Multilevel Conference

Post-conference: ‘Multilevel SEM Using the lavaan Package’

The post-conference workshop on ‘An Overview of Multilevel Structural Equation Modeling Using the lavaan Package on April 14 is taught by Terrence D. Jorgensen.

Note that this workshop will be taught online.

The open-source R package lavaan provides a user-friendly interface for researchers to specify and fit structural equation models (SEMs) to their data (Rosseel, 2012).  Since version 0.6 (May 2018) lavaan has also provided functionality for multilevel continuous data. This workshop will demonstrate the current functionality for MLSEMs in lavaan (as of version 0.6-8), including:

  • cluster-robust solutions for single-level analyses
  • two-level SEMs with random intercepts
  • alternative maximum likelihood estimators (quasi-Newton vs. expectation–maximization algorithms)
  • full-information maximum likelihood (FIML) estimation for incomplete data
  • evaluating level-specific model fit and reliability
  • measurement equivalence across clusters
  • the wide-format approach
  • random slopes

Software syntax and data for all examples will be provided via the Open Science Framework (OSF) .  An overview of multilevel features still under development will also be provided.  Participants are expected to have the latest software installed and to be familiar with R and the basics of lavaan, as covered in Rosseel’s (2012) Journal of Statistical Software article or the tutorial pages on the lavaan website.

This workshop has a morning session from 9:00 till 12:00 and an afternoon session from 13:00 till 16:00.

Note that you can also register for this workshop without participating in the conference.