Abstract
When using structural equation modeling, comparison across time or groups can be misleading if measures are not invariant. Partial least squares structural equation modeling (PLS-SEM) is a method widely used in business research, but its ability to test for measurement invariance is limited. This study introduces a comprehensive approach for measurement invariance testing in reflective measurement models in PLS-SEM. The methodology diverges from the traditional measurement invariance of composite models (MICOM) approach and expands the possibilities of measurement invariance testing in three areas: 1) providing statistical tests to validate the comparison of latent means across groups; 2) measurement invariance testing in longitudinal studies; and 3) the ability to simultaneously assess measurement invariance across multiple groups. Additionally, this study proposes a strategy to address measurement invariance rejections in large-sample studies. The paper offers guidelines for the MI tests, and an empirical example illustrates their utility in facilitating experimental approaches in PLS-SEM.
| Original language | English |
|---|---|
| Article number | 114581 |
| Journal | Journal of Business Research |
| Volume | 177 |
| ISSN | 0148-2963 |
| DOIs | |
| Publication status | Published - Apr 2024 |
Keywords
- Longitudinal
- Measurement
- Measurement invariance
- Multigroup analysis
- Partial least squares
- structural equation modeling
- Structural equation modeling