서베이앤헬스폴리시리서치센터

Hong, S. J., and Kim, S. W. (2012). “New model-optimized sampling techniques,” Paper presented at the Joint Statistical Meetings, San Diego, California.

Abstract

The work of Kim, Heeringa, and Solenberger (2006) provided a theoretical basis for what is called model-optimized sampling methods for yielding sampling designs that give large variance reductions as well as the stability of the variance estimates. Their methods were based on a simple linear regression superpopulation model. Hong et al. (2009) suggested modified sampling methods based on the same model. However, in many real populations, using more complicated superpopulation models would be better concerning efficiency. For this, we suggest model-optimized sampling methods using general polynomial superpopulation models. We illustrate the benefits of our new approaches by comparing the efficiencies between the different models.

Key Words: superpopulation, general polynomial model, model optimization

For more information about model-optimized sampling techniques, click on the following title of the Ph.D. dissertation (Korean) by Hong, S. J. (홍성준, 2013) or search for it at https://riss.kr/index.do

일반다항회귀 초모집단 모형 기반의 크기비례포함확률추출법(πPS Sampling Strategies under Generalized Polynomial Regression Super-population Model)