Approximating Power for Multilevel and Longitudinal Studies with Missing Data

• Activity Number: 169
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Brandy Ringham*+ and Sarah Kreidler and Keith Muller and Deborah Glueck
• Companies: University of California, Los Angeles and University of Colorado Denver and University of Florida and University of Colorado Denver
• Keywords: non-central F power approximation ; multiple outcomes ; multilevel ; longitudinal ; missing completely at random

Abstract

Catellier and Muller proposed a data analytic technique to account for data missing at random in multilevel and longitudinal studies. They suggested modifying the degrees of freedom for the Hotelling-Lawley trace and its null case reference distribution. We propose parallel adjustments to approximate power for multilevel and longitudinal studies with missing data. The power approximations modify the non-central F statistic with one of three functions of the expected effective sample size: 1) the expected number of complete cases, 2) the expected number of non-missing pairs of responses, or 3) the trimmed sample size, which is the planned sample size reduced by the anticipated proportion of missing data. The accuracy of the method is assessed by comparing the theoretical results to the empirical power, which is estimated using a Monte Carlo simulation. Over all experimental conditions, the closest approximation to the empirical power is obtained by adjusting power calculations with the expected number of complete cases. The utility of the method is demonstrated with a power analysis for a hypothetical oral cancer biomarkers study.