Targeted Machine Learning for Latent Outcomes in Causal Research

• In Event: Statistical Frontiers: Bayesian Inference, Machine Learning, and Longitudinal Modeling

Abstract

Prior research has developed methods specifically designed to accommodate latent outcomes (e.g., structural equation models [SEMs]). A critical limitation of conventional methods is that they typically operate under assumptions of linear effects and almost always assume correct a priori model specification (e.g., all non-linearities and n-way interactions are known and included). Machine learning (ML) has relaxed such limitations by developing data adaptive methods that empirically construct these relationships. However, ML methods have been almost exclusively developed for observed or manifest outcome variables and do not accommodate latent variables. In this study, we developed and evaluated methods to implement ML with a super learner within the context of the targeted learning framework that supports causal inference with latent outcomes.

Authors

• Amota Ataneka, University of Cincinnati

• Benjamin Kelcey, University of Cincinnati

• Jean Baptiste Habarurema, University of Cincinnati