My name is Adam Garber and I am a PhD student in Education at the University of California, Santa Barbara.The purpose of this website is to make SEM modeling accesible for applied researchers and students. My work focuses on finite mixture modeling in-line with my advisor Dr. Karen Nylund-Gibson's research (Latent_Variable_Group). My other interests are Autism and Special Education Research. In particular, the benifits of physical activity for children with disabilities.
- The R package {
MplusAutomation} is used for creating organized project workflows (Hallquist & Wiley, 2018) - All models are estimated in Mplus allowing for highly flexible SEM model specification (Muthén & Muthén, 1998-2017)
- R-Projects and the {
here} package allow for reproducibility across operating systems. - The {
tidyverse}'s highly coherent functions are used whenever possible to increase accessibility for applied audiences (Wickham et al., 2019)
... instead of estimating SEM models with R packages?
- Currently, most SEM packages in R such as {
lavaan} have significant limitations in their capacity for flexibly specifying the full range modeling approaches available in SEM software (i.e., Mplus). This includes the ability to specify categorical latent variables (LCA/LPA/LTA), multi-level models (MLM), non-normal outcomes (GLM), and their combinations (i.e., multi-level LTA with covariates & distals). - Other packages, although highly flexible {
OpenMX} are less accessible to applied researchers.
... instead of doing all modeling entirely in Mplus?
- Running complex models in Mplus can be error prone due to the workflow involving a proliferation of files and reliance on other software (Excel, SPSS).
- In practice, this results in organizational challenges and lots of copying and pasting.
- Importantly, R provides convenient methods for conducting fully reproducible research projects.
- In R documenting all research decisions and data presentation is straightforward from start to finish (e.g., data cleaning, transformation, re-coding, plotting, and table construction).
Data examples: All lab exercises utilize public-use data repositories.
- Lab 1: Getting Started with MplusAutomation - Explore and Prepare Data. Run Models. $\color{darkred}{\text{(PDF)}}$
- Lab 2: Exploratory Factor Analysis (EFA) - Univariate and Multivariate Diagnostics.
- Lab 3: EFA - R coding excercise: What's Missing?
- Lab 4: Splitting samples, Iterators, and Dealing with Large Datasets
- Lab 5: EFA Rotation - Confirmatory Factor Analysis (CFA) - Path Diagrams
- Lab 6: CFA Roulette - A Game of Chance
- Lab 7: Principle Component Analysis - PCA
- Lab 8: Multiple Indicator, Multiple Causes, MIMIC Models $\color{darkred}{\text{(PDF)}}$
- Lab 9: Measurement Invariance $\color{darkred}{\text{(PDF)}}$
- Lab 10.1: Weighted Least Squares Estimator - EFA, CFA, Invariance $\color{darkred}{\text{(PDF)}}$
- Lab 10.2: Higher Order Factor Models $\color{darkred}{\text{(PDF)}}$
- Lab 1: Path Analysis - Multiple Indirect Path Model - Single Indicator Factor Model $\color{darkred}{\text{(PDF)}}$
- Lab 2: Competing Path Models $\color{darkred}{\text{(PDF)}}$
- Lab 3: Moderation $\color{darkred}{\text{(PDF)}}$
- Lab 4: Mediation $\color{darkred}{\text{(PDF)}}$
- Lab 5: Conditional indirect effects (moderated mediation) $\color{darkred}{\text{(PDF)}}$
- Lab 6: Latent growth models $\color{darkred}{\text{(PDF)}}$
- Lab 7: Advanced growth models $\color{darkred}{\text{(PDF)}}$
- Lab 8: Introduction to mixture models (LCA) - Enumeration and Plotting
- Lab 9: Mixture Models with Covariates and Distal Outcomes - Manual Three Step Approach
- Lab 10.1: Observed Response Patterns in Latent Class Analysis
- Lab 10.2: Introduction to Latent Profile Analysis
- Lab 1: Tidy Enumeration with MplusAutomation - Latent Class Analysis (LCA)
- Random Intercept Latent Transition Analysis (RI-LTA): Replication of analyses presented in Muthén & Asparouhov (2020)
- Multi-level Latent Class Analysis (MLCA) : Replication of models presented in Henry & Muthén (2010) $\color{darkred}{\text{(PDF)}}$
- Monte Carlo Simulation of Power & Sample Size: Replication of examples presented in Muthén & Muthén (2002)
Hallquist, M. N., & Wiley, J. F. (2018). MplusAutomation: An R Package for Facilitating Large-Scale Latent Variable Analyses in Mplus. Structural equation modeling: a multidisciplinary journal, 25(4), 621-638.
Henry, K. L., & Muthén, B. (2010). Multilevel latent class analysis: An application of adolescent smoking typologies with individual and contextual predictors. Structural Equation Modeling, 17(2), 193-215.
Horst, A. (2020). Course & Workshop Materials. GitHub Repositories, https://https://allisonhorst.github.io/
Muthén, B. & Asparouhov, T. (2020). \textcolor{blue}{Latent Transition Analysis with Random Intercepts (RI-LTA)}. Under review. Version 3.
Muthén, L.K. and Muthén, B.O. (1998-2017). Mplus User’s Guide. Eighth Edition. Los Angeles, CA: Muthén & Muthén
Muthén, L. K., & Muthén, B. O. (2002). \textcolor{blue}{How to use a Monte Carlo study to decide on sample size and determine power.} Structural equation modeling, 9(4), 599-620.
R Core Team (2017). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-project.org/
Wickham et al., (2019). Welcome to the tidyverse. Journal of Open Source Software, 4(43), 1686, https://doi.org/10.21105/joss.01686
