Using Ranked-Set Sampling in Group-Randomized Studies

• Activity Number: 640
• Type: Topic Contributed
• Date/Time: Thursday, August 8, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Xinlei Wang*+ and Johan Lim and Lynne Stokes
• Companies: SMU and Seoul National University and Southern Methodist University
• Keywords: cluster randomized design ; mean square error ; relative efficiency

Abstract

Motivated by important educational studies, we will examine the application of ranked set sampling (RSS), a cost-efficient sampling method, to replace SRS for randomization in group randomized designs (GRDs). Though well-established in Statistics and many application fields, RSS has not yet been applied in studies involving GRDs. Outcome data in such studies typically have a multilevel structure, and so hierarchical linear models (HLMs) become a natural choice for data analysis. However, the theoretical developments in RSS to date are all within the structure of one-level models. Thus, implementation of RSS for randomization at one or more levels of a HLM will require development of new theory and methods. In this research, we develop new statistical methods to analyze RSS-structured data under HLMs and new theory for examining the properties of proposed estimators and assessing the power of inferential procedures.