Construction of Weights in Surveys
Co-sponsored by AAPOR and DC-AAPOR
Presented by David Haziza
March 16-17, 2017
Bureau of Labor Statistics Conference Center
Bureau of Labor Statistics
Postal Square Building
2 Massachusetts Ave., N.E.
Washington, DC 20212-0001
THIS IS A LIVE, IN-PERSON COURSE, NOT AVAILABLE ONLINE.
2017 Members: $100
2017 Non-Members: $300 (includes AAPOR and DC-AAPOR membership)
Most surveys are designed to provide statistics for a possibly (very) large number of characteristics of interest. Typically, the data collected are stored in a rectangular data file, each row corresponding to a sample unit and each column corresponding to a characteristic of interest. Made available on the data file is a weighting system. The idea is to construct a single weighting system applicable to all the characteristics of interest. The typical weighting process involves three major stages. At the first stage, each unit is assigned a base weight, which is defined as the inverse of its inclusion probability. The base weights are then modified to account for unit nonresponse. At the last stage, the weights adjusted for nonresponse are further modified to ensure consistency between survey estimates and known population totals. When needed, the weights undergo a last modification through weight trimming or weight smoothing methods in order to improve the efficiency of survey estimates. The goal of the workshop is to provide a detailed (theoretical and practical) description of each stage of the weighting process. Throughout the workshop, different concepts will be illustrated through live simulation studies.
About This Course:
About the Instructor:
David Haziza is an Associate Professor in the Department of mathematics and statistics at the Université de Montréal.
Day 1 Morning
Day 1 Afternoon
- Introduction to weighting: multipurpose surveys, weighting system, notation, finite population parameters
- Sampling designs: definition and commonly used sampling designs in practice (stratified simple random sampling, Poisson sampling, proportional-to-size sampling, two-stage sampling, two-phase sampling, indirect sampling), first and second-order inclusion probabilities
- The basic weighting system: Unweighted estimator, Horvitz-Thompson estimator, weighting in two-stage sampling, weighting in two-phase sampling, weight-share method
Day 2 Morning
- Calibrated weighting system: definition, general formulation of calibration, calibration methods (linear method, exponential method, truncated linear and logit methods), properties of calibration methods, calibration in two-stage sampling, calibration in two-phase sampling, model calibration
Day 2 Afternoon
- Weighting system adjusted for nonresponse: estimation of the response probabilities (parametric and nonparametric methods), empirical double expansion estimator and its properties, one-step approach, two-step approach
- Weight trimming and weight smoothing methods
- Use of a single weighting system: internal and external consistencyIdentify a proper scale for the measurement task.
Day 1 and 2
|Lunch on your own
For more information contact: Heidi Diederich at firstname.lastname@example.org or call 847-686-2230