Journal Article

Robust Machine Learning for Treatment Effects in Multilevel Observational Studies Under Cluster-level Unmeasured Confounding

Suk, Youmi, Kang, Hyunseung

2022

Overview

Recently, machine learning (ML) methods have been used in causal inference to estimate treatment effects in order to reduce concerns for model mis-specification. However, many ML methods require that all confounders are measured to consistently estimate treatment effects. In this paper, we propose a family of ML methods that estimate treatment effects in the presence of cluster-level unmeasured confounders, a type of unmeasured confounders that are shared within each cluster and are common in multilevel observational studies. We show through simulation studies that our proposed methods are robust from biases from unmeasured cluster-level confounders in a variety of multilevel observational studies. We also examine the effect of taking an algebra course on math achievement scores from the Early Childhood Longitudinal Study, a multilevel observational educational study, using our methods. The proposed methods are available in the CURobustML R package.

• Publisher: Springer US,Cambridge University Press
• Subject: Academic Achievement / Algebra / Artificial Intelligence / Assessment / Behavioral Science and Psychology / Bias / Causality / Child, Preschool / Children / College Science / Confounding Factors, Epidemiologic / Educational Background / Educational Equity (Finance) / Educational Finance / Family (Sociological Unit) / Humanities / Humans / Inferences / Law / Learning algorithms / Longitudinal Studies / Machine Learning / Mathematics / Mathematics Achievement / Mathematics Education / Mathematics Instruction / Maximum Likelihood Statistics / Observational studies / Psychology / Psychometrics / Statistical Theory and Methods / Statistics for Social Sciences / Student Experience / Students / Testing and Evaluation / Theory and Methods / Young Children