Short Course: Structural Equation Modeling Using R and SAS

.

Thursday (September 5): 08:30 am – 13:00 pm

.

Abstract:

Originated from social sciences, structural equation modeling (SEM) is becoming more popular in other fields such as education, health science, and medical sciences. This short course is aimed to provide an overview of SEM and to demonstrate its applications by using R and SAS software based on the newly published book: “Chen and Yung (2023). Structural Equation Modeling Using R/SAS: A Step-by-Step Approach with Real Data Analysis. Chapman and Hall/CRC”. We will cover some main SEM topics, including path analysis, confirmatory factor analysis, structural relations with latent variables, and latent growth-curve modeling. Real-world application examples, most of which are based on our newly published SEM book (see reference), are compiled to demonstrate SEM in social, educational, behavioral, and marketing research.  Mathematical and statistical foundations of SEM are discussed at a level suitable for general understanding.  This course is designed for statisticians and data analysts who like to learn SEM techniques for their own research and applications. Both R package “lavaan” (latent variable analysis) and the CALIS procedure of SAS/STAT will be used to demonstrate model specifications, fitting, and result interpretations.

Attendees should have a basic understanding of regression analysis. Experience using R and SAS software is not required for understanding the general SEM techniques.

.

Outline (Two 2-hr Sessions):

.

Session 1 (2-hr)

  1. Overview of structural equation modeling
  • Historical background
  • Path analysis and SEM as an extension of regression analysis
  • Path diagram representations
  • Measurement errors in regressors
  • Confirmatory factor models for instrument validations
  • Combining measurement models and structural models for latent variables.

2. Statistical and mathematical backgrounds of structural equation modeling

  • Functional equations and matrix formulations
  • Estimation of parameters: Maximum-likelihood, generalized least squares, and asymptotically distribution-free method

.

Session 2: (2-hr)

  1. Latent growth-curve modeling
  • Latent growth-curve modeling with a single outcome
  • Latent growth-curve modeling with multiple outcomes
  • Latent growth-curve modeling with covariates
  • Illustrations using the Cancer Surgery data

.

4. Using SEM to assess direct, indirect, and total effects (if time permits)

  • Definitions of direct, indirect, and total effects
  • Why SEM is useful to study these effects
  • Illustration: How mental abilities are affected remotely by social status

.

Data analysis with SEM using R/lavaan and PROC CALIS

.

Presenters

Dr. (Din)Ding-Geng Chen is an elected fellow of American Statistical Association and an elected member of the International Statistical Institute. Currently he is the executive director and professor in biostatistics at the College of Health Solutions, Arizona State University. Dr. Chen is also an extraordinary professor and the SARChI research chair in biostatistics, at the Department of Statistics, University of Pretoria, South Africa, and an honorary professor at the School of Mathematics, Statistics and Computer Science, University of KwaZulu-Natal, South Africa. He was the Wallace H. Kuralt distinguished professor in Biostatistics in the University of North Carolina-Chapel Hill, a professor in Biostatistics at the University of Rochester, and the Karl E. Peace endowed eminent scholar chair in biostatistics at Georgia Southern University. He is also a senior statistics consultant for biopharmaceuticals and government agencies with extensive expertise in Monte-Carlo simulations, clinical trial biostatistics and public health statistics. Dr. Chen has more than 200 referred professional publications, co-authored and co-edited 40 books on clinical trial methodology and analysis, meta-analysis, data sciences, causal inferences and public health applications. He has been invited nationally and internationally to give speeches on his research.

Dr. Yiu-Fai Yung is an analytic solution manager at the SAS Institute Inc. He has been developing commercial software for causal analysis, factor analysis, and structural equation modeling for more than 20 years. He has held several workshops and taught courses about causal analysis and structural equation modeling in conferences such as SAS Users’ Group meetings, Joint Statistical Meetings, and International Meetings of Psychometric Society. Prior to joining SAS, he taught psychological and behavioral statistics at the University of North Carolina at Chapel Hill. He has published articles in Psychometrika, British Journal of Mathematical and Statistical Psychology, and Journal of Educational and Behavioral Statistics.  His main research interests include latent variable modeling, mixture modeling, mediation analysis, and causal inferences.