Abstract:
|
We study the use of ranked set sampling (RSS) with binary outcomes in cluster randomized designs, where a generalized linear mixed model (GLMM) is used to model the hierarchical data structure involved. Under the GLMM-based framework, we develop different estimators of the treatment effect, including the nonparametric estimator (NP), maximum likelihood estimator (MLE) and pseudo likelihood estimator (PL), and study their properties and performance via numeric evaluation and/or simulation. We also develop procedures to test the existence of the treatment effect based on the three RSS estimators, and examine the power and size of the RSS tests vs. simple random sampling (SRS) tests. Further, we illustrate the proposed RSS methods with two data examples, one for rare events and the other for non-rare events. Imperfect ranking is within our consideration throughout our study. Recommendations will be given on whether to use RSS over SRS with binary outcomes in CRDs, and if yes, when to use which RSS estimator among NP, MLE and PL.
|