Structural equation models (SEM) and multilevel models (MLM) offer vast flexibility for analysing data. With these techniques, researchers can work with observed and latent variables in many dynamic ways to estimate direct, mediated/indirect, non-linear, and interaction effects. With Mplus, these effects can be embedded in larger models of many types.
In terms of SEM, special cases with only observed variables include multiple regression and path analysis. With observed and latent variables a special case is confirmatory factor analysis (CFA). In terms of MLM with clustered or nested data, such as with people in work teams or students in classrooms, latent variables are used to model random intercepts and random slopes. This allows estimating relationships at multiple levels of analysis. Importantly, both SEM and MLM can be used to model longitudinal processes and effects, for example with latent growth modeling (LGM). Further, SEM and MLM are special cases of a more general modeling framework that can be called multilevel SEM, which allows SEM at multiple levels of analysis.
Conveniently, Mplus can estimate all of these with both traditional and Bayesian estimation methods, allowing researchers a wide variety of options for hypothesis testing. This 6-day workshop will provide an overview of various models and estimators in Mplus, including how to interpret results.
At the conclusion of this course participants will have a working understanding of SEM, MLM, multilevel SEM, and LGM in Mplus. Real data and examples will be used throughout the course.
Participants will also gain a working understanding of traditional and Bayesian methods of estimation and inference in Mplus. This will be accomplished by covering the following topics, with a special 6th bonus day to cover LGM:
- Probability and Estimation, including regression and hypothesis testing with traditional p-values and sampling distributions versus posterior probabilities.
- Mplus, including modeling options, commands, model specification, and output
- Path Analysis, including model fit, interactions/moderation, mediation/indirect effects, and instrumental variables
- Structural Equation Modeling, including latent variables, confirmatory factor analysis, model identification, and multi-group models
- Multilevel Modeling, including random intercepts, random slopes, multilevel path analysis, and multilevel structural equation modeling
- Latent growth modeling, including longitudinal data and processes, modeled as an MLM, modeled as an SEM, and extensions for more dynamic models
Participants are encouraged to bring a laptop. All models will be estimated and interpreted during the course so a laptop is not necessary, but participants will find it very helpful to use Mplus software.
- Probability and estimation
- Hypothesis testing
- Mplus: Modeling options, commands, and output
Day 2 - Path analysis
- Specification and estimation
- Model fit, selection, and equivalence
- Mediation/indirect effects & instrumental variables
Day 3 - SEM
- Latent variables
- Confirmatory factor analysis
- Structural equation modeling
- Model identification
Day 4 - MLM
- Multilevel data and effects
- Random intercepts
- Random slopes
- Cross-level interactions
Day 5 - Multilevel SEM
- Multilevel path analysis
- Multilevel mediation
- Multilevel confirmatory factor analysis
- Multilevel structural equation modeling
Day 6 - LGM
- Longitudinal processes and data
- Latent growth models as Multilevel models
- Latent growth models as SEMs
- Dynamic latent growth modeling
Who should attend?
This course is targeted at researchers and PhD students from social, health, and physical sciences with a working knowledge of regression. A review of regression and the logic of hypothesis testing before the Mplus course will greatly enhance uptake for those who are rusty with basic the concepts. An introduction to these topics will be given at the start of the course.
February 2015 TBC
Dr. Michael Zyphur
+61 3 8344 1905
The last Mplus course taught by Michael Zyphur was attended by 70 people from around the globe. Average participant ratings were 4.8/5 on a questionnaire assessing the overall quality of the course.
This was an exceptional course - educational and entertaining on what is a difficult subject otherwise. Michael Zyphur should be commended for his energy and grasp of the material over the 6 days.
Really well organised. Mike is an excellent presenter - really animated and responsive to questions. He made a tough/dry subject fun and interesting. A great course.
Exactly what I needed for my research. Very engaging presenter.
Great coverage of different techniques. Topics covered are abundant. Great illustrations with examples. Quite a practical approach (not only theoretical) especially with the examples in Mplus. Lots of materials that can be reviewed post workshop for self-learning.
Loved the course. Mike was engaging and made the content easily accessible. Very pleased to have references provided for further learning and feel like I've gained a thorough base understanding of the material that can now be built on (with greater enthusiasm than I expected).