Weighting Survey Samples (using R)
This course will cover the steps that are typically used in computing weights in surveys and the rationale behind each step. Weighting in probability samples will be emphasized, although approaches for non-probability samples, like volunteer web surveys, will also be considered. The particular steps to be covered are computation of base weights, adjustments for unknown eligibility, nonresponse adjustments, and calibration to population control totals. General techniques will be covered that can be applied in single- and multi-stage designs and to surveys of various populations, including households, establishments, schools, and other institutions. The course will include techniques for creating cells for nonresponse adjustments based on response propensities and regression trees. We will also discuss methods for correcting frame coverage errors and for reducing variance by calibrating to auxiliary information. Post-stratification, raking, and general regression (GREG) estimation will be covered. Use of weights in variance estimation will be addressed, including reasons for using replication methods, like the jackknife and bootstrap, in which series of replicate weights are created. Techniques will be illustrated using R software, particularly the R survey package.
Valliant, R., Dever, J.A., and Kreuter, F. (2013). Practical Tools for Designing and Weighting Survey Samples. New York: Springer.
Richard Valliant has been a Research Professor at the University of Michigan and the Joint Program for Survey Methodology at the University of Maryland since 2003. He has over 35 years of experience in survey sampling, estimation theory, and statistical computing. He was formerly an Associate Director at Westat and a mathematical statistician with the U.S. Bureau of Labor Statistics where he worked on some of the major U.S. economic surveys. He is a fellow of the American Statistical Association and has been an associate editor of the Journal of the American Statistical Association, Survey Methodology, and the Journal of Official Statistics.