year in which a popular war was being successfully prosecuted by a parliamentary government formed from an all-party coalition, the degree of public cynicism then about the patriotism of Britsh politicians seems remarkable. It is not impossible that suspicion of MPs and governmental institutions has been widespread amongst the ordinary public for a long period and only increased significantly in recent years amongst the small minority of the politically active and concerned.