It sounds like you are encountering a situation where different statistical methods are giving you somewhat contradictory results regarding the relationship between your independent variable (IV) and dependent variable (DV), and the mediation effect. Here’s a breakdown of what might be happening and how you could approach it:

1. Simple Regression Shows Significant IV-DV Relationship

- In simple regression, you're testing the direct effect of the IV on the DV, without considering any mediating variables. A significant result here means that, without considering the mediator, the IV does appear to impact the DV significantly.

2. PLS-SEM and Process Macro Show Insignificant Direct Effect (Full Mediation)

- When you move to PLS-SEM or Process macro, you include a mediator in the analysis. These techniques are testing **both the direct and indirect effects** (via the mediator) of the IV on the DV.

- The fact that these methods show a non-significant direct effect but a significant mediation (full mediation) suggests that the impact of the IV on the DV goes entirely through the mediator.

- This is not uncommon—many times, when mediation is tested, the direct effect that was significant in a simple regression model becomes insignificant once the mediator is accounted for. This happens because the **mediator absorbs the effect** of the IV on the DV.

Why Is This Happening?

- In simple regression, you’re not accounting for the mediator, so you see a direct effect of the IV on the DV.

- In mediation models (PLS-SEM or Process Macro), you are adding a layer of complexity by including the mediator. This model essentially says: “The IV impacts the DV, but it does so by influencing the mediator, which then affects the DV.”

- In this case, your simple regression results likely reflect total effects (the combined direct and indirect effects), while the mediation models are showing that the direct effect is actually being carried by the mediator, leading to full mediation.

What Can Be Done?

- Report both findings: It’s important to report the results from all analyses. You can explain that while simple regression shows a significant IV-DV relationship, mediation analysis reveals that this effect is fully mediated by the mediator, leading to an insignificant direct effect when accounting for the mediator.

- Reconcile the results: Highlight in your discussion that the simple regression reflects the total effect, whereas the mediation model provides a more nuanced understanding, showing that the effect of the IV on the DV operates entirely through the mediator.

- Check for suppression effects: Sometimes, if the mediator and the IV are highly correlated, including the mediator might change the effect of the IV on the DV. Ensure that this isn't happening by examining multicollinearity between the IV and the mediator.

In summary, the difference between your simple regression and mediation analyses is due to the mediator’s role in carrying the effect. Your analysis is revealing a more detailed mechanism, where the IV impacts the DV indirectly through the mediator, rather than directly. This is why the direct effect becomes insignificant once the mediator is included in the model.

The explanation I provided is based on well-established statistical concepts and mediation analysis principles, which are covered in various methodological and statistical texts. Here are some key sources you can reference:

1. Baron, R. M., & Kenny, D. A. (1986). "The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations." Journal of Personality and Social Psychology, 51(6), 1173-1182.

- This is the seminal paper that introduced the framework for mediation analysis and the distinction between direct, indirect, and total effects, which is relevant for understanding why mediation affects the relationship between the IV and DV.

2. Preacher, K. J., & Hayes, A. F. (2004). "SPSS and SAS Procedures for Estimating Indirect Effects in Simple Mediation Models." Behavior Research Methods, Instruments, & Computers, 36(4), 717-731.

- This article discusses the Process macro used in mediation analysis and its role in estimating direct and indirect effects.

3. Hair, J. F., Hult, G. T. M., Ringle, C., & Sarstedt, M. (2017). A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM) (2nd ed.). Sage Publications.

- This book explains PLS-SEM in depth, including how it handles mediation and how the model allows for testing direct and indirect effects.

4. Hayes, A. F. (2013). Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach. The Guilford Press.

- Andrew Hayes’ book is one of the definitive guides on mediation and moderation analysis, including the use of PROCESS macro, and provides examples of how direct and indirect effects are handled in regression-based analyses.

These sources provide a foundation for understanding why the results of simple regression and mediation analysis can differ and why you see full mediation in your PLS-SEM and Process Macro results.

-Erin Fry

In addition to what Erin Fry explained, also check the measurement model of the variables. The same latent variable scores must be used for both PLS-SEM and simple regression.