Sample Size Re-Estimation for Laboratory Experiments

• Activity Number: 657
• Type: Contributed
• Date/Time: Thursday, August 2, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Lai Wei*+ and David Jarjoura
• Companies: The Ohio State University and The Ohio State University
• Keywords: Adaptive designs ; Conditional error function ; Conditional power ; Small sample size

Abstract

We developed a small sample size hypothesis testing procedure based on a potential need for repeating a laboratory experiment a second time to achieve significance. The sample size for a second experiment or stage is chosen to achieve sufficient conditional power, given the first stage results. The testing procedure is based on an adaptation of Bauer and Kohne's (1994) method. We emphasize the importance of achieving significance in the first stage, and this results in simplifying the multiple considerations usually associated with interim analysis and decision making in clinical trials. Within this context of laboratory experiments, we show that a range of reasonable options including the futility cut-off, effect size estimation, and significance levels for the first stage provide similar power and expected overall sample size. We contrast this approach to a naive procedure in which a second unplanned experiment is run based on non-significance in the first experiment, and data are combined as if they were obtained from one run.