New article: The Spatial Articulation of Urban Political Cleavages

By Jan Doering, Dan Silver, and Zack Taylor in Urban Affairs Review.

Sociologists and geographers have long placed space and place at the center of their analyses. They have shown that people’s identities and attitudes are inflected by their social and physical contexts—who their neighbors are and what kind of place they live in—although they have not always extended this to politics. Studies of urban politics, on the other hand, have focused on individual characteristics such as race and gender rather than space or place. In their important study of exit polls in American big-city elections, Trounstine and Hajnal (2014) find that race overwhelms all other factors. Elections in large American cities are predominantly contests between cohesive groups defined by race. 

In our Urban Affairs Review article, “The Spatial Articulation of Urban Cleavages,” we develop a framework for rooting urban politics in space. We theorize three sources that turn urban politics into spatial divisions: group-, place-, and location-based interests. Group-based interests emerge from shared experience and linked fate on the basis of race, class, religion, sexuality, language, and other salient markers of difference. In segregated cities, where groups cluster in particular districts or neighbourhoods, conflicts between groups are articulated in space. A distinction between “us” and “them” may manifest politically as, for example, a distinction between “East Side” and “West Side.”

Place and location-based interests derive from different sources. Regarding place, cities offer a range of urban experiences, because neighborhoods vary in terms of their built environments, density, amenity composition, and so on. These idiosyncratic characteristics in turn generate unique identities, interests, and lifestyles. The consumption patterns and policy demands of privatistic “homevoters” in low-density, car-dependent neighborhoods differ from inhabitants in dense neighborhoods with mixed uses and collective amenities that are accessible by walking and cycling. If place-based interests emerge from lifestyles and identities associated with neighborhoods’ internal features, location-based interests derive from the fact that areas have distinct needs and priorities depending on their relative position within the city as a whole. Triggered by transit cuts and gas price increases, France’ Yellow Vest movement, for example, developed in part as a conflict between peripheral, car-dependent locations and centrally located areas with comprehensive transit access.

To examine the spatial basis of political cleavages, we constructed an original dataset of neighborhood-scale mayoral election results over two decades in three cities, Chicago, Toronto, and London. These cities are large in population and territory, socially diverse, and contain many different kinds of places, from dense central business districts to apartment tower neighborhoods to low-rise residential tracts. At the same time, each is also located in a different national context with distinct institutional features and social histories. 

Our analysis unfolded in three stages. First, we sought to uncover to what extent political divisions occur between rather than within neighbourhoods. Strikingly, we found that most elections are landslides at the neighborhood scale even if close citywide. Precinct margins of victory were greater than 20 percentage points in at least 60% of precincts in each of the three cities. This suggested that election results are spatially articulated to a similar degree in Chicago, Toronto, and London, and that the national U.S. trend toward urban-rural polarization in voting despite close national results has a counterpart within urban politics, even when elections are nonpartisan. 

Our next step was to uncover the nature of these divisions. As Chicago and Toronto have nonpartisan elections featuring multiple candidates, and London has a shifting array of party-nominated and independent candidates, we used principal component analysis, a data reduction technique, to reduce this complexity to three latent variables (“components”) that account for over 80% of variation in neighborhood vote shares across time in each city. We then correlated neighborhood scores on these components with a range of census and other data representing group-, place-, and location-based interests. We found that the positive and negative correlations on each component correspond to recognizable bundles of identities and interests, although their strength varies across the three cities. (See Table 1.)

Table 1: Summary of cleavages

Black ↔ White/Latino (56%)
African American neighborhoods support Black candidates. White and Latino neighborhoods support their opponents.
Periphery ↔ Core (56%)
Dense core supports progressive candidates. Low-density, auto-dependent periphery areas support conservative candidates.
Periphery ↔ Core (41%)
Dense core supports progressive parties. Lower density periphery areas support conservative parties.
White ↔ Latino (14%)
White, well-off, privileged areas support business-friendly white candidates. Mostly Latino, blue-collar areas support Latino candidates when they run.
Establishment ↔ Marginality (20%)
Privileged areas support establishment conservative candidates. Disadvantaged areas support right-populist candidates when they run. 
Establishment ↔ Marginality (27%)
Privileged areas support mainstream parties (Labour, Conservatives, Liberal Democrats). Disadvantaged areas support anti-establishment parties (BNP, UKIP, Respect).
Periphery ↔ Core (11%)
Ethnically diverse core areas support more progressive candidates. Auto-dependent periphery areas support more conservative candidates.
South Asian ↔ Southern European (7%)
Established southern European neighborhoods support Italian candidates when they run. Asian immigrant neighborhoods cohere when appealed to.
Immigrants ↔ UK-born (13%)
UK-born areas support Lib Dems and xenophobic parties. Established, mostly South Asian and Muslim immigrant neighborhoods support Conservatives and Respect.
Maps are not to scale.

As might be expected in a city that is highly and durably segregated by race, and where class and race are closely linked, Chicago’s dominant cleavage is race. Across all six elections, the strongest, most persistent division is between Black neighborhoods and White and Latino neighborhoods. A second, much weaker cleavage also revolves around race and class, dividing well-off White neighborhoods and blue-collar Latino neighborhoods. A third, very weak cleavage is based on place-based lifestyle and location, pitting dense centrally located neighborhoods against peripherally located, car-dependent single-family, neighborhoods. 

The cleavage structures found in Toronto and London are similar to one another and almost the inverse of Chicago’s. The dominant cleavage in both cities is rooted in place-based lifestyles. On one side are dense and accessible core neighborhoods whose residents tend to be young, single, and work in creative occupations. On the other side are peripherally located neighborhoods featuring low-density housing, auto-dependency, and traditional lifestyles. The next-strongest cleavage turns on class, pitting higher-income educated professionals against lower-income, blue-collar residents. The weakest cleavage divides native-born from foreign-born residents. In stark contrast with Chicago, lifestyle is the strongest cleavage, and ethno-racial group identity the weakest. 

Overall, our findings demonstrate that urban politics are strongly articulated in space, although this happens in different ways in different cities. This opens up new possibilities for research comparing cities within and across national borders. For example, we suggest that group-based cleavages will dominate in cities that are segregated by race, language, religion, and class, while place- and location-based cleavages will appear in large jurisdictions that contain highly differentiated built environments. Furthermore, our findings call more for research on the actions and processes that produce and activate group-, place-, and location-based cleavages.

Post forthcoming in Urban Affairs Forum.