All Times EDT
Virtual
Comparing Methods for Randomized Complete Block Designs with Repeated Measures Over Time (310015)
*Laurie Moulton, Procter & GambleYuanshu Zou, Procter & Gamble
Keywords: Repeated Measures, Randomized Complete Block, Kenward Rogers, Mixed Model, Simulation, False Positive Rate, Power, Variance-covariance Structure, Adjustments to Control Type I Error
Analysis methods for Randomized Complete Block designs with repeated measures over time (i.e. Repeated Measures Analysis and By Visit analysis, both using a Mixed Model) are discussed and compared through simulations. Various analysis approaches, to address the nuances of the RCB design, were assessed for sensitivity by looking at the power of the method and for the control of the false positive rate. With repeated measurements, variance-covariance structure for the random error and adjustments needed to the degrees of freedom and adjustments to the variance-covariance estimates to control the false positive rate were taken into consideration. Recommendation were given at different scenarios as a conclusion.