Political Behaviour/Sociology

F11(b) - Political Decisions and Cognitive Biases / Décisions politiques et biais cognitifs

Date: Jun 4 | Time: 10:15am to 11:45am | Location:

Chair/Président/Présidente : Catherine Ouellet (Université de Montréal)

Discussant/Commentateur/Commentatrice : Sophie Borwein (University of British Columbia)

Optimism-Pessimism and Trust in Artificial Intelligence Technology: Jordan Mansell (Mcmaster University), Anwar Sheluchin (McMaster University), John McAndrews (McMaster University), Cliff Van der Linden (McMaster University)
Abstract: Governments are facing increasing pressure to develop policies to regulate the use of artificial intelligence (AI). We study optimism and pessimism towards technology to explain individuals’ attitudes towards trust in artificial intelligence (AI) and its growing role in society. Optimism and pessimism are psychological tendencies to expect good or bad things in the future. In contrast to other motivational accounts based on interest, this psychological approach to technological outlook focuses on how dispositions, the tendencies to act in certain ways, shapes individuals’ attitudes towards the use and regulation of technology in society. In a recent study with the McMaster Digital Society Lab, we find that optimism and pessimism are the single best predictor of attitudes towards technology. Building on this research, we ask two questions: 1) what makes someone a technology optimistic vs. pessimistic, and whether this outlook differs from general psychological disposition and, 2) how do optimism and pessimism bias individuals’ perceptions of the benefits and risks of AI. We answer this question using an online survey of Canadians (n=700) attitudes towards AI along with a within-subject experiment. We hypothesize that: 1) technological optimism will negatively correlate with behavioral updating in response to information about the perceived risks of AI and, 2) technological pessimism will negatively correlate with behavioral updating in response to information about the perceived benefits of AI. An objective of this project is to create insights for the development of better public policies for the regulation of technology.

Do cognitive styles improve election prediction? Need for cognition and its effect on citizen forecasts: Brian Thompson Collart (Université Laval), Yannick Dufresne (Université Laval)
Abstract: Citizens form expectations about the results of future elections through the process of citizen forecasting. Several studies identify political sophistication as a source of citizen forecasting ability. However, political sophistication also drives partisanship, which can reduce forecasting ability by inducing motivated reasoning. Why do some politically sophisticated individuals forecast elections accurately while others succumb to partisan motivated reasoning? This paper proposes that cognitive styles can explain the variation in citizen forecasting ability among partisans. In this paper we explore whether individual-level variations in Need for Cognition and Need for Evaluation influence citizen forecasting accuracy. To answer this question, we employ an original dataset collected during the 2025 Canadian federal election. We expect to find that higher levels of Need for Cognition and lower levels of Need to Evaluate predict higher citizen forecasting accuracy.

Speaking to Power: How Linguistic Minority Accents Shape Voter Perceptions of Party Leaders: Richard Nadeau (Université de Montréal), Florence Laflamme (Université de Montréal)
Abstract: In countries with multiple linguistic communities, does the way minority group members speak the majority language pose a barrier when they seek the highest political office? Can their accent undermine their claim to represent all citizens? Is it associated with certain stereotypes? We aim to address these questions through a randomized survey experiment conducted in Canada, a bilingual country with two official languages: English and French. Specifically, we examine how a perceptible Québécois accent in English influences anglophone Canadians’ evaluations of federal party leaders outside Québec. Participants (n = 1,200) listen to a 30-second audio recording featuring a fictional party leader speaking in English. While the content of the message remains identical for all participants, the delivery varies: the first experimental group (n = 600) hears the leader speaking with a “standard” Canadian English accent, whereas the second group (n = 600) hears the leader speaking with a Québécois accent. Participants evaluate the candidate’s competence, intelligence, and trustworthiness, and answer questions exploring whether the candidate’s accent evokes stereotypes about their political orientations. They also assess the candidate’s capacity to represent the interests of all Canadian citizens. Our study contributes to the growing body of literature on cognitive shortcuts by shedding light on language-based judgments, a relatively understudied area of political behaviour.

Party or Policy? The Role of Policy Partisanship in Voter Decision-Making: Sarah Lachance (University of Calgary), Clareta Treger (University of Toronto)
Abstract: Which matters more for voters, the political party or the policy positions of electoral candidates? We contribute to this longstanding debate by analyzing the relative importance of policy content and party cues in the multi-party Canadian context. Using the Canadian case allows us to disentangle the effects of policy and party on vote choice, which are closely intertwined in the more polarized and extensively-studied U.S. case. First, we employ a conjoint survey experiment to test whether the effect of an electoral candidate's policy position on their evaluation by voters depends on the implicit party cues that are embedded in the policy. We find that while Canadians often associate policies with specific parties, they do not seem to use these implicit party cues in their evaluation of the candidates, focusing on policy congruence instead. Second, we test whether explicit party cues reduce the weight of policy information in candidate evaluations and find that they do not. Overall, our findings suggest that party cues are not as useful for voters in multi-party systems with low polarization, and that voters rely on policy information to make a decision in these contexts.