A practical bootstrap method for testing hypotheses from survey data

Abstract:

The bootstrap method is popular in the field of survey because it can be easily implemented to some situations that could not be dealt with by others. However, most of the attention has been paid to the estimation problems, such as variance estimation, and the literature on hypotheses testing by bootstrap is very limited in survey. In this talk, a practical bootstrap method, which takes the sampling weight into account, is introduced to test hypotheses in survey sampling. We will compare the proposed method with some commonly used ones in a linear regression problem under four different sampling scenarios to demonstrate its advantages.