One-day workshop: Design-based approach to sampling from a finite population
The aim of this one-day workshop is to present the core concepts of the design-based approach to sampling from a finite population.
Probabilistic - as opposed to non-probabilistic - designs are presented in some detail, and potential sources of sampling errors are investigated. The unbiased estimation of population parameters under each such design is considered, and the important problem of quantifying uncertainty in estimation is discussed.
The practical aspects of implementing some classic (simple random sampling, cluster, stratified) and modern (general π-ps, adaptive cluster) designs using R are presented, together with a quick overview of currently available computational tools for sampling, and of some useful graphical tools for data visualization (e.g., choropleths).
By the end of the day, participants should be able to identify and understand the main components of complex sampling strategies and empirically investigate their properties via simulation.
Previous knowledge of statistical inference is recommended; basic command of R is desirable, but not required.
Registration and contact
This workshop is supported by the Strategic Research Program in Epidemiology (SFO-Epi) at Karolinska Institutet.Contact person: Nicola Orsini