Module 9: PLS-SEM Structural Model

Purpose

After the measurement model is acceptable, researchers assess the structural model: collinearity, path coefficients, explanatory power, effect sizes, and significance.

Learning Objectives

After this module, learners should be able to:

• Assess collinearity using VIF.
• Interpret path coefficients.
• Assess R-square and adjusted R-square.
• Interpret f-square effect sizes.
• Use bootstrapping for path significance.

Assessment Sequence

1. Check structural collinearity.
2. Assess path coefficients.
3. Assess coefficient significance using bootstrapping.
4. Assess R-square and adjusted R-square.
5. Assess f-square effect sizes.
6. Assess predictive performance when relevant.

Collinearity

Check VIF for predictor constructs.

Common guideline:

VIF < 3.3 or VIF < 5.0 depending on discipline and source

High VIF suggests overlapping predictors and unstable estimates.

Path Coefficients

Path coefficients are standardized regression-like effects between constructs.

Interpret:

• Direction
• Magnitude
• Significance
• Theoretical meaning

R-Square

R-square is the explained variance of an endogenous construct.

Always interpret R-square in context. A low R-square can still be meaningful in behavioral research, while a high R-square does not prove causality.

f-Square

f-square assesses how much an exogenous construct contributes to an endogenous construct's R-square.

Common rough guideline:

0.02 small
0.15 medium
0.35 large

Structural Reporting Template

The structural model was assessed after confirming measurement quality. Predictor collinearity was not problematic because all VIF values were below the selected threshold. Bootstrapping with ... subsamples showed that [path] was significant (beta = ..., t = ..., p = ...), supporting H1. The model explained ...% of the variance in [endogenous construct] (R2 = ...).

Checklist

Use Structural Model Checklist.

Next Module

Continue to Module 10: Prediction and PLSpredict.