A Random Effect Model Approach to Survey Data Integration
Kim, Jae Kwang
Combining information from several surveys, or survey integration, is an important practical problem in survey sampling. When the samples are selected from similar but different populations, random effect models can be used to describe the sample observations and to borrow strength from multiple surveys. In this paper, we consider a prediction approach to survey integration assuming random effect models. The sampling designs are allowed to be informative. The model parameters are estimated using a version of EM algorithm accounting for the sampling design. The mean squared error estimation is also discussed. Two limited simulation studies are used to investigate the performance of the proposed method.
This article is published as E. Gwak, J.K. Kim, and Y. Kim (2018). “A Random Effect Model Approach to Survey Data Integration," Statistics and Applications 16, 227-243. Posted with permission.