Skip to main content

Resources

Fit Indices for Multiple-Group Measurement Invariance (MI) (in progress)

  • This project is in the early stages. How sensitive are fit indices to different violations of metric and scalar MI? Can focusing on residuals separated by type (correlations, variances, means) help here? Are nested fit indices better than absolute ones? This Shiny app begins to explore this.

Composite-Level Measurement Invariance (Savalei, Lemmons, & Brace, 2026)

  • A GitHub repository with a Shiny app and sample lavaan code to implement the method.
  • A link to a preprint (coming soon).

Are SEM fit indices sensitive to cross-loadings? (Savalei & Huang, 2025)

  • Suppose you fit a CFA model (perfect simple structure) to EFA-like data (lots and lots of cross-loadings). Surely your fit indices will tell you your “clean” model is wrong? Right? Wrong. Download and run this Shiny app to explore this phenomenon.

Break coefficient alpha! (Savalei & Reise, 2019)

  • Have you heard that alpha is not appropriate for congeneric (one-factor) data, and omega should be used? Well, actually, alpha is extremely robust to violations of tau-equivalence! You can use this Shiny app to try to “break” it! Good luck.
  • The associated paper is here.

R Functions for Mardia’s Multivariate Kurtosis With Missing Data

  • R functions to compute Mardia’s multivariate kurtosis with missing data, based on the paper by Yuan, Lambert, and Fouladi (2004), are available in this GitHub repository.