¹ If Northern Ireland is included, the range is even greater. Turnout in Fermanagh and South Tyrone in 1970 was 92.1 per cent. Turnout figures throughout are derived from the following sources: Butler, D. E., The British General Election of 1955 (London: Macmillan, 1955)Google Scholar; Butler, D. E. and Rose, Richard, The British General Election of 1959 (London: Macmillan, 1960)Google Scholar; Butler, D. E. and King, A., The British General Election of 1964 (London: Macmillan, 1965)CrossRefGoogle Scholar; Butler, D. E. and King, A., The British General Election of 1966 (London: Macmillan, 1966)CrossRefGoogle Scholar; Butler, David and Pinto-duschinsky, Michael, The British General Election of 1970 (London: Macmillan, 1971)CrossRefGoogle Scholar; Boehm, K. and Mitchell, B. R., British Parliamentary Election Results 1950–1964 (Cambridge: Cambridge University Press, 1966).Google Scholar Discrepancies were resolved by reference to the following Returns of Election Expenses published as House of Commons Papers by HMSO: Nos. 141 of session 1955/6, 173 of session 1959/60, 220 of session 1964/5, 162 of session 1966/7, 305 of session 1970/71.

² See, however, Birch, A. H., ‘The Habit of Voting’, Manchester School of Economic and Social Studies, XVIII (1950), 75–83CrossRefGoogle Scholar, and Blondel, J., Voters, Parties and Leaders (Harmondsworth: Penguin, 1963), pp. 53–4.Google Scholar By contrast, variations in local election turnout have recently attracted attention. See Fletcher, P., ‘The Results Analysed’ in Sharpe, L. J., Voting in Cities (London: Macmillan, 1967), 290–328CrossRefGoogle Scholar; Fletcher, P., ‘An Explanation of Variations in “Turnout” in Local Elections’, Political Studies, XVII (1969), 495–502CrossRefGoogle Scholar; Newton, K., ‘Turnout and Marginality in Local Elections’ British Journal of Political Science, II (1972), 251–5.CrossRefGoogle Scholar

³ See for instance Budge, I. and Urwin, D. W., Scottish Political Behaviour (London: Longmans, 1966), Chap. 6Google Scholar; Sharpe, L. J., A Metropolis Votes (London: London School of Economics, 1962)Google Scholar; Bealey, F., Blondel, J.andMccann, W. P., Constituency Politics (London: Faber, 1965), pp. 228–35.Google Scholar

⁴ See the description of the Conservatives’ ‘critical seats’ exercise in the 1970 election in Butler, and Duschinsky, Pinto-, The British General Election of 1970, pp. 288–91.Google Scholar

⁵ Hereafter, references to the Nuffield studies will consist of the names of authors and the year of the election in question.

⁶ Butler and Rose, 1959, pp. 232–3.

⁷ Steed, M., ‘An Analysis of the Results’ in Butler and King, 1964, 337–59, p. 345.Google Scholar

⁸ Steed, in Butler and King, 1966, 271–95, pp. 284–6.Google Scholar

⁹ Steed, in Butler and King, 1966, p. 286.Google Scholar

¹⁰ Steed, in Butler and Pinto-Duschinsky, 1970, 386–415, p. 410.Google Scholar

¹¹ These are Cirencester and Tewkesbury, Cities of London and Westminster and Southampton, Itchen.

¹² Throughout the paper, correlation coefficients have been tested for significance using the ‘F’ test described in, for example, Anderson, T. R. and Zelditch, M., A Basic Course in Statistics, 2nd edn. (New York: Holt, Rinehart and Winston, 1968), p. 278.Google Scholar Where N=615, as in this case, then a coefficient greater than about .08 is statistically significant at the 5 per cent level.

¹³ See Bealey, , Blondel, and Mccann, , Constituency Politics, pp. 229–32Google Scholar and Birch, A. H., Small Town Politics (London: Oxford University Press, 1959), p. 106.Google Scholar

¹⁴ We employ both measures because although ‘non-manual workers’ is generally thought to approximate to ‘middle class’, there is some doubt about the true class position of some of the groups included in the former category. The percentage of professional and managerial workers gives an indication of the size of the relatively ‘pure’ middle-class group in a constituency. See Butler, David and Stokes, Donald, Political Change in Britain (London: Macmillan, 1969), pp. 66–73.Google Scholar

¹⁵ General Register Office, Sample Census 1966: United Kingdom General and Parliamentary Constituency Tables (HMSO, 1969)Google Scholar. ‘Non-manual’ is defined to include socio-economic groups 1, 2, 3, 4, 5, 6 and 13; ‘Professional and Managerial’ includes groups 1, 2, 3, 4 and 13.

¹⁶ Crewe, I. and Payne, C., ‘Analysing the Census Data’, in Butler, and Pinto-duschinsky, , 1970, 416–34, p. 425.Google Scholar

¹⁷ For both Labour and Conservativeseats, correlation coefficients of about 13 are significant at the 5 per cent level.

¹⁸ J. Rex, and R. Moore, , Race, Community and Conflict (London: Oxford University Press, 1967), PP.273–4.Google Scholar

¹⁹ Boehm, and Mitchell, , BritishParliamentary Election Results 1950–1964, p. vi.Google Scholar In constructing our measure, we defined urban areas to include, in the case of England and Wales, (i) boroughs or urban districts having a population of 5,000 or more, (ii) parishes in rural districts having a population of 5,000 or more and a density of more than 5 persons per acre; in the case of Scotland, (i) counties of city and large burghs, (ii) small burghs and other towns with a population of 5,000 or more. The figures refer to 1966 and were derived from various sources, namely: Municipal Year Book 1967 (London: Municipal Journal Ltd.);Google ScholarGeneral Register Office, Sample Census 1966, England and Wales, CountyReports (London: HMSO, 1967);Google ScholarGeneral Register Office, Edinburgh, Sample Census 1966, Scotland, County Reports (Edinburgh: HMSO, 1967);Google ScholarGeneral Register Office, Edinburgh, Place Names and Population: Scotland (Edinburgh: HMSO, 1967).Google Scholar

²⁰ See Steed, in Butler and Pinto-Duschinsky, 1970, p. 394.Google Scholar

²¹ Of course, where a constituency's electorate declined, its score on this variable is negative.

²² These are listed in fn. 1.

²³ We have included in this category all constituencies in Inner London, Glasgow, Birmingham, Liverpool, Manchester, Edinburgh, Bristol, Leeds, Sheffield, Bradford, Nottingham and Coventry. Correlation coefficients of about .18 are statistically significant at the 5 per cent level where N = 123.

²⁴ Butler, Steed and Duschinsky, Pinto, 1970, pp. 387–9.Google Scholar

²⁵ See, for instance, Butler, and Stokes, . Political Change, pp. 135–43Google Scholar; Pulzer, P., Political Representation and Elections in Britain (London: Allen and Unwin, 1967), pp. 110 ff.Google Scholar

²⁶ Butler and Pinto-Duschinsky, 1970, p. 355. Our regions differ slightly from those defined by Butler and Pinto-Duschinsky in that we have included the Isle of Wight and the Buckinghamshire constituencies in the South East and Lincoln in the East Midlands.

²⁷ The correlations needed for statistical significance at the 5 per cent level for each region are approximately as follows: Scotland, .23; Wales, .33; Inner London, .31; South East, .16; South West, .30; East Midlands, .33; West Midlands, .27; North West, .22; North, .32; East Anglia, .47; Yorkshire, .27.

²⁸ We have assumed that the relationships between turnout and the other variables is approximately linear. In the case of the class variables this might seem dubious in the light of our earlier discussion. However, a perusal of scatter diagrams suggests that the assumption of linearity is not unreasonable, and this avoids the complexities of non-linear regression analysis.

²⁹ See, for example, Sprent, P., Models in Regression (London: Methuen, 1969), Chap. 5.Google Scholar

³⁰ The significance of the coefficients was assessed by means of a ‘t’ test, and we eliminated variables whose coefficients were not significant at the 5 per cent level. Generally, a coefficient is significant at this level if it is at least twice as large as its standard error.

³¹ The equations we arrived at in this way are as follows: The symbols employed in the equations are as follows: TO, Turnout; OO, Percentage of Owner Occupiers; CT, Percentage of Council Tenants; NM, Percentage Non-Manual; PM, Percentage Professional and Managerial; MP, Minor-Party Vote; GE, Growth of Electorate; PC, Population Concentration; M, Marginality. R2 is the square of the multiple correlation coefficient, and is a measure of the proportion of variance in the dependent variable accounted for by the equation. The figures in brackets below the coefficients are their standard errors.

³² The new equations are as follows:

³³ Crewe and Payne in Butler and Pinto-Duschinsky, 1970, p. 425.

³⁴ There is one major exception to this consistent pattern. In the 1970 election, turnout in the three Stoke seats and in the constituencies of Leek and Newcastle-under-Lyme was very much lower than predicted due to the fact that the election coincided with local holidays.

³⁵ Campbell, A., Converse, P., Miller, W. and Stokes, D., The American Voter (New York: Wiley, 1960), p. 99.Google Scholar

³⁶ See Budge, and Urwin, , Scottish Political Behaviour, pp. 81–2;Google ScholarAbrams, M., ‘Social Trends and Electoral Behaviour’, British Journal of Sociology, XIII (1962), 228–42, pp. 232–8.CrossRefGoogle Scholar

³⁷ Bochel, J. M. and Denver, D. T., ‘Canvassing, Turnout and Party Support: An Experiment’, British Journal of Political Science, I (1971), 257–69CrossRefGoogle Scholar; Bochel, J. M. and Denver, D. T., ‘The Impact of the Campaign on the Results of Local Government Elections’, British Journal of Political Science, II (1972), 239–44.CrossRefGoogle Scholar

³⁸ Butler and Pinto-Duschinsky, 1970, Chap. II.

³⁹ Expenditure figures are derived from the House of Commons papers listed in fn. I.

⁴⁰ For another attempt to use campaign expenditure as an indicator of the strength of party organization, see Taylor, A. H., ‘The Effect of Party Organization: Correlation between Campaign Expenditure and Voting in the 1970 Election’, Political Studies, xx (1972), 329–31.CrossRefGoogle Scholar