The Inclusion of Covariates in Constrained Longitudinal Data Analysis for Pre-Post RCTs

• Activity Number: 443 - Latent Variables, Causal Inference, Machine Learning and Other Topics in Mental Health Statistics
• Type: Contributed
• Date/Time: Thursday, August 12, 2021 : 4:00 PM to 5:50 PM
• Sponsor: Mental Health Statistics Section
• Author(s): Joseph R Rausch*
• Companies: Nationwide Children's Hospital, Ohio State University Medical Center
• Keywords: randomized clinical trials; constrained longitudinal data analysis; causal inference; analysis of covariance; covariates; treatment effects

Abstract

I review the practical advantages of constrained longitudinal data analysis (cLDA; see, e.g., Coffman, CJ, Edelman, D, & Woolson, RF, BMJ Open, 2016 for a relatively recent discussion of this approach) when compared to the analysis of covariance (ANCOVA) for the analysis of treatment effects in pre-post RCTs. For example, when data are missing on the outcome or any of the covariates of interest, cLDA offers a valid but flexible approach for analysis of data, which are assumed to be missing at random, when compared to ANCOVA. This is because cLDA does not generally require list-wise deletion for a participant when data is missing on the outcome or any of the covariates of interest, in contrast to ANCOVA. I will also discuss different methods for including covariates in the analysis and the practical ramifications for each of these approaches when employing cLDA. The cLDA approach and ANCOVA will be compared using an illustrative example from the behavioral health sciences to demonstrate how these different analytic methods compare to one another and can be used in practice.