A Hybrid Method in Combining Treatment Effects from Matched and Unmatched Studies

• Activity Number: 631
• Type: Contributed
• Date/Time: Thursday, August 5, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Jinyoung Byun* and Dejian Lai+ and Sheng Luo and Robert Hardy
• Companies: The University of Texas School of Public Health and The University of Texas School of Public Health and The University of Texas School of Public Health and The University of Texas School of Public Health
• Keywords: nonparametric hybrid test statistic

Abstract

In many biomedical studies,subjects are usually paired for the more effective comparison of the treatment.For example,one may want to verify an interventional treatment on one or both diseased eyes. For both diseased cases,if one eye is randomized into treatment, then the other is automatically allocated to the control or vice versa that is a matched design.The patient may only have one diseased eye.The diseased eye would be randomly allocated to either treatment or control which is an unmatched study.It is common type of studies on ears, knees, kidneys and other clustered measurements.We will introduce a new nonparametric hybrid test statistic combined on the Wilcoxon's signed rank and rank sum statistics for matched and unmatched samples.We will also present preliminary simulation and comparative results on the size and the power for samples observed from continuous and discrete cases.