IASS Webinar 53: Multiphase sampling: design, analysis, computation

Watch the webinar recording

Biomedical and health studies often benefit from subsampling, because a new or expensive assay is desired for stored blood and tissue samples or because electronic health records need to be validated against clinical notes, or for other reasons. Notably, many of these applications do not involve non-response, so optimisation of the design and analysis is worthwhile.  As with all statistical methods, these require software, and it is desirable for analysts with subject-matter expertise can run as much of the analysis as possible.  I will talk about two-phase sampling in medical research and about software development for survey analysis. A unifying principle for these two topics is the concept of influence functions as an approach to generalising classical theory and methods from survey statistics.