Monash University, Melbourne
About the Webinar
A key ingredient in sample size calculations for cluster randomized trials is the intracluster correlation coefficient—a quantity that describes the similarity of the outcomes from participants in the same cluster. When considering designs like the stepped wedge, where clusters provide measurements in multiple time periods, Dr. Jessica Kasza recommends that researchers ask whether the intracluster correlation coefficient could decay over time. That is, does the similarity between participants wane as the time between their observation increases? If so, such decaying correlations should be accounted for in the design and analysis of the trial.
In this presentation, Dr. Kasza describes correlation structures for stepped wedge and related designs that allow for decaying within-cluster correlation coefficients. She discusses the implications of these decays for planning studies, the consequences of incorrectly omitting a decay in the analysis of studies, how to pick a within-cluster correlation structure, and a repository of intracluster correlation coefficient estimates that researchers can use to help plan their studies.
About Jessica Kasza
Jessica Kasza is an Associate Professor in biostatistics in the School of Public Health and Preventive Medicine at Monash University in Melbourne, Australia. After completing a Ph.D. in 2010 at the University of Adelaide, she spent time at the University of Copenhagen in Denmark. She has been at Monash University since 2013. At Monash, she leads the development of statistical methodology for longitudinal cluster randomized trials, including the stepped wedge and cluster cross over designs, with a current focus on the development of “incomplete” stepped wedge designs. She also has interests in the comparison of health care providers and in causal inference. In addition, Dr. Kasza is the President of the Statistical Society of Australia and, in that role, is keen to ensure that the Australian statistical community is diverse, welcoming, and inclusive.