How Much of Electoral Politics Is in the District? Measuring District Effects on Party Support

Abstract

Abstract The electoral district is the fundamental unit of representation in single- and multi-member electoral systems, yet most research shows little interest in district effects on election outcomes, focusing instead on national and individual factors. This is problematic as parties and candidates often put a great deal of effort into district-based campaigns. How, then, can we best capture district effects on party support? We propose a new method using official election returns and geospatial techniques. The result is a measure of how much of a party's vote share is explained by district effects. Using data from the 2006–2019 Canadian federal elections, we find that, on average, 6 to 10 per cent of the variation in a party's vote in Canada is explained by district effects. While district effects on party support are trivial for some districts, in others they account for more than 80 per cent of the variance in party vote shares. The effect of districts on party support is composed, in part, of electoral context, province, socio-economic factors and district campaign intensity. Importantly, the size and sources of district effects on party support vary across parties, suggesting heterogeneity. The benefits of our approach are threefold: (1) it is cost-effective, (2) it can be easily replicated in any setting—past or present—where districts are relevant electoral units and where districting is nonpartisan, and (3) it is responsive to differences in district composition and parties’ campaign effort.