¹ Merton, R. K., ‘The Self-Fulfilling Prophecy’, Antioch Review, VIII (1948), 193–210CrossRefGoogle Scholar; Henshel, R. L., ‘The Boundary of Self-Fulfilling Prophecy’, British Journal of Sociology, XXXIII (1982), 511–28.CrossRefGoogle Scholar

² Simon, H. A., ‘Bandwagon and Underdog Effects in Election Prediction’, in Models of Man: Social and Rational (New York: Wiley, 1957), pp. 79–87Google Scholar; Gartner, M., ‘Endogenous Bandwagon and Underdog Effects’, Public Choice, XXV (1976), 83–139CrossRefGoogle Scholar; Straffin, P. D., ‘The Bandwagon Curve’, American Journal of Political Science, XXI (1977), 695–709Google Scholar; Zech, C. E., ‘Leibenstein's Bandwagon Effect as Applied to Voting’, Public Choice, XXI (1975), 117–22.CrossRefGoogle Scholar

³ Dizney, H. F. and Roskens, R. W., ‘An Investigation of the “Bandwagon Effect” in a College Straw Election’, Journal of Educational Sociology, XXXVI (1962), 108–14CrossRefGoogle Scholar; Navazio, R., ‘An Experimental Approach to Bandwagon Research’, Public Opinion Quarterly, XLI (1977), 217–25CrossRefGoogle Scholar; Cantril, A. H., ed., Polling on the Issues (Cabin John, Md.: Seven Locks Press, 1980)Google Scholar; Tyson, J. L. and Kaplowitz, S. A., ‘Attitudinal Conformity and Anonymity’, Public Opinion Quarterly, XXXV (1977), 226–34.CrossRefGoogle Scholar

⁴ Lang, K. and Lang, G., Voting and Nonvoting: Implications of Broadcasting Returns before the Polls are Closed (Waltham, Mass.: Blaisdell, 1968)Google Scholar; Mendelsohn, H., ‘Election Day Broadcasts and Terminal Voting Decisions’, Public Opinion Quarterly, XXX (1966), 212–25CrossRefGoogle Scholar; Fuchs, D., ‘Election Day Radio-Television and Western Voting’, Public Opinion Quarterly, XXX (1966), 226–236CrossRefGoogle Scholar; Tuchman, S. and Coffin, T. E., ‘The Influence of Election Night Television Broadcasts in a Close Election’, Public Opinion Quarterly, XXXV (1972), 315–26.Google Scholar

⁵ Laponce, J. A., ‘An Experimental Method to Measure the Tendency to Equibalance in a Political System’, American Political Science Review, LX (1966), 434–8Google Scholar; Fleitas, D. W., ‘Bandwagon and Underdog Effects in Minimal Information Elections’, American Political Science Review, LXV (1971), 434–8.CrossRefGoogle Scholar

⁶ Rothman, in Teer, F. and Spence, J. D., Political Opinion Polls (London: Hutchinson University Library, 1973), pp. 131–2Google Scholar; Gaskell, G., ‘Polls and the Voters’, New Society, 4 04 1974, pp. 23–4.Google Scholar

⁷ Ceci, S. J. and Cain, E. L., ‘Jumping on the Bandwagon with the Underdog: the Impact of Attitude Polls on Polling Behaviour’, Public Opinion Quarterly, XLVI (1982), 228–42.CrossRefGoogle Scholar

⁸ Beniger, J. R., ‘Winning the Presidential Nomination: National Polls and State Primary Elections, 1937–1972’, Public Opinion Quarterly, XL (1976), 22–38.CrossRefGoogle Scholar

⁹ See Marsh, Catherine, ‘Do Polls Affect What People Think?’ in Turner, C. F. and Martin, E., eds, Survey Measurement of Subjective Phenomena, Report of the Panel on Survey Measurement of Subjective Phenomena, National Research Council, National Academy of Sciences, 2 vols. (New York: Sage, 1983)Google Scholar

¹⁰ Gollob, H. R., Rossman, B. B. and Abelson, R. P., ‘Social Inference as a Function of the Number of Instances and Consistency of the Information Presented’, Journal of Personality and Social Psychology, XXVII (1973), 19–33.CrossRefGoogle Scholar

¹¹ Marsh, Catherine and Calderbank, David, ‘Attitudes to Abortion 1962–1982’ (mimeograph, Social and Political Sciences Committee, University of Cambridge, 1983).Google Scholar

¹² In a pilot of seventy-one individuals we asked the interviewers to record if the experimental information was ever challenged, and it was on only one occasion.

¹³ Marsh, Catherine, ‘Judging Bias in Question Wording: A Research Note’ (mimeograph, Social and Political Sciences Committee, University of Cambridge, 1982).Google Scholar

¹⁴ ‘Do you think on balance that most people want the law left as it is now, or do they think that abortion should be easier to get, or do they think it should be more difficult to get?’

¹⁵ ‘In which direction do you think public opinion will move in the future? Will public opinion not change in the future, or will it move towards making abortion easier to get, or towards making it more difficult to get?’

¹⁶ We had an interesting degree of difficulty in the pilot study in devising a suitable measure of strength of feeling. We included a question after the main abortion question, saying simply: ‘and how strongly do you feel about that?’ The pilot interviewers reported that many people replied, ‘About what?’ Even with respondents who immediately gave a reply, many of them seemed to be answering a very vague question: ‘How strong are your views on abortion?’ The question was therefore modified in the main survey to read ‘How strongly do you feel that the law on abortion should be … left as it stands … the law should be altered (etc.)?’ This worked much better.

¹⁷ ‘In fact most people want abortion to be made easier/harder to get.’

¹⁸ ‘In fact the trend in recent polls has been towards making abortion easier/harder to get.’

¹⁹ ‘Abortion should be legally available for anyone who wants it’ – ‘Strongly agree … agree … neither agree nor disagree … disagree … disagree strongly.’

²⁰ Kruskal, W. and Mosteller, F., ‘Representative Sampling I, II and III’, International Statistical Review, XLVII (1979), 13–24, 111–27, 245–65.CrossRefGoogle Scholar

²¹ Note that, since survey variables correlate with one another and are not independent, we would not, pace Selvin and Stuart (Selvin, H. C. and Stuart, A., ‘Data-dredging Procedures in Survey Analysis’, American Statistician, XX (1966), 20–3)Google Scholar, expect to find one in ten producing differences significant at the 0·1 level across randomized conditions.

²² Bowen, Jenifer, ‘A Survey of the General Public's Attitudes to Market Research’, Journal of the Market Research Society, XXI (1979), 75–102.Google Scholar

²³ NOP Market Research Ltd., NOP Political Bulletin, No. 124 (1974a)Google Scholar; No. 131 (1974b); Market and Opinion Research International, British Public Opinion: General Election 1983: Final Report (MORI, 32 Old Queen Street, London SW1H 9HP).Google Scholar

²⁴ It has been estimated that there is, on average, at least one poll finding in every newspaper issued in Britain and the USA (Turner, and Martin, , Survey Measurement of Objective PhenomenaGoogle Scholar, Chap. 2.). MORI calculates that 4 per cent of the front page news coverage in the 1983 general election was devoted to opinion polls (Market and Opinion Research International Ltd. General Election 1983Google Scholar, Appendix 1).

²⁵ The percentages have to be run in this direction since the perception variable was manipulated among subgroups on the assumption that the direction of cause was from perception of others to own views.

²⁶ I am very grateful to NOP Market Research Ltd for making this available.

²⁷ Marsh, and Calderbank, , ‘Attitudes to Abortion 1962–1982’.Google Scholar

²⁸ Noëlle-Neumann, E., ‘The Spiral of Silence: A Theory of Public Opinion’, Journal of Communication, XXXIV (1974), 43–51.CrossRefGoogle Scholar

²⁹ Schuman, H. and Presser, S., Questions and Answers in Attitude Surveys (New York: Academic Press, 1981).Google Scholar

³⁰ Schuman, and Presser, , Questions and Answers in Attitude Surveys, Chap. 9.Google Scholar

³¹ When there are only three categories, the largest percentage difference is also the index of dissimilarity.

³² Lazarsfeld, P. F., Berelson, B. R., and Gaudet, H., The People's Choice (New York: Duell, Sloan and Pearce, 1944).Google Scholar

³³ Revelations, 3: 15.Google Scholar

³⁴ The design of four experimental groups was not drawn up for the experiment on abortion attitudes, and this was the most efficient way to use the four groups available.

³⁵ Orne, M. T., ‘On the Social Psychology of the Psychological Experiment’, American Psychologist, XVII (1962), 776–83.CrossRefGoogle Scholar

³⁶ Rosenthal, R., Experimenter Effects in Behavioural Research (New York: Appleton-Century-Croft, 1966).Google Scholar

³⁷ Festinger, L., ‘Laboratory Experiments’, in Festinger, L. and Katz, D., eds, Research Methods in the Behavioural Sciences (New York: Holt, Rinehart and Winston, 1965).Google Scholar

³⁸ An example of this can be found in the main part of this same investigation. The main purpose of the rest of the survey was to test hypotheses about the effects of information about sentencing on attitudes to crime; despite varying the sentencing information quite markedly in different conditions (telling one group a criminal got a six-month prison sentence and another that he got probation), no significant differences were produced.

³⁹ E.g. Fraser, C., ‘Group Risk-taking and Group Polarisation’, European Journal of Social Psychology, 1 (1971), 493–551.CrossRefGoogle Scholar

⁴⁰ E.g. Tyson, J. L. and Kaplowitz, S. A., ‘Attitudinal Conformity and Anonymity’.Google Scholar

⁴¹ Paicheler, G., ‘Norms and Attitude Change 1: Polarisation and Styles of Behaviour’, European Journal of Social Psychology, VI (1976), 405–28.CrossRefGoogle Scholar

⁴² Hovland, C. I., ‘Reconciling Conflicting Results Derived from Experimental and Survey Results of Attitude Change’, in Warren, N. and Jahoda, M., eds, Attitudes (Harmondsworth: Penguin, 1966).Google Scholar

⁴³ The error associated with change in party support is, moreover, much greater than that associated with a single estimate (Stuart, A., ‘Public Opinion Polls’, Journal of the Royal Statistical Society, A, CXLII (1979), 443–67).CrossRefGoogle Scholar It is ironic that the most unreliable poll results are not only the most likely to be picked up by the press (Crewe, Ivor, ‘Improving but Could Do Better: A Report on the Media and the Polls in the 1979 General Election’, in Harrop, M. and Worcester, R., eds, Political Communication and the General Election of 1979 (London: Macmillan, 1981))Google Scholar but they are possibly the most influential.