9th Biennial ACSPRI Social Science Methodology Conference

Individuals’ interactions with administrative systems (e.g., health, criminal justice, and education systems) produce data about those interactions. In some jurisdictions these administrative data have been aggregated, deidentified, linked at the person level across administrative systems, and made available for research.

Several features of linked administrative data make them ideal for understanding life-course development. First, there are developmentally relevant data at all stages of the life course, from in utero development through to old age and death. Second, the data are longitudinal by design. Administrative records often consist of a timeline of date-stamped events (e.g., births, immunizations, school enrollments, hospitalizations, and monthly earnings), essentially forming a longitudinal record of people's lives as they develop. Third, data linkage across domains permits broad assessments of exposures and outcomes. Fourth, administrative data collections often permit intergenerational investigations through the ability to link between parents and children (e.g., through birth records, legal-parenting-status information, census information, and insurance records). Fifth, administrative collections which capture address or geographic information allow for neighborhood factors (e.g., disadvantage, crime rates, air pollution) to be linked to individual records and assessed in relation to life-course outcomes.

This presentation will give example of how analysis of linked administrative data resources has advanced our understanding of life-course development by utilizing the advantages of these data. Specifically, I will highlight how advances have been made through: (a) understanding small or difficult-to-study populations, (b) using causally-informative designs, (c) identifying early factors predictive of outcomes later in the life-course, and (d) identifying important neighborhood and environmental influences.

I will finish by emphasizing that linked administrative data resources cannot answer all research questions and have a range of limitations, including that (a) the data are not typically collected for research purposes and can be of lower quality (and quality of the data may change over time); (b) the data primarily consist of service contacts, and cases that come into contact with services or seek treatment may form only the tip of the iceberg of those with the condition (e.g., those with more severe problems) and may represent a biased subset; and (c) important factors that rely on observation, self-report, or subjective ratings (e.g., self-control, attachment, parenting quality, loneliness, social cohesion, and measures of the home environment) are typically not captured in administrative systems, so they cannot be the focus of investigations using population-level administrative data.