From health research to social research: privacy, methods, approaches.

Information-rich environments in Canada, Australia, and the United Kingdom have been built using record linkage techniques with population-based health insurance systems and longitudinal administrative data. This paper discusses the issues in extending population-based administrative data from health to additional topics more generally connected with well being. The scope of work associated with a multi-faceted American survey, the Panel Study in Income Dynamics (PSID), is compared with that of the administrative data in Manitoba, Canada. Both the PSID and the Manitoba database go back over 30 years, include families, and have good information on residential location. The PSID has emphasized research design to maximize the opportunities associated with expensive primary data collection. Information-rich environments such as that in Manitoba depend on registries and record linkage to increase the range of variables available for analysis. Using new databases on education and income assistance to provide information on the whole Manitoba population has involved linking files while preserving privacy, scaling educational achievement, assessing exposure to a given neighborhood, and measuring family circumstances. Questions being studied concern the role of the socioeconomic gradient and infant health in child development, the comparative influence of family and neighborhood in later well being, and the long-term effects of poverty reduction. Issues of organization of research, gaps in the data, and productivity are discussed.