¹ See, for example, Butler, David and Stokes, Donald, Political Change in Britain (Harmonds-worth, Middx.: Penguin, 1971), pp. 509–12.CrossRefGoogle Scholar

² McAllister, Ian and Rose, Richard, The Nationwide Competition for Votes: The 1983 British Election (London: Frances Pinter, 1984).Google Scholar

³ A similar multiple regression analysis which provides figures only for the Alliance vote is contained in Curtice, John and Steed, Michael, One in Four (London: Association of Liberal Councillors, 1983).Google Scholar

⁴ For a full description see McAllister, and Rose, , Nationwide Competition for VotesGoogle Scholar, Appendix B. The relative importance of each factor is discussed and described as follows: ‘The order in which the factors emerged from the analysis does not indicate that one is more important than the other, for example, that socio-economic status is a more important social structural measure than immigrants; on the contrary, it merely reflects the number of variables included in the final solution which loaded on each factor. The first factor, with an eigenvalue of 4.65, explained 29 percent of the variance, while the second, immigrants, had an eigenvalue of 3.73 and explained 23 percent of the variance. The remaining two factors, the elderly and agriculture, had eigenvalues of 3.00 and 1.89 and explained 19 percent and 12 percent of the variance respectively.’

⁵ McAllister, and Rose, , Nationwide Competition for Votes, p. 176.Google Scholar

⁶ The full list is as follows: Shore/Hattersley 11, Shore/Dunwoody 7, Shore/Davies 3, Hattersley/Hattersley 50, Hattersley/Davies 2, Hattersley/Dunwoody 1, Kinnock/Hattersley 50, Kinnock/Davies 15, Kinnock/No vote 3, Kinnock/Meacher 32, Heffer/Meacher 27, Heffer/Davies 1, Heffer/No vote 1, No vote/Davies 1, No vote/Hattersley 1, No vote/No vote 4.

⁷ When the population is normal (or in this case approximately normal) and when the standard deviation of the population is known, it is possible to make use of a simple Z-test rather than either the t-test that is used in estimation exercises or any of its non-parametric counterparts. The value of Z largely depends on two variables, first the size of the difference between the population and sample means, second the size of the sample. It is essentially the difference between the sample mean and the population mean expressed in units of the standard deviation of the population. See Blalock, H. M., Social Statistics, revised 2nd edn (Tokyo: McGraw-Hill, 1979), pp. 96–101.Google Scholar

⁸ Further control variables could, of course, be introduced but examination of the revised data suggests nothing of quantitative importance. Deselected Labour MPs opposed Labour as Independent Labour candidates in only two of the sixty-one seats won by Labour left-wingers. In general Labour right-wingers have served in Parliament longer than left-wingers, 54 per cent of whom were first elected at or after the 1979 general election.

⁹ The regional figures come from McAllister, and Rose, , Nationwide Competition for VotesGoogle Scholar, Appendix A. The residual mean for Labour candidates nationwide was 0.4 which compares with a figure of – 8.2 for the North of Scotland and +5.1 for the North of England. The adjustment equalizes the residual means of all ten British regions (McAllister and Rose definition). The principle involved is the same as that which underlies the analysis of variance. It constructs the distribution of residuals that would have occurred if all candidates had been exposed to the same regional influences. See Blalock, , Social Statistics, pp. 354–5.Google Scholar

¹⁰ As before, the adjustment equalizes the means of the relevant different groupings – in this case the candidates who were opposed by SDP defectors and the candidates who did not face such opposition. We add 3.7 per cent to the (already regionally adjusted) residual scores of all members of the former group and subtract 0.2 per cent from the scores of the latter.

¹¹ If we turn length of parliamentary service into a dichotomized nominal scale and compare the mean residual scores of MPs elected before the 1979 General Election with those elected at or after it, we find a difference of 2.2 on the unadjusted residuals but one of only 0.1 on the adjusted scores. ‘Time first-elected’ clearly correlates with ‘region’.

¹² These were: Militant – Pat Wall (Bradford North); Terry Fields (Liverpool Broadgreen); Rod Fitch (Brighton Kemptown); Dave Nellist (Coventry South East); Cathy Wilson (Isle of Wight); loosely associated with Socialist Organizer Alliance – Jeremy Corbyn (Islington North); Valerie Veness (Hornsey and Wood Green); Paul Boateng (Hertfordshire West); Harry Cohen (Leyton); Chris Smith (Islington South and Finsbury); Audrey Wise (Woolwich); Tony Banks (Newham NW); John Denham (Southampton Itchen); Alan Whitehead (Southampton Test).

¹³ Ben Ford (Ind. Lab.) fought Bradford North against Pal Wall. Arthur Lewis (Ind. Lab.) contested Newham North West against Tony Banks. Michael O'Halloran (Ind. Lab.) and ex-SDP stood against Jeremy Corbyn in Islington North.

¹⁴ My thanks to Linda Anderson at the Political Research Unit for allowing me to use the results of this survey.