The Effect of Electoral Systems on Voting Behavior and Party Development: The Case of Canada
This thesis will explore three major areas of political science research: The timing of elections, Duverger's law and voter turnout. Each of these topics has been covered extensively in the literature. However, as with most areas of political science, there remain facets of each of these topics that deserve further examination. Chapter 1 begins this discussion with an introduction to each of these three topics and a description of the Canadian electoral and party systems. Chapter 2 examines the impact of the timing of elections on the government party in Canadian parliamentary elections. This chapter shows that there are aspects of the study of election timing in presidential regimes that also pertain to the study of election timing in parliamentary regimes. In particular, this chapter finds that the "honeymoon period" described by Shugart and Carey (1992) and Shugart (1995) for presidential regimes relates to early election calls by the government in parliamentary regimes. The issues of sample size and susceptibility to operationalization limit the explanatory power of the analysis in this chapter. While these results show that future research in this area is warranted, these problems must be addressed in order to enhance these initial findings. Duverger's law states that "the simple-majority single-ballot system favours the two-party system" (1964, 217). An important implication of the law is that the electoral system is a significant determinant of the number of parties. Chapter 3 shows that even when multi-partism appears at the national level, the district level can present a different picture depending on how the number of parties is counted. This number of competitive parties is determined by using the same LT index method as Gaines (1999). Using this vote share method, the results for the 1997 and the 2000 election confirm the pattern established by Gaines for the elections from 1935 through 1993. By using seat shares instead of vote shares in the LT index, I find that Canadian elections at the riding level are more appropriately characterized as two-party or even one-party competition. This finding, which is the most significant contribution of this chapter, concurs with Duverger's (1964) hypothesis that Canadian exceptionalism at the national level is the result of different regional parties competing with national parties in different regions of the country. There are two main methods that are used to study voter turnout. Single country studies, particularly in the case of the U.S., tend to focus on socioeconomic and behavioral variables while comparative studies tend to favor an analysis of institutional variables. In Chapter 4, I present an analysis of voter turnout using a combination of the two approaches. Chapter 4 presents a longitudinal analysis of Canadian turnout using the institutional variables explored in Chapters 2 and 3 as well as a snapshot analysis of the 1997 Canadian election. I use regional analysis to demonstrate the problem with using an aggregate analysis to explain a phenomenon that is affected by different variables in different regions. This analysis shows that different independent variables have a distinctly different effect on voter turnout in different regions. When combined in an aggregate analysis, these independent variables lose some of their explanatory power because of the contradictory effects observed in different regions.
