Evaluation and Development of Approaches for Handling Missing Data in Complex Longitudinal Studies
Chief Investigator: Melissa Middleton
Synopsis: In Australia, and globally, there has been a growing focus on larger and more complex epidemiological studies. With these studies producing ever-increasing volumes of data, there is a clear need for developments in biostatistical methodology to handle the data produced. My proposal seeks to extend and evaluate statistical methods that are needed to address two major issues that arise together in modern epidemiological studies: complex study designs with differential probability of selection, and missing data. While the former is a powerful tool for the efficient and cost-effective use of data, the latter is an inevitable nuisance that threatens the validity of study results. There is however no clear guidance on best methods for handling them jointly. This project aims to address this gap in the literature, developing and evaluating methods for handling missing data in the context of studies with unequal probability sampling, with a focus on multiple imputation and inverse probability weighting in case-cohort studies and surveys that oversample small populations. This is guided by the issues encountered in two prominent Australian longitudinal cohort studies, the Barwon Infant Study (BIS) and the Longitudinal Study of Australian Children (LSAC). Translational outcomes will be guidance on best practice for the applied researcher.
Funding: This project is supported by and part of larger research being conducted by the Victorian Centre for Biostatistics. Melissa Midleton is funded through an NHMRC Postgraduate Scholarship (GNT1190921) and the Australian Government Research Training Program.