Umeå
Incomplete data: semi-parametric and Bayesian methods (Winter Conference)
Incomplete data is a common phenomenon in longitudinal studies based on surveys and/or population registers. In such studies individuals are followed through time and data may be incomplete due, e.g., to drop out/attrition (individuals intentionally leave the study, individuals leave the study because they move or die, etc.) and censoring (due to end of study, death, etc). Incomplete data may also arise due to selection mechanisms, for instance, in meta-analyses (publication bias) and causal inference in observational studies.
Statistics for register based research
The course covers the following topics related to register based research:
- Registers, measurement and quality issues
-Causality and inference in observational studies
-Graphical models, multivariate analysis
-Registers, population based studies and surveys, combining different sources of data
