3 Sampling design of the CLSA
The Canadian Longitudinal Study on Aging (CLSA) uses a stratified single-stage sampling design and has a large sample of 51,338 Canadian residents aged 45-85 years at baseline. Respondents in the initial sample are to be followed in subsequent survey waves for at least 20 years or until death or loss to follow-up. There are two types of respondents in the sample:
- CLSA Tracking cohort, which has a target size of 20,000 people from across the 10 Canadian provinces;
- CLSA Comprehensive cohort, which aims to have 30,000 people living within 25-50 km of one of the 11 Data Collection Sites (DCS) across 7 Canadian provinces.
Respondents in both cohorts are surveyed through telephone for a common set of baseline questions. Respondents in the comprehensive cohort are required to visit one of the DCSs to take additional measurements for key health related variables. Further details on the CLSA survey design and sampling procedures as well as measurements collected for each respondent can be found in (Canadian Longitudinal Study on Aging 2020) and (Raina et al. 2019).
The geographic variables are the stratum indicator variables for the original CLSA survey design. The variables are namely for the tracking cohorts, for the comprehensive cohorts, and for the pooled dataset of both cohorts. Researchers may also consider to stratify further using gender, age groups, education level, etc. We do not consider stratification beyond the geographic variables in this paper.