The Shape of Public Opinion and Politicians' Perceptual Errors

• In Event: Beyond the Mean: New Directions in Elite Behaviour and Political Representation

Abstract

Politicians in advanced democracies consistently misperceive their constituents’ preferences and ideological positions. Even more worryingly, this perceptual error tends to be biased, with accumulating work showing that most politicians overestimate the conservatism of their constituents’ preferences. This pervasive and systematic perceptual error has serious implications for political representation: politicians’ understanding of public opinion shapes their own policy priorities, communicative strategies, and legislative voting behaviour. The sources of these biases and errors are so far poorly understood. However, recent research suggests that the way that politicians are being asked to think about public opinion when making their assessments could be a cause for this systematic bias: when individuals are prompted to think more reflectively about how attitudes are distributed in a given population, rather than simply making point estimates, they become more accurate.

In this paper, we draw on evidence from large surveys of politicians and the general public to explore this “distributions versus point estimates” comparison among elected politicians for the first time. Using data from nearly 2,000 unique Canadian local politicians, together with surveys of more than 100,000 Canadian citizens, we first replicate past studies, demonstrating pervasive conservative overestimation among the politicians in our sample. Then, using an experiment embedded in a survey of more than 600 local politicians, we explore whether politicians’ perceptual error is improved by a more reflective module focusing on the distribution of constituents’ ideological positions, rather than their mean values. We conclude with a discussion of the implications of our findings for research on elite perceptual accuracy and political representation more generally.

Authors

• Nicholas Carlo Dias, University of Pennsylvania

• Jack Lucas, University of Calgary

• Lior Sheffer, Tel Aviv University