¹ Goldthorpe, John H., Lockwood, David, Bechhofer, Frank and Platt, Jennifer, The Affluent Worker: Political Attitudes and Behaviour (Cambridge: Cambridge University Press, 1968)Google Scholar, henceforth abbreviated to PAB; and, by the same authors, The Affluent Worker in the Class Structure (Cambridge: Cambridge University Press, 1969)Google Scholar, henceforth abbreviated to CS. A third volume on industrial attitudes and behaviour was also published.

² Or, as the authors put it, ‘Because of the nature of the social structure in which he lives, the worker is seldom exposed to influences that might alter his political convictions; on the contrary, everything in his social environment at work and at leisure, combines to render them beyond question. Party allegiance is less a conscious and deliberate choice than a by-product of the more primitive forms of collectivism in which he is involved in the workplace, in the local union branch, in the working men's club, and, at an even deeper level, in the communal sociability of everyday life. These forms of social relationship make his support for Labour the “natural” or “instinctive” one…’ PAB, p. 75.

³ PAB, pp. 79–80.

⁴ The authors placed considerable emphasis, for instance, on the answer of one respondent, who was regarded as peculiarly typical of the sample as whole, and had never voted other than Labour, but who said ‘I’d be the first to vote Tory though, if a Labour Government started mucking us about’ PAB, p. 18.

⁵ For example, Dennis, N. et al. , Coal is our Life (London: Eyre & Spottiswoode, 1956)Google Scholar; Young, M. and Willmott, P., Family and Kinship in East London (London: Routledge and Kegan Paul, 1957: University of Liverpool Press, 1954)Google Scholar; Tunstall, J., The Fishermen (London: McGibbon and Kee, 1962)Google Scholar; and Klein, J., Samples of British Culture, Vol. I (London: Routledge and Kegan Paul, 1965)Google Scholar, which summarizes many of the previous community studies.

⁶ See Wedderburn's, Dorothy review of The Affluent Worker: Industrial Attitudes and Behaviour in New Society, 25 06 1968.Google Scholar

⁷ See Brown, Richard and Brennen, Peter, ‘Social Relations and Social Perspectives amongst Shipbuilding Workers — a Preliminary Statement’, Sociology, IV (1970), pp. 71–84, and pp. 197–212.CrossRefGoogle Scholar Wallsend is both economically and culturally dominated by the one ‘heavy’ and traditional industry of shipbuilding; has a homogeneously working class social structure; a stable population with a collective experience and memory of unemployment and deprivation; and close-knit networks of kin, friends and neighbours. It is therefore difficult to think of Wallsend as anything other than a traditional working class community.

⁸ The study produced no data on political attitudes. But similar proportions of both Wallsend and Luton workers held favourable attitudes towards top management, possessed high job and educational aspirations for their children, and subscribed to complex prestige models of the social structure.

⁹ Two arguments in defence of Goldthorpe et al. can be made here. One is that the ‘traditional working class’ is an ideal-type only, based more on plausibly and consistently related sociological categories than on any existing working class community. But to this it could be replied that the same is true of the ‘affluent working class’ community. The second is that the only working class communities which sufficiently resemble the traditional working class model existed in the past (such as South Wales mining valleys early this century, or Lancashire mill towns in the nineteenth century) and no contemporary working class community will do. But the studies which form the source of the model constructed by Goldthorpe et al. were themselves conducted in the 1950s; and at one point the authors state ‘in the post war period distinctive elements of working class traditionalism still persist in various parts of the country’ (p. 74).

¹⁰ See Butler, David and Freeman, Jennie, British Political Facts 1900–68, 3rd ed. (London: Macmillan, 1969) for evidence on opinion polls (p. 162) and local government elections (p. 243)Google Scholar and Butler, David and King, Anthony, The British General Election of 1964 (London: Macmillan 1965), pp. 332–5CrossRefGoogle Scholar for the figures on by-elections.

¹¹ PAB, Table 4, p. 13.

¹² Against this point it might be argued that, owing to the well-known tendency for some respondents to project their present voting intention incorrectly onto their recalled voting in the past, the reported level of Labour support in 1955 and 1959 is an overestimate, and thus masks what was really a small rise in Labour support between 1955 and 1963. But in that case it would no longer be safe to claim that the sample's vote for Labour exceeded the national average for manual workers, which is an important part of Goldthorpe et al.'s objection to the embourgeoisement theory. Thus whatever interpretation is placed on this set of figures, they do not constitute a convincing empirical refutation of the embourgeoisement theory.

¹³ See Butler, David and Stokes, Donald, Political Change in Britain (London: Macmillan 1969), pp. 55–6Google Scholar, and Campbell, Angus et al. , The American Voter (New York: John Wiley and Sons Inc., 1960), p. 126.Google Scholar

¹⁴ In this context it is interesting to note that studies of working class kinship patterns have shown how rehousing from a traditional slum to new estates may produce only a temporary deviation of about one generation's duration from conventional kinship behaviour. See, for instance, Willmott, P., The Evolution of a Community: A Study of Dagenham after Forty Years (London: Routledge and Kegan Paul, 1963).Google Scholar

¹⁵ See Butler and Stokes, Political Change in Britain, Chap. 7; Ingham, Geoffrey, ‘Plant Size: Political Attitudes and Behaviour’, Sociological Review, XVII (1969), pp. 235–50CrossRefGoogle Scholar, and Nordlinger, Eric A., The Working Class Tories (London: MacGibbon & Kee, 1967), pp. 205–9.Google Scholar

¹⁶ Census 1966 United Kingdom General and Parliamentary Constituencies (London: HMSO, 1969)Google Scholar, a more detailed description of the information it contains, and of its uses for the analysis of voting behaviour can be found in Ivor Crewe and Clive Payne, ‘Analysing the Census Data’ in Butler, David and Pinto-dushinsky, Michael, The British General Election of 1970 (London: Macmillan, 1971).CrossRefGoogle Scholar

¹⁷ A subsequent check showed that the inclusion of Welsh and Scottish constituencies would have made no difference to the final list of affluent working class seats: ‘affluence’ turned out to be a strictly English phenomenon and no seat in Scotland or Wales even approached eligibility for the AW group.

¹⁸ Of course, it is possible that car ownership among affluent industrial workers gives some the opportunity and incentive to live in rural areas and commute to work. But whilst rural and semirural constituencies probably contain ‘affluent workers’ among their electorate, it is difficult to think of any that contain affluent working class communities as described by Goldthorpe et al.

¹⁹ Crewe, and Payne, , ‘Analysing the Census Data’, pp. 435–6.Google Scholar

²⁰ See Lockwood, David, ‘Sources of Variation in Working Class Images of Society’, Sociological Review. XIV (1966), 249–67CrossRefGoogle Scholar, in which he distinguishes between the ‘traditional proletarian’, ‘traditional deferential’ and ‘privatized’ social consciousness of different working class communities.

²¹ The correlation coefficient between the two lists of residuals scores was 0.89 which can be considered remarkably high.

²² The admittedly arbitrary figure of 100 was adopted simply for convenience. But as James Alt correctly pointed out in his commentary on an earlier draft of this paper, such arbitrariness could and should be avoided by basing the cut-off point on the same measure of standard deviation for both lists of residual scores. This would cope with the possibility that the two lists contained very different distributions of scores. In fact both lists were revealed in a subsequent check to consist of near-perfect normal distributions; and the cut-off point of 100 was equivalent in both cases to a 0.6 standard deviation.

²³ I.e., the Registrar General's socio-economic groups 1, 2, 3,4 and 12. These do not include routine, junior or unqualified non-manual workers.

²⁴ A further test of the validity of the AW group was applied. The Census contains figures on the number of skilled manual workers, and the extent of recent migration for each constituency. It seemed plausible that in comparison with both traditional working class constituencies and the country at large, the AW group would contain (i) a higher ratio of skilled to semi-skilled and unskilled manual workers, and (ii) a higher proportion of residents who had moved into the constituency within the previous five years. The figures below show that this is indeed the case:

²⁵ Bassetlaw, Newark, Morpeth and Durham North West were therefore excluded because over 3 per cent of their labour force was employed in agriculture.

²⁶ A similar description to that of Lockwood, 's in ‘Sources of Variation in Working Class Images of Society’, can be found in CS, pp. 118–20Google Scholar and PAB, pp. 74–6.

²⁷ See Dennis et al., Coal is our Life, and Brennan, T., Cooney, E. W. and Pollins, H., Social Change in South West Wales (London: Watts, 1954).Google Scholar

²⁸ Finding a valid method for locating traditional working class communities which escaped both tautology and arbitrary subjectivity proved troublesome. To select constituencies from the oldest industrial areas would have immediately taken one to the textile belt of Lancashire and the West Riding which on a number of counts would have been inappropriate: there is an exceptionally high degree of working class home-ownership and female employment (but little ‘affluence’) in this part of the country. City-centre constituencies were unsuitable because, as mentioned above, they tend to have a recent history of immigrant settlement and depopulation, which gives them a distinctive social and political character, but one very different from that intended in Lockwood's model of the traditional working class community (see Hindess, Barry, The Decline of Working Class Politics (London: McGibbon & Kee, 1971)).Google Scholar It was also too difficult to choose a group on the basis of personal intuition: problems of definition and distinction immediately arose (e.g., whether constituencies containing large pre-war council estates, like Dagenham or Manchester Wythenshawe, should be included; or whether isolated railway towns like Crewe, Darlington or Carlisle should be included). What one notices so strongly from combing through the Census data is the enormous variety of local social structures that exist in contemporary Britain.

²⁹ See Butler, and Stokes, , Political Change in Britain, pp. 468–74Google Scholar for a good discussion of this point.

³⁰ Ample evidence of this is given in the Nuffield election studies.

³¹ PAB, p. 73.

³² Henceforth referred to simply as ‘English constituencies’, ‘comparable constituencies’ or ‘the average’ for convenience. The figure of 268 was arrived at after eliminating from the group of 361 urban English constituencies all those (93) with more than 40 per cent non-manual workers.

³³ As Berrington, H. first pointed out in ‘The General Election of 1964’, Journal of the Royal Statistical Society, Series A, 128 (1965)Google Scholar, the problem with the use of swing as a measure is that it can simply reflect the widely varying distributions of partisanship in constituencies. In 1955 the mean Labour share of the two-party vote was 73.9 per cent in traditional working class constituencies and 52.6 per cent in affluent working class constituencies. Other things (such as turnout and the presence of minor parties) being equal, one would expect a uniform national swing to the Conservatives to produce a higher pro-Conservative swing in the TW group than in the AW group because in the former there are relatively more Labour voters ‘at risk’. Similarly a uniform national swing to Labour should produce the higher pro-Labour swing in the AW group where there are relatively more previous Conservative voters vulnerable to defection. Table 2 shows that the reverse occurred in 1959 and 1970 and it therefore seems safe to assume that differences in mean swing between the two groups in these two years were substantively significant. An artefact of this type, however, may have had a slight influence on the mean swing differentials in 1964 and 1966 but, given their magnitude, not enough to alter their direction. Only the very small difference in the mean long term swing between 1955 and 1970 should be treated with caution as a result of the disparate vulnerability of the two groups to defection by Labour supporters.

³⁴ PAB, p. 80.

³⁵ Of course economic trends between elections varied amongst the regions. For example, relative prosperity continued in the Midlands and the South East between 1959 and 1964 whereas the North East suffered from a local recession, and this differential in economic growth (which is clearly related to, but is not the same as, more enduring differences in the level of affluence) may have influenced the differences in swing between the AW and TW constituencies. But we should then expect those traditional working class seats located in the North East to behave differently from those unaffected by a recession (such as those in the West Riding) and this did not appear to be the case.

³⁶ This result may be partly due to the marginality of many of the AW constituencies. The eight marginal seats (Bedfordshire South, Birmingham Perry Barr, Bristol North East, Coventry South, Derbyshire South East, Gloucester, Luton and Wellingborough) had an average residual score of 5·7 per cent; this compares with 1·2 per cent for the seven remaining safe seats. But marginality had no apparent effect on turnout change between 1966 and 1970, which was –5·2 per cent for the marginal seats and –5·4 per cent for the safe seats. And the fact that the group of safe seats still had a positive mean residual score suggests that the fourth hypothesis should still be rejected.

³⁷ For an earlier attempt to locate proletarian and traditional working class communities by a similar method, see Piepe, Anthony, Prior, Robin and Box, Arthur, ‘The Location of the Proletarian and the Deferential Worker’, Sociology, III (1969), pp. 239–44.CrossRefGoogle Scholar Their analysis, however, consisted of a very elementary form of residuals analysis which, without computing a linear regression equation, merely looked at the variations in the per cent Labour vote (standardized by age and sex) of communities (based on constituencies) with a similar proportion of manual workers (based on figures for boroughs). Moreover, their analysis was unsystematic; their residuals were unranked; and their demographic figures were taken from the 1951 Census.