A methodology is proposed for assessing the impact of televised debates on electoral outcomes, and it is applied to a specific case, that of the 1988 Canadian election. We present four tests of the debates' impact: first, a cross-sectional group comparison, which contrasts the voting behaviour of those who did and those who did not see the debates; secondly, a panel analysis of the shift in party support, before and after the debates, among those who watched the debates and those who did not; thirdly, a panel study of the impact of reactions to the debates on voting behaviour; and, fourthly, a time-series analysis, which examines the evolution of vote intentions over the course of the campaign and, more precisely, before and after the debates.
It is argued that because non-watchers are influenced by what their friends or the media tell them about the debates, the first two designs, based on a comparison of debate watchers and non-watchers, are not appropriate. The empirical analysis of the 1988 Canadian election substantiates this point. While these first two designs seem to indicate no debate impact, panel reaction and time-series analyses show that the debates had a substantial and enduring impact on the vote and that they were decisive in the contest for second place between the Liberals and the NDP.
* Department of Political Science, Université de Montreal; and Department of Risk and Insurance, University of Pennsylvania, respectively. We thank the Social Sciences and Humanities Research Council of Canada for its financial support, Claude Blanchette for his research assistance, and Richard Nadeau, Marcel Dagenais and David Sanders for their comments on previous drafts. A replication set of the data used in this article can be obtained from the first author.