Using Small Domain Estimation to obtain better retrospective Age Period Cohort insights
Recent changes in housing costs relative to income are likely to affect people's propensity to Housing Affordability Stress (HAS), which is known to have a detrimental effect on a range of health outcomes. The magnitude of these effects may vary between subgroups of the population, in particular across age groups. Estimating these effect sizes from longitudinal data requires Small Domain Estimation (SDE) as available data is generally limited to small sample sizes. In this paper we develop the rationale for smoothing-based SDE using two case studies: (1) transitions into and out of HAS and (2) the mental health effect associated with HAS. We apply cross-validation to assess the relative performance of multiple SDE methods and discuss how SDE can be embedded into g-computation for causal inference.
READ FULL TEXT