How do I do PLS-SEM in SmartPLS 4?

Answer 1

Hey! Here's something that I thoguht based on how your question was asked and this might be what you're looking for. Some steps that I think you can follow are:

1. Project Setup & Data Import:

1. Create a New Project: Open SmartPLS 4, click "New Project" in the "Projects" panel, and name it.
2. Import Data: Right-click your project, select "Import Data File," choose your .csv, .txt, .sav, or .xlsx file. Review and confirm data types (Metric, Ordinal, Nominal) and missing values in the import dialog.

2. Build Your PLS Path Model:

1. Create New Model: Select your project, click "PLS-SEM" in the toolbar, and name your model.
2. Add Constructs & Indicators: Drag indicators from the "Indicators" list (left panel) onto the modeling canvas. As you drop the first indicator for a concept, name the latent variable (construct). SmartPLS defaults to reflective (arrows from construct to indicators); if formative (arrows from indicators to construct), right-click the construct and select "Invert measurement model."
3. Draw Paths: Use the "Connect" tool (arrow icon) to draw hypothesized relationships (arrows) between your constructs.
4. Arrange: Organize your model visually for clarity.
5. Save: Regularly save your model.

3. Run PLS-SEM Algorithm:

1. Calculate: In the toolbar, click "Calculate" and then select "PLS-SEM Algorithm."
2. Default Settings: The default algorithm settings are typically suitable for most analyses. Ensure "Open report" is checked.
3. Start: Click "Start calculation."

4. Interpret Results - First, the Measurement Model: (This assesses the reliability and validity of your constructs)

1. Outer Loadings: (Under "Outer Loadings" in report) Check if indicators load strongly on their respective constructs. Aim for > 0.708.
2. Convergent Validity: (Under "Construct Reliability and Validity")
3. Average Variance Extracted (AVE): Should be > 0.50.
4. Construct Reliability: (Under "Construct Reliability and Validity")
5. Composite Reliability (CR): Should be > 0.70.
6. Cronbach's Alpha (α): Should be > 0.70.
7. Rho_A (ρA​): Should be > 0.70.
8. Discriminant Validity: (Under "Discriminant Validity")
9. Heterotrait-Monotrait Ratio (HTMT): Most important. Values should be < 0.90 (or < 0.85).
10. Formative Construct Evaluation: (If applicable)
11. Outer Weights: Evaluate significance and magnitude.
12. Collinearity (VIF): (Under "Collinearity Statistics (VIF)") Indicator VIFs for formative constructs should be < 5 (ideally < 3).

5. Interpret Results - Second, the Structural Model: (This assesses the relationships between your constructs)

1. Path Coefficients (β): (Under "Path Coefficients") These represent the strength and direction of relationships.
2. R-squared (R2): (Under "R-squared") Indicates the variance explained in endogenous constructs (e.g., 0.25 weak, 0.50 moderate, 0.75 substantial, depending on field).
3. Predictive Relevance (Q2): (Run "PLS Predict" algorithm first, then check "Predictive Relevance (Q2)") Values > 0 indicate predictive relevance.

6. Run Bootstrapping for Significance Testing:

1. Return to Model: Go back to your model drawing ("Edit" button).
2. Calculate: Click "Calculate" again, then select "Bootstrapping."
3. Settings: Set subsamples to 5,000 or 10,000. Choose "Two-tailed" for general hypotheses.
4. Start: Click "Start calculation."

7. Interpret Bootstrapping Results:

1. Path Coefficients (t-values and p-values): (Under "Path Coefficients")
2. t-values > 1.96 (for a two-tailed 0.05 significance level) and p-values < 0.05 indicate statistical significance.
3. Check confidence intervals (should not include zero).
4. Specific Indirect Effects: (If mediating variables are present) Check this section for the significance of indirect paths.

8. Reporting Your Results: When writing up your findings, present them clearly:

1. Measurement Model: Tables for reliability, validity, and discriminant validity.
2. Structural Model: Table for path coefficients (β), t-values, p-values, and R2.
3. Discuss the implications of your findings, supported hypotheses, and any limitations.