Multilevel Structural Equation Modeling
A 5-Day Seminar Taught by Kristopher Preacher, Ph.D.
To see a sample of the course materials, click here.
Hierarchically clustered (multilevel or nested) data are common in the social sciences, medical fields, and business research. Clustered data violate the assumption of independence required by ordinary statistical methods. Increasingly complex research designs and hypotheses have created a need for sophisticated methods that go beyond standard multilevel modeling (MLM). This course will introduce a variety of extensions to MLM, including cutting-edge multilevel structural equation modeling (MSEM) to handle complex designs and modeling objectives. Throughout the workshop, empirical examples will be presented to illustrate key concepts. A strong background in structural equation modeling (SEM) is not necessary.
Please note, this course will include all material from the 2-day Multilevel Modeling: A Second Course.
On Day 1 we will begin by reviewing the basics of MLM. Next, Mplus will be introduced as a flexible and powerful software environment for fitting basic and advanced multilevel models. Then we will cover several advanced MLM topics.
Basic MLM topics include:
- The motivation for MLM
- Key concepts
- Equation conventions
- The univariate two-level MLM with fixed and random coefficients
Advanced MLM topics include:
- Conducting power analysis for MLM using a general Monte Carlo technique
- Fitting multivariate multilevel models
- Modeling cross-classified data
On Day 2 multilevel structural equation modeling will be introduced as a general approach for more complex modeling tasks. After a brief overview of single-level SEM, we will turn to the development of MSEM and the important advantages of MSEM over MLM (e.g., inclusion of latent variables, complex causal pathways, upper-level outcomes, and model fit assessment). Standard SEM and MLM will be recast as special cases of MSEM. Next we will cover a variety of MSEM topics:
- Multilevel path analysis
- Multilevel exploratory and confirmatory factor analysis
- Model fit in MSEM
On Days 3-5 we will continue to explore special applications of MSEM. Advanced topics will include:
- Multilevel structural models with latent variables
- Applications to three-level (and higher-level) data
- Multilevel reliability estimation
- Multilevel mediation analysis
- Multiple group models
- Estimating, plotting, and probing interaction effects
- Moderation in MLM and MSEM
- Modeling discrete (e.g., binary, count) dependent variables
- Interval estimates for nonnormal statistics
- Handling convergence problems: A bag of tricks
- Conducting Monte Carlo simulation studies
Days 3-5 will also involve small group exercises to get practice using Mplus to fit models and conduct power analyses and Monte Carlo studies. Informal homework assignments will be given on Days 1-4, and discussed the following morning.
Throughout the five-day course, models will be presented in several formats—path diagrams, equations, and software syntax. Data and Mplus syntax for all of the examples will be included in the workshop materials.
Participants in this seminar can expect to gain:
- Mastery of advanced topics in MLM
- A deeper understanding of the relationship between MLM and SEM
- The ability to use multilevel SEM to test complex structural hypotheses
- Resources to conduct power analysis for virtually any multilevel design
- The ability to fluently interpret and translate among path diagrams, model equations, and Mplus syntax for advanced MLM and MSEM
- Strategies for tackling convergence problems and estimation errors
- Programming skills for conducting Monte Carlo studies to assess model feasibility prior to data collection.
- Documented Mplus syntax templates for fitting a variety of models to multilevel data.
Who should attend?
This seminar is designed for researchers who have had some exposure to multilevel modeling and/or structural equation modeling (e.g., from seminars, workshops, or courses) and who want to deepen and extend their knowledge. At a minimum, participants should have a good working knowledge of basic principles of statistical inference (e.g., standard errors, hypothesis tests, confidence intervals), and should also have a good understanding of the theory and practice of linear regression.
Mplus will be used for all worked examples, but prior knowledge of Mplus is not essential. You are welcome and encouraged to bring your own laptop computer with Mplus installed (including the multilevel or combination add-on). However, this is not required. Participants will still benefit from the comprehensive set of slides and syntax that they can apply at home.
Location, Format, and materials
The class will meet from 9 am to 5 pm each day with a 1-hour lunch break at Temple University Center City, 1515 Market Street, Philadelphia, PA 19103.
Participants receive a bound manual containing detailed lecture notes (with equations and graphics), examples of computer printout, and many other useful features. This book frees participants from the distracting task of note taking.
Registration and Lodging
The fee of $1795.00 includes all course materials. The early registration fee of $1595.00 is available until July 2.
Lodging Reservation Instructions
A block of guest rooms has been reserved at the Club Quarters Hotel, 1628 Chestnut Street, Philadelphia, PA at a special rate of $137 per night. This location is about a 5 minute walk to the seminar location. In order to make reservations, call 203-905-2100 during business hours and identify yourself by using group code STA729 or click here. For guaranteed rate and availability, you must reserve your room no later than Friday, June 29, 2018.
If you make reservations after the cut-off date, ask for the Statistical Horizons room rate (do not use the code) and they will try to accommodate your request.
“This is an excellent workshop. I have learned more in a week than in months (years!) of working with Mplus on my own. The content and structure of the course, including his presentation, exercises, discussion, and consultation is excellent – never a dull moment! Kris is also an amazing educator; his mastery of the topics is impressive and his availability, kindness, and eagerness to help is second to none. Also remarkable is how he answers everyone’s questions and how he communicates the materials. I think his course is a must for anyone who wants to accurately run MSEM.”
Elizabeth Grandfield, University of Kentucky
“This was my first statistical workshop post graduate school, so I wasn’t sure what to expect. I was pleasantly surprised to find it fun as well as incredibly helpful. I went in with a specific dataset and analysis in mind, and walked out with all the tools I needed to complete for publication, as well as new directions to keep learning on my own. The material was accessible and we were given plenty of help trouble-shooting and working with real-life problems.”
Meredith Martin, University of Nebraska-Lincoln
“The course helped me solidify statistical knowledge as well as learn new knowledge and skills.”
Wenjuan Ma, Michigan State University
“I came into this course with minimal statistical training, especially in multilevel models or structural equation modeling, and I had never used Mplus so I was very nervous about how the course would go. Dr. Preacher is a phenomenal teacher and I have learned and understood more than I ever expected to! I have gotten a lot out of this course and he made complex topics very accessible.”
Rebecca Lipschutz, Tulane University – School of Medicine
“This was a great course on a topic that was not covered in my graduate courses. Dr. Preacher is an excellent teacher!”
Robert Nichols, Ohio State University
“Excellent course! Professor Preacher does a masterful job of communicating a complex subject matter in an easy to digest, stepwise manner. Would recommend the course without hesitation.”
Shikhar Sarin, Boise State University
“This course is state-of-the-art and features cutting edge statistical techniques for modeling hierarchical data. It’s a must take course for those interested in testing simple to complex structural equation models with nested data.”
Alan K. Goodboy, West Virginia University