Abstract: Why are voters more likely to support candidates of the same ethnic group? There is considerable evidence supporting affinity voting but, while interest-based explainations dominate theoretical frameworks, there are few direct tests of the mechanism. This paper examines two kinds of explainations for affinity voting: interest based explainations and identity based expainations. In an experiement which manipulated the ethnicity of hypothetical candidates, racialized voters are indeed more likely to support racialized candidates. To explore why they do so the main part of the paper uses a series of experimental manipulations and observed variables, including a self-interest prime, measure of percieved ideological position, and a measure of ethnic self-identification. Contrary to the presumption of most existing research, there is little support for interest-based explainations. Conversely, affinity voting is highly conditional on the degree of ethnic self-identification. The paper concludes by arguing for increased attention to the effects of self-identification, including in-group bias and persuasion effects.

Abstract: Two models of vote choice appear to contradict each other. The spatial model (Downs 1957) relies on the idea that voters hold positions of their own and vote for candidates whose positions are closest to theirs. The valence model (Stokes 1992), in contrast, argues that voters do not have strong positions on the issues, and instead vote for the candidate they believe is the most competent. These two models appear contradictory. Yet, they both find empirical support in studies examining the vote of Canadian voters.   The paper tries to resolve this tension. It is hypothesized that the spatial and valence models each accurately describe the behavior of a subset of the electorate. The spatial model describes the behaviour of individuals who have a high degree of political sophistication, while the valence model describes the behaviour of voters who do not. To shed light on this issue, the paper uses data from the Canadian Electoral Study. It uses a quantitative research design which makes use of interaction variables to assess differences in behaviour across different subsets of voters. This finding has important implications for studies of voting behavior. It shows that models of voting cannot be expected to apply equally to all voters. On the contrary, when designing a model, researchers should be careful to state to which subset of the population the model should be applied. This conclusion echoes that of Anderson and Stephenson (2010), who stress the importance of studying heterogeneity effects in models of voting behaviour.

Abstract: In the last 15 years Freedom House has downgraded Russia from "intermediary" into "consolidated authoritarian" regime. Despite its gradual sliding in terms of political rights and civil liberties since the arrival of Vladimir Putin in Kremlin, Russia has not seen any significant change, either upward or downward, in terms of voting turnout in the federal elections. Why people vote in Russia after all, a country where voting does neither change the government nor seems to influence its decisions? What is the meaning to make such an inconsequential political act without been physically forced? This paper answers this question by hermeneutically analyzing Russian public opinions expressed mainly within the frame of the World Value Survey. Beyond the specific Russian case, this research bridges the gap existing between statistical and interpretative approaches in political science. It provides cultural studies with rigidity of statistical methodology, and statistics with the interpretative richness of hermeneutics. Far from making personal interest or material expectations central for explaining voting turnout, this study finds clues to a Russia version of what André Blais calls "sense of civic duty" as a voting determinant. In the Russian case, where the notion of citizenship lacks capacity to change peacefully the government, the sense of duty refers more to the national belonging and the emotional connection to the leader.

Paper / Communication

Abstract: It is widely known that choosing a candidate in an election is a complex and costly task for voters. Even if cues and heuristic facilitate this task, vote choice is still subject to multiple biases. In an effort to defend the integrity of democratic collective preferences, some students of electoral behavior argued that individual errors caused by voters biases cancel out in the aggregate, meaning that democratic decisions of the entire electorate should be considered meaningful. While some studies critic this argument using simulations, this paper intent to challenge the "aggregationnist'' argument by using real election results. Using provincial and municipal elections from the province of Quebec, we measure the ballot order effect in a high and a low cue settings. Contrary to previous research on the topic, our setting also has the advantage that the level of information required from the electorate is more reasonable. We argue that if there is still a ballot order bias in these simple settings, it falsifies the aggregationnist argument by highlighting a real instance in which noise took over the signal. Our analysis shows there is a ballot order effect at the municipal level and therefore, that the errors do not cancel out.