Tutorial: Survey analysis in R

Thomas Lumley,  Dept of Biostatistics, University of Washington, Seattle


The 'survey' package in R provides a fairly comprehensive set of methods for analyzing data from complex sampling designs. In particular, it provides facilities for regression modelling and exploratory data analysis, and also for calibration and raking.  The workshop will look at both data analyses and simulations with the survey package, based on real surveys and populations -- including the complete public-use data from BRFSS, the world's largest telephone interview survey.


1) Basic structure of the package
2) Describing surveys to R
3) Summary statistics
4) Graphical techniques in the presence of sampling weights
5) Regression models
6) Using multiply imputed data
7) Simple two-phase designs.


The tutorial should be suitable either for people with experience in data analysis and modelling in R or for people with experience analyzing complex surveys in other statistical systems and some elementary knowledge of R.