Understanding Complex Design Features via Design Effect Models

Abstract Survey research, data is commonly collected through a sample design with complex design features thatallow the relative efficiency on the precision of an estimator to be measured using the concept of the designeffect compared to simple random sampling as a reference design. This concept is most useful when thedesign effect can be expressed as a function of various design features. We propose a design effect formulasuitable under a stratified multistage sampling by generalizing Gabler et al. (1999, 2006)’s approaches formultistage sampling. Its use can either guide improvement in the design efficiency when in design stage orenable the evaluation of the adopted design features afterwards.Keywords: stratified multistage sampling, haphazard weighting, intracluster correlation coefficient, mixedeffect model, effective sample size 1. 서론 조사연구를 위한 자료수집은집락화(clustering), 불균등가중(unequal weighting), 층화(stratificaiton)와 같은복잡한 설계요소를 포함한 복잡표본설계(complex sample design)를 통해 이루어진다. 추정량에 대해 설계요소가 갖는 효율성은주로 단순확률추출을기준으로 비교한 설계효과(design effect)라는개념을통해 평가할 수있다. 설계효과는 Cornfield (1951)에 의해 처음으로 소개되었고 Kish (1965)를통해 널리 알려지기 시작한 후 그의일련의논문 (Kish, 1987, 1992)을통해 지속적으로 연구발표되었다.표본설계