GIS Mapping of Social Indicators

CRISP has developed a flexible computer program that provides a GIS Mapping of Social Indicatorsmeans for conducting certain types of spatial analyses with data from the large surveys conducted by Statistics Canada. The program integrates NLSCY survey data describing individuals with various indicators describing enumeration areas (EAs) and other spatial units, such as the new Dissemination Area (DA), which are derived from the 1991, 1996, and 2001 Censuses. The computer program can: (a) estimate particular “statistics” at the EA level, (b) display these statistics on maps using ARCVIEW, the geographical information system used by Statistics Canada, and (c) make inferences pertaining to the spatial variation of these statistics. The set of health outcome “statistics” can include simple descriptive indicators, such as the prevalence of teenage smokers or overweight children, or more complex parameter estimates that depict multivariate relationships, such as the odds-ratios for teen smoking associated with living in a deprived area, or the socioeconomic gradient for BMI.

In developing the program, the researchers (Willms, Chan, and Daoust) built a “contiguity matrix” that identifies the EAs that are contiguous to a target EA (and at the second and third levels of contiguity), for the full set of approximately 40,000 EAs in Canada. This matrix, which in most spatial analysis programs is determined as part of an analysis, is hard-wired into the program, thereby enabling researchers to quickly calculate a wide range of spatial statistics. For example, the program enables researchers to spatially “smooth” data describing childhood outcomes, making it easier to identify patterns in the data (Fotheringham, & Charlton, 1994), and provide results that comply with guidelines regarding confidentiality. In the future, CRISP plans to extend the program to estimate geographically weighted regression models (Fotheringham, Charlton, and Brunsdon, 1997) and spatially auto-correlated models (Haining, 1997), and to build in a procedure for conducting cluster analyses with a geographical constraint (Gordon, 1996).


  • Statistics Canada

Research Team

  • Richard Chan
  • Douglas Willms