Using challenges and obstacles encountered during the 2nd South African Comparative Risk Assessment Study (SACRA-2) as motivating examples, the talk presents an overview of methods and approaches for estimating population distributions of risk factors when available data sources are sparse – i.e. they don’t cover the whole range of subpopulations of interest – and heterogeneous in terms of available measurements, time frame and quality. While all the approaches presented rely on statistical modelling in their implementation, the focus of the talk is not statistical. It rather aims at clarifying concepts, assumptions and limitations underlying the various approaches and stimulating discussion on applications beyond the examples discussed.
