PLS-SEM methods, explained
Practical, plain-English guides to the methods and quality criteria behind a PLS-SEM analysis. Each guide explains what the metric means, how to read it, the common pitfalls, and how to run it in OpenPLS. More are on the way.
The HTMT ratio
What the heterotrait-monotrait ratio measures, the 0.85 and 0.90 thresholds, HTMT2, bootstrap inference, and how to assess discriminant validity.
Group comparisonMulti-group analysis (MGA)
How to test whether path coefficients differ across groups: the permutation test, measurement invariance (MICOM), reading p-values, and common pitfalls.
Predictive powerPLSpredict
Out-of-sample predictive power via k-fold cross-validation: Q²predict, the RMSE/MAE linear-model benchmark, CVPAT, and common pitfalls.
Interaction effectsModeration
When an effect depends on a third variable: the interaction effect, reading it and simple slopes, the two-stage approach, and common pitfalls.
PrioritizationIPMA
Importance-performance map analysis: combining total effects (importance) with rescaled scores (performance) to prioritize where to act.
MechanismMediation
How an effect works through a mediator: direct, indirect and total effects, full versus partial mediation, bootstrapping, and common pitfalls.
Hidden segmentsFIMIX-PLS
Uncover unobserved heterogeneity: latent segments with different path coefficients, fit criteria, entropy, and common pitfalls.
Hierarchical modelHigher-order constructs
Model an abstract concept through lower-order components: the four types, estimation approaches, validation, and common pitfalls.