Online Program Home
My Program

Abstract Details

Activity Number: 472 - Statistical Methods for Causal Inference
Type: Contributed
Date/Time: Wednesday, July 31, 2019 : 8:30 AM to 10:20 AM
Sponsor: Section on Statistics in Epidemiology
Abstract #305167 Presentation
Title: A Two-Stage Estimation Procedure for Nonlinear Structural Equation Models
Author(s): Esben Budtz-Jorgensen* and Klaus Holst
Companies: University of Copenhagen Dept. of Biostat and Mærsk
Keywords: latent variable; structural equation; non-linear effect; distributional assumptions

Maximum likelihood (ML) estimation in non-linear structural equation with latent variables require numerical integration and results are sensitive to distributional assumptions. In this talk, we introduce a two-stage technique for estimation of non-linear associations between latent variables. Here both steps are based on fitting linear structural equation models: first a model is fitted to data measuring the latent predictor and terms describing the non-linear effect are predicted. In the second step, these predictions are included in a model for the latent outcome variable. We show that this procedure is consistent and illustrate that it allows the association between latent variables to be modeled using restricted cubic splines. We also discus robustness and compare the method to relevant alternatives including ML.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program