British Journal of Political Science

Research Article

Back on the Bandwagon: The Effect of Opinion Polls on Public Opinion

Catherine Marsh*

The bandwagon was the caravan in a circus that carried the band, and usually took the lead in a procession. It has come to stand as a symbol for a party or a cause which is successful; we talk of people wanting to climb on to a bandwagon when their desire to be associated with the winning party or cause is strong. A ‘bandwagon effect’ is the label given by social scientists to a situation where the information about majority opinion itself causes some people to adopt the majority view for whatever reason; conversely, an ‘underdog’ effect is held to exist if the information causes some people to adopt a minority view. Processes of this kind are of theoretical interest because they affect the possibility of stable prediction in the social sciences; if the very act of predicting that one party will win an election can be a self-fulfilling prophecy then the natural scientific model of the social sciences may be compromised. Bandwagon processes are also of practical importance to pollsters, since the professional nature of their trade might also be compromised if their predictions could be shown to be interfering in political reality. These effects have therefore received a fair amount of attention in the literature. The main context studied has been the effect of exposure to an opinion poll on the general public. In this article, I shall first examine the existing evidence for poll effects of this kind, and then present and discuss a study of such effects.


* Social and Political Sciences Committee, University of Cambridge. The research here was carried out with the support of SSRC grant no. E 230018. I am very grateful to Nigel Walker, Wolfson Professor of Criminology at the University of Cambridge, for giving me space on a survey of penal attitudes towards crime to conduct this experiment. Thanks are due to NOP Market Research Ltd., especially to Nick Moon, not just for carrying out the fieldwork for the research but also for advice and help throughout the survey, and for helpful comments on an earlier draft of this paper. Brian Omotani in the Computer Laboratory at the University of Cambridge gave invaluable technical advice. Others who have given advice or helped are: Bob Blackburn, David Calderbank and Colin Fraser.