An empirical comparison of generalized structured component analysis and partial least squares path modeling under variance-based structural equation models

The article compares four ways of estimating variance-based structural equation models: GSCA using reflective or formative indicators and partial least squares path modeling using Mode A or Mode B. A proposed data-generation procedure allows these approaches to be evaluated within a common framework. Monte Carlo experiments assess parameter recovery and statistical power. Although estimators from all four approaches were consistent under the examined conditions, GSCA with reflective indicators showed the greatest estimation efficiency.