Sampling Error in Generalizability Theory

• Activity Number: 297
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Education
• Author(s): Ye Tong*+
• Companies: Pearson
• Keywords: generalizability theory ; bootstrap ; variance components ; sampling error

Abstract

Generalizability theory is a broadly defined measurement model that utilizes the application of analysis variance and permits a multifaceted perspective on measurement error. This theory is widely applied in the testing and educational research, with a focus often on variance components estimation. Estimated variance components are subject to sampling errors. Bootstrap technique can be applied to estimate the sampling errors. Its application, however, is not straightforward. One layer of complexity deals with the biasness of this estimator; the other layer of complexity is associated with the many possible ways of extracting bootstrap samples and the identification of the "best" approach. Using simulated data, these complexities are investigated. Explicit rules are proposed for bias-corrections and how to select a bootstrap sample.