Consistent multigroup analysis is the multigroup analysis (MGA) procedure applied to models estimated with the consistent PLS-SEM (PLSc-SEM) algorithm rather than the standard PLS-SEM algorithm. Like standard MGA, it lets researchers test whether pre-defined data groups, such as countries or customer segments, differ significantly in their group-specific parameter estimates. Researchers use consistent MGA when their model has been estimated with PLSc-SEM and they need a group comparison based on the disattenuated, factor-consistent PLSc-SEM estimates rather than standard PLS-SEM results.

Consistent MGA follows the same logic as standard MGA: group-specific results are compared using the permutation procedure or bootstrapping-based tests (bias-corrected confidence intervals, the PLS-MGA test, a parametric test, and the Welch-Satterthwaite test). The key difference is that every group-specific estimation underlying these tests is computed with the consistent PLS-SEM (PLSc-SEM) algorithm instead of the standard PLS-SEM algorithm, so the resulting comparisons reflect the disattenuated, factor-consistent parameter estimates that PLSc-SEM produces. As with standard MGA, group comparisons are only meaningful once measurement invariance has been established; the consistent permutation procedure supports the MICOM test for PLSc-SEM models.

In SmartPLS, consistent MGA is available whenever a multigroup comparison is run on a model that uses the PLSc-SEM algorithm. It offers the same group-selection and significance-testing settings as standard MGA, applied on top of the PLSc-SEM estimation for each group.

Standard MGA compares group-specific results from the standard PLS-SEM algorithm. Consistent MGA applies the same comparison logic, but each group is estimated with the consistent PLS-SEM (PLSc-SEM) algorithm, so it is used for group comparisons in PLSc-SEM models.

Use consistent MGA when your model has been estimated with PLSc-SEM. Since PLSc-SEM produces different (factor-consistent) parameter estimates than standard PLS-SEM, group comparisons for such a model should also be based on PLSc-SEM re-estimations.

Yes. As with standard MGA, group comparisons are only meaningful once measurement invariance has been established. The consistent permutation procedure supports the MICOM test for PLSc-SEM models.