Workshop on the Use of Weights in Survey Data Analyses
The one-day workshop on the use of weights in survey data analysis will take place on Thursday June 8. There will be two distinct sessions: one in the morning (Introduction and theoretical issues) and one in the afternoon (on practical implications).
In the first part, I will give an introductory presentation about weighting survey data using the Swiss Household Panel (SHP) as an example. As it is the case with most panel studies on the general population, the SHP has a nontrivial structure and a plethora of variables reaching from the field of social change to living conditions or social representations. Consequently, the weighting system is quite complex. I will present different weights and explain how they are built up, their properties and the differences between them. Knowing and understanding the theories behind and the construction of these weights helps to decide which weight is the most adequate to use.
The afternoon session will focus more on practical issues. First, through an example, I will present why weights should be used when building models for analyses. Second, I will show how weights should be modified whenever subsamples are used or different samples are merged. Finally, the application of weights with some commonly used regression models will be presented in stata.