SmartPLS Algorithms and Techniques

Estimation & Core Algorithm

• PLS-SEM Algorithm — the standard PLS-SEM estimation algorithm and its settings.
• Weighted PLS Algorithm (WPLS) — incorporates sampling weights so results better represent your target population.
• Consistent PLS-SEM (PLSc-SEM) — corrects PLS-SEM estimates to approximate common-factor (CB-SEM-like) results.
• Principal Component Analysis (PCA) — reduces a set of indicators to their principal components.

Resampling & Inference

• Bootstrapping — the standard resampling procedure for testing the significance of path coefficients, loadings, and weights.
• Consistent Bootstrapping — bootstrapping for consistent PLS-SEM (PLSc-SEM) models.
• Blindfolding — legacy procedure for in-sample predictive relevance (Q²), superseded by PLSpredict and CVPAT.
• Permutation — nonparametric significance testing used for multigroup comparisons and measurement invariance (MICOM).
• Consistent Permutation — permutation testing for consistent PLS-SEM (PLSc-SEM) models.
• Cross-validated Predictive Ability Test (CVPAT) — statistical test for comparing the out-of-sample predictive power of competing models.

Validity, Reliability & Model Fit

• Confirmatory Composite Analysis (CCA) — tests whether composites adequately represent the assumed constructs.
• Confirmatory Tetrad Analysis in PLS (CTA-PLS) — tests whether a construct is better modeled as reflective or formative.
• Discriminant Validity Assessment and HTMT — the heterotrait-monotrait ratio of correlations for discriminant validity testing.
• Goodness of Fit (GoF) — a legacy overall fit measure; we explain why we generally advise against relying on it.
• Model Fit — fit indices adapted from CB-SEM (e.g., SRMR) for PLS-SEM models.
• Model Comparison — compares competing PLS-SEM models using predictive and information-theoretic criteria.
• Prediction-oriented Model Selection — a Bayesian-information-criterion-based approach for selecting among competing models.

Heterogeneity & Multigroup Analysis

• Measurement Invariance Assessment (MICOM) — establishes measurement invariance, a prerequisite for meaningful multigroup comparisons.
• Multigroup Analysis (MGA) — tests whether path coefficients differ significantly between groups.
• Consistent Multigroup Analysis (MGA) — multigroup analysis for consistent PLS-SEM (PLSc-SEM) models.

Prediction & Segmentation

• PLSpredict — evaluates a model's out-of-sample predictive power.
• PLS Prediction-oriented Segmentation (PLS-POS) — a distance-based approach for identifying latent classes of respondents.
• Finite Mixture Partial Least Squares (FIMIX-PLS) — a probabilistic approach for uncovering unobserved heterogeneity through latent class segmentation.
• Importance-performance Map Analysis (IPMA) — combines construct importance and performance to identify priorities for improvement.

Extended Relationships

• Moderation — tests whether a third variable changes the strength or direction of a relationship.
• Mediation — tests indirect effects that explain how or why an effect occurs.
• Nonlinear Relationships — models quadratic and other nonlinear effects between constructs.
• Higher-order Models — models constructs composed of several lower-order sub-dimensions.
• Endogeneity and Gaussian Copulas — an approach for detecting and correcting for endogeneity without instrumental variables.

Generalized Structured Component Analysis (GSCA)

• GSCA algorithm and bootstrapping

Regression, Path Analysis and PROCESS

• Logistic Regression
• Necessary Condition Analysis (NCA)
• Path Analysis and PROCESS
• Path Analysis and PROCESS Bootstrapping
• Regression
• Regression Bootstrapping

CB-SEM and CFA

• CB-SEM - Covariance-based Structural Equation Modeling (CB-SEM)
• CB-SEM Bootstrapping
• CB-SEM Model Comparison
• CB-SEM Multigroup Analysis (MGA)
• CB-SEM Measurement Invariance Assessment
• CB-SEM Moderator Analysis
• Confirmatory Factor Analysis (CFA)
• Principal component analysis (PCA)