Modern Modeling Methods (M3) Conference
Pre-Conference Workshop: May 19, 9:00-5:30 pm
Advances in Latent Variable Modeling using the new Mplus Version 7.2!
Dr. Bengt Muthén
Contact person and lecturer:
Bengt Muthen: firstname.lastname@example.org
Bengt Muthén obtained his Ph.D. in Statistics at the University of Uppsala, Sweden and is Professor Emeritus at UCLA. He was the 1988-89 President of the Psychometric Society and the 2011 recipient of the Psychometric Society’s Lifetime Achievement Award. He has published extensively on latent variable modeling and many of his procedures are implemented in Mplus. He is responsible for determining statistical directions for Mplus, formulating new models, methodological writing, and teaching.
This 1-day short course gives an overview of new latent variable modeling opportunities in the newly released Mplus Version 7.2. New features relate to mediation analysis, factor analysis, IRT, latent class modeling, and growth modeling. Examples with Mplus input and output are discussed. The emphasis is on modeling ideas and applications. Three topics will be given special emphasis: causal inference in mediation analysis, IRT modeling, and mixture modeling.
Causal inference in mediation analysis
Causal inference in mediation analysis offers counterfactually-based causal definitions of direct and indirect effects for models with categorical and count outcomes, drawing on research by Robins, Greenland, Pearl, VanderWeele, Imai and others. Background information is provided in the following papers on the Mplus website
Muthén, B. (2011). Applications of causally defined direct and indirect effects in mediation analysis using SEM in Mplus. www.statmodel.com
Muthén, B. & Asparouhov, T. (2014). Causal effects in mediation modeling: An introduction with applications to latent variables. Forthcoming in Structural Equation Modeling.
This type of modeling is still little known and seldom used among practitioners of mediation analysis. Part of the reason is that the literature is difficult to penetrate for non-experts. The latest version of Mplus offers ways to carry out such analyses in an easy way, including moderator analysis of both treatment-baseline interactions and treatment-mediator interactions. A unique feature of Mplus is the ability to do the causal analysis with latent variables. The aim of the workshop is to inform about these new analysis opportunities in an accessible way and discuss examples of how to perform the analyses in Mplus.
Item response theory modeling
Item response theory modeling in the general latent variable framework of the Mplus program offers many unique features. For an overview with examples and references, see
Muthén & Asparouhov (2013). Item response modeling in Mplus: A multi-dimensional, multi-level, and multi-timepoint example. Forthcoming in Linden & Hambleton (2013). Handbook of item response theory: Models, statistical tools, and applications. www.statmodel.com.
Asparouhov & Muthén (2013). New methods for the study of measurement invariance with many groups. Submitted for publication. www.statmodel.com.
The unique features include weighted least-squares, maximum-likelihood, and Bayes estimators for multidimensional analysis (Asparouhov & Muthén, 2012a); two-level, three-level, and cross-classified analysis (Asparouhov & Muthén, 2012b); mixture modeling (Muthén, 2008; Muthén & Asparouhov, 2009); multilevel mixture modeling (Asparouhov & Muthén, 2008; Henry & Muthén, 2010); and the alignment approach to the study of measurement invariance with many groups (Asparouhov & Muthén, 2013; Muthén & Asparouhov, 2013). The workshop gives an overview of these techniques with examples from mental health and achievement testing.
Mixture modeling includes latent class analysis, item response mixture modeling, factor mixture analysis, latent transition analysis, latent class growth analysis, growth mixture modeling, and survival mixture analysis. For an overview, see
Muthén, B. (2008). Latent variable hybrids: Overview of old and new models. In Hancock, G. R., & Samuelsen, K. M. (Eds.), Advances in latent variable mixture models, pp. 1-24. Charlotte, NC: Information Age Publishing, Inc. www.statmodel.com
Asparouhov, T. & Muthén, B. (2014). Skewed structural equation models and mixture models with continuous non-normal distributions. Mplus Web Note 19. www.statmodel.com.
The latest version of Mplus adds many new features to such analyses including 3-step methods for investigating the relationships between latent classes and auxiliary variables; residual correlations in latent class and latent transition modeling; and mixture modeling with non-normal distributions. The workshop gives an overview of these new techniques and discusses Mplus analysis of several examples.
This course is intended for both methodologists as well as applied researchers in the social and behavioral sciences. The presentation will be on a non-technical level where the modeling ideas are emphasized, but many of the ideas are novel and are therefore of interest to methodologists.
No internet connections or laptops are needed.