Table 2 Spatial characterization of different field of neighbourhood

Domain 1: socio-economic environment

Population

Total population

All socio-economic variables

Zonal data available at census block level (2000 inhabitants on average)

Using the ArcGIS software zone-clipping algorithm, we disaggregated the variables according to real weighting interpolation methods. Because the value of the information transferred to the cell was thus a function of the area common to both the initial area (here, the census block) and the grid cell, these variables were able to be integrated into the final analysis

Domain 2 : public resources

Healthcare system

Location of doctors’ surgeries

Location of healthcare centres

Point data: address

We assigned to each cell centroid the road distance (non-Euclidian) to the nearest healthcare centre or doctor’s surgery

Public parks and gardens

Location and area of public parks and gardens

Polygon data:

Location and size

We built an attractiveness index for public parks and gardens, derived from French studies showing that attractiveness is a function of size. Using GIS tools, we drew concentric zones of attractiveness by area: 100 m (area less than 1 ha), 500 m (area 1–10 ha), and 1000 m for larger areas. We subsequently computed this index for each cell

Sports facilities

Location of sport facilities

Point data: address and coordinate X, Y

The road network distance to the nearest sports facility was attributed to each cell centroid

Public transportation supply

Location of bus and tram stop and the number of lines served at each

Point data: coordinate X, Y

Using GIS tools, and on the basis of modal differential attractiveness between these two types of public transportation, we constructed a public transportation availability indicator, with a catchment area attributed to each stop (300 m for a bus stop, 400 m for a tram station), weighted by the number of lines at each stop or station. This indicator was then assigned to each cell

Domain 3: psychosocial environment

Local businesses

Location of retail outlets

Point data: address and coordinate X, Y

Using GIS tools, we attributed to each unit the quantity of retail stores relative to all available retail space within a radius of 200 m around the spatial unit centroids. The resulting values associated with the retail store scoring (quantity of retail stores relative to all available retail space) by category (itinerant vendors; retail food stores; retail non-food stores and other services) were attributed to each unit^a

Location of food markets

Point data: address and coordinate X, Y

Characterization of educational facilities

Violence in schools

Schools’ social scores

Primary/middle and secondary (high) schools

ZEP (priority) and successful (AR) middle schools

Map showing primary and middle schools

Secondary (High) schools

Point data: address and coordinate X, Y

The French school environment is graded as: (1) Priority education zones (ZEP-Zone d’éducation prioritaire), where establishments receive additional resources and have greater autonomy for dealing with educational and social difficulties, (2) “successful ambition” zones (AR), having fewer (but definite) needs and thus fewer resources), and (3) others. All non-private schools in the city and their catchment area were geocoded, using information provided by local authorities. We computed an indicator taking into account school density and classification (primary/middle or secondary/high schools)

Voting rates

Voting rates

Zonal data available for each center of vote

Civic associations

Civic associations

Point data: address and coordinate X, Y

The fairly exhaustive and georeferenced SIRENE database allowed calculation of the ratio of the number of (official) civic associations per 100 inhabitants in each unit, taking into consideration their type (religious, political, other)

Type of civic associations: religious, political, volunteer

Point data: address and coordinate X, Y