PLS-SEM
Estimate a justified composite path model
PLS-SEM estimates linked constructs using component-based algorithms and is often used for composite modelling and prediction-oriented questions. Reflective factors and formative composites imply different measurement theories and evaluation criteria. PLS is not a universal substitute for covariance-based SEM, and claims of automatic suitability for small samples or any distribution are misleading.
Use when the construct specification and modelling objective justify a composite-based approach. Plan power and precision, distinguish formative from reflective measurement, and assess predictions beyond the estimation sample where relevant.
Strengths
- Accommodates explicitly justified composite constructs
- Supports structural modelling with predictive evaluation
Limitations
- Small samples can still produce unstable estimates
- Measurement-model misspecification undermines path interpretation
Know the boundary
PLS-SEM does not remove the need for theory, adequate information or causal identification.