Heterogeneity in Longitudinal School Intervention Effects: Causal Inference with Latent Trajectory Types

• Activity Number: 103
• Type: Invited
• Date/Time: Monday, July 30, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Social Statistics Section
• Author(s): Booil Jo*+
• Companies: Stanford University
• Keywords: causal inference ; growth mixture modeling ; intervention ; longitudinal development

Abstract

Statistical analyses that identify heterogeneous outcome trajectories and trajectory-specific intervention effects may hold important treatment implications and useful information for future intervention trials. Growth mixture modeling (GMM) helps to identify subpopulations that develop into heterogeneous trajectory strata, as well as to identify heterogeneous intervention effects for these strata in the context of intervention/treatment studies. This latter use of GMM involves causal inference about the effect of the treatment. However, little is known about the possibility of formal causal inference in the GMM framework. In general, GMM is heavily dependent on empirical model fitting, and therefore typical strategies of contemporary causal modeling may not directly apply to causal inference using GMM. Nonetheless, the broad framework - defining identifying assumptions and conducting sensitivity analyses based on these assumptions - may well apply to causal inference using GMM. In this work, I examine the feasibility of adopting this broad framework for causal inference in the GMM framework.