A GUI Alternative to Hayes #Process Macro in #SmartPLS4
Based on Research With Fawad's video on YouTube. If you like this content, support the original creators by watching, liking and subscribing to their content.
SmartPLS4’s Process option provides a GUI workflow for Hayes-style mediation models, reducing reliance on SPSS coding-based Process Macro setups.
Briefing
SmartPLS4’s Process option offers a graphical, GUI-based way to run Hayes-style mediation models—without relying on SPSS Process Macro’s common “one IV, one DV” constraint. Instead of coding or switching tools, researchers can build mediation paths directly inside SmartPLS4, including models with multiple independent variables (IVs) and multiple dependent variables (DVs), which is a key limitation when Process Macro is used in SPSS.
The workflow starts in SmartPLS4 by using the Process toolbar. A new process project is created, then the researcher assigns constructs to the model roles: an independent variable (example: “leadership style”), a mediator (example: “knowledge initiation”), and a dependent variable (example: “sustainable competitive advantage”). Indicators for each construct are positioned in the model builder, and the mediation structure is specified by linking IV → mediator, mediator → DV, and IV → DV.
To test a specific Hayes model—such as Model 4 for mediation—the interface provides a model list (commonly available through Hayes Process documentation). Once the model is selected, results can be generated through either path analysis or bootstrapping. Initial path coefficient output may show estimates but not significance testing (no T statistics or P values), so the recommended step is bootstrapping.
Bootstrapping is run with 10,000 resamples to stabilize estimates, using bias-corrected and accelerated (BCa) bootstrap intervals. The example also uses a one-tailed test setting, a fixed seed for reproducibility, and standardized results. After bootstrapping, the output includes path coefficients with significance values for each direct relationship and a separate test for the indirect (mediated) effect.
In the mediation example, all three paths are significant: the IV to mediator link, the mediator to DV link, and the direct IV to DV link. The specific indirect effect (IV → mediator → DV) is also significant. That combination—significant indirect effect alongside a significant direct effect—leads to the conclusion of partial mediation. In practical terms, part of leadership style’s impact on sustainable competitive advantage flows through knowledge initiation, while another portion remains direct (not explained by the mediator).
The session ends by positioning this GUI approach as a foundation for more complex Hayes-style testing later, including running different models from the model list within SmartPLS4’s Process model-building options. The core takeaway is that SmartPLS4 can replicate and extend classic mediation testing in a more interactive, GUI-driven way, while supporting richer model structures than the SPSS Process Macro setup often allows.
Cornell Notes
SmartPLS4’s Process option brings Hayes-style mediation testing into a graphical interface. Researchers assign an IV, mediator, and DV (e.g., leadership style → knowledge initiation → sustainable competitive advantage) and select a Hayes model such as Model 4. Significance for path coefficients may require bootstrapping; using 10,000 BCa bootstrap resamples provides P values and T statistics. The key mediation decision comes from comparing direct and indirect effects: when the indirect effect is significant and the direct effect is also significant, the result is partial mediation. This GUI workflow supports later expansion to more complex models, including scenarios with multiple IVs and DVs.
How does SmartPLS4’s Process option differ from using Hayes Process Macro in SPSS?
What steps are used to set up a mediation model (Hayes Model 4) in SmartPLS4?
Why might initial path coefficient output lack significance values, and what fixes it?
How is partial mediation determined in the example results?
What does the “specific indirect effect” represent in mediation testing?
Review Questions
- When would you choose bootstrapping in SmartPLS4 Process analysis instead of relying on path coefficients alone?
- What pattern of significance (direct effect vs. indirect effect) indicates partial mediation versus full mediation?
- How does the model structure IV → mediator → DV differ from a model that tests only direct effects?
Key Points
- 1
SmartPLS4’s Process option provides a GUI workflow for Hayes-style mediation models, reducing reliance on SPSS coding-based Process Macro setups.
- 2
The Process toolbar supports building mediation models by assigning an IV, mediator, and DV and linking IV → mediator, mediator → DV, and IV → DV.
- 3
Model 4 (mediation) can be selected from a Hayes model list within SmartPLS4’s Process interface.
- 4
Bootstrapping (e.g., 10,000 BCa resamples) is used to generate T statistics and P values for both direct paths and indirect effects.
- 5
A significant indirect effect plus a significant direct effect indicates partial mediation, meaning the mediator explains part of the relationship while a direct effect remains.
- 6
The GUI approach is positioned as a base for running additional, more complex Hayes models later using SmartPLS4’s Process model-building options.