Don’t skew it up: on the statistical efficiency of continuous composite outcomes when one component is skewed.

Seminar Series

Friday, May 8, 2020 - 01:30
Zoom Conference
Berd Core Logo

Abstract: Under the right conditions, statistical efficiency can be gained in clinical trials by using composite outcomes. Little work has been done in the literature to examine the properties of continuous composite outcomes (i.e. composites where each individual component is a continuous variable). We have previously examined the statistical properties of composites formed from normally distributed continuous outcome measures. In this seminar we will review these properties and extend them to the case where one of the component outcome measures is skewed. These concepts will be discussed from the standpoint of planning a new trial.

To attend this meeting, please contact Lauren Howard at lauren.howard@duke.edu to obtain the Zoom link.

Ryan Simmons, MB
Biostatistician II