Intellectual Merit: Conventional wisdom asserts that elites drive opinion change in advanced democracies (i.e. Converse 1964; Chubb, Flanigan, and Zingale 1980; Carmines and Stimson 1989). In Canada, the behavioral ramifications of the ongoing transformation of the party system, a process that began in 1993 after fifty years of relative stability, calls into question this assumption. This proposal hypothesizes that: a) vote choice in Canada is best modeled using a variant of the "issue priorities" model (c.f. Budge and Farlie 1983) in which positional issues are mass rather than elite driven; and b) tests this model against rivals contending that both positional and valence issues vary by region (c.f. Gidengil et al. 1999); and c) a multidimensional variation of the spatial model with both positional and valence issues (c.f. Groseclose 2001). To test these models, pseudo-panels employing national samples of the Canadian electorate generated in a number of cross-sectional and panel studies since the 1980s will be employed. These include both studies of national elections and studies conducted in the interim between elections. A variety of multivariate statistical techniques such as modeling both partisanship and ideology on a latent variable level and making use of multinomial probit techniques will assess the validity of the hypotheses. Advances in adapting these quantitative approaches for use in the analysis of political behavior will enable the modeling of the stability of both ideology and partisanship within the multi-wave panels while controlling for measurement error (e.g. Green, Palmquist, and Schickler 2002). These measurement techniques will also provide insight as to whether the much noted unstable nature of partisanship in Canada (c.f. Meisel 1975; Clarke et al. 1996; Clarke et al. 2000) was "time bound" and a byproduct of the party system that existed before 1993. Since the set of parties available to the Canadian voter differs from region to region, and even between neighboring ridings, the "independence of irrelevant alternatives" (IIA) assumption is likely violated in standard analyses of voting behavior in Canada (Greene 2003). If this is the case estimation of modeling parameters will be done using Breslaw's (2002) GAUSSX multinomial probit procedure. Since the Canadian national election of 1993, two new parties, the Bloc Quebecois (BQ) and the Reform/Alliance have become major players in national electoral politics, whereas two old-line parties, the Progressive Conservatives (PC) and the left-of-center New Democratic Party (NDP) have become distinctly minority participants in national elections. However, a merger has been arranged between the old-line Tory and the Alliance who will henceforth run under the label, Conservative Party of Canada (CPC). The models to be tested have the additional merit of analyzing and explaining whether the proposed merger of these two parties will be consummated, and also whether the merged party will be successful in the sense of either winning the next election or seriously challenging the governing Liberals. Funding will be utilized to add several questions to a pre-post study of an election anticipated during spring 2004 that is being undertaken by Dr. Allan Kornberg. Additional funds are sought for a mail- back questionnaire that will employ issue questions that are pertinent to the investigation of the hypotheses and the testing of the above models. The pre-post telephone interviews will constitute the fourth and fifth wave of a panel begun by Dr. Kornberg with a pre-post 2000 election study and a 2002 interim study. b) Broader Impact: The study will be the first to systematically assess whether the so-called "Absent Mandate" model (c.f. Clarke et al. 1996) that has largely driven the study of Canadian elections since the 1970's is time bound and in need of serious revision given the changes that hypothesized to have occurred at both the elite and mass level of one of the world oldest, continuing democracies.
