Unless you have a measurement model, moderation is nothing more than testing for interaction effects in ordinary least squares regression. Hence, there is no need for SmartPLS.

Thank you, Professor Morgan, for your valuable insight.

Indeed, my research involves a measurement model with latent constructs, which is why I considered SmartPLS as a potential tool. I’m trying to assess whether it provides more accurate or interpretable results for moderation in such models compared to SPSS. I appreciate your clarification and would love to hear more from your experience if you’ve worked with moderation in SEM contexts.

My suggestion is you read and compare the type of SEM model in SmartPLS and the type in SPSS (which would be the extension AMOS). While there is an overlap, you must decide which type of SEM would be most appropriate (if, indeed, you use SEM). Search for partial least squares SEM versus covariance-based SEM. PLS-SEM certainly has grown in use and popularity, so there are many more examples than a decade ago.

Thank you, Dr. Coker, for your thoughtful and detailed suggestion.

I’ll definitely take your advice and explore the differences between PLS-SEM and covariance-based SEM to determine which method best suits my research needs.

My model involves latent constructs and aims to examine predictive relationships, so understanding the strengths of each approach is important at this stage.

I truly appreciate your guidance and the reference to more recent examples — it’s very helpful for moving forward with clarity.

For computing path analysis, Smart PLS is considered over SPSS

Thanks you, Dr. Munagala — I really

appreciate you taking the time to share your insight!

Do you intend to perform a CB-SEM analysis or PLS-SEM analysis? The current version of SmartPLS allows to perform both types of SEM analysis, while in AMOS (SPSS) you can only perform a CB-SEM analysis. What you must know is that, in the current version of SmartPLS, the two-step method is used for estimating the moderation effect. I don't know what method uses AMOS (SPSS) for the moderation effect for CB-SEM models. You should check that and compare the advantages and disadvantages for the estimation methods of the moderation effects used by the two software for CB-SEM in order to decide which one is better for the objective of your research.

Thank you, Professor Margalina, for pointing out the CB-SEM support in SmartPLS — that’s an important distinction!

I’m currently focusing on moderation within a PLS-SEM context, but your comment reminded me to also look into how AMOS handles moderation in CB-SEM.

If anyone has experience comparing the two approaches — statistically or practically — I’d really appreciate your insights!

The softwear used in the analysis depends on the measurement model used to formulate the interactions . Smart PLS is recommended for the analysis of moderated mediated relationships of formative latent constructs .

Thank you, Dr. Noel! That’s a great point — I’m currently working with reflective constructs, but your clarification about formative models and moderated mediation is very helpful. I appreciate you sharing this!

Thank you again to everyone who contributed valuable insights to this discussion.

I’d like to share a quick follow-up based on preliminary findings from my moderation analysis:

➡️ I tested a moderating effect using both SPSS (hierarchical regression) and SmartPLS (two-stage approach).

📊 Interestingly, the interaction was statistically significant in SPSS, but not significant in SmartPLS.

I’m now trying to better understand what might explain this difference. Could it be related to:

• Differences in estimation methods?
• Sample size sensitivity (n = 185)?
• Model characteristics (e.g., reflective measurement model)?

If anyone has experienced similar results or can suggest resources or explanations, I would really appreciate your thoughts!

Hi Amaal Mohammad Khalil Alshaar ! First, of all the difference can be the result of applying two different types of regression, hierarchical regression vs. pls-sem. In the case of the estimation methods for the moderator, usually in SPSS it is used the product approach. I don't know if you used SPSS or its extension for SEM analysis, AMOS. In AMOS, I don't know what estimation method for the moderator it's implemented. If it is the first case, than the moderator estimation model shouldn't be the reason because the second stage implemented in SmartPLS it's also a product approach.

The measurement model can be the cause but not because it's reflective but because of the difference in the estimation method of the measurement model itself. The PLS-SEM algorithm uses a composite approach, but in SPSS the estimation can be different (E.g. the common variance approach used by CB-SEM).

Thank you again, Professor Vasilica, for the detailed and thoughtful clarification.

Your point about the difference in estimation methods between SPSS (product term in hierarchical regression) and SmartPLS (composite-based PLS-SEM) was especially helpful.

To better understand the variation in my results, I plan to re-analyze the moderation effect using AMOS (CB-SEM). This will allow me to compare all three approaches—SPSS, SmartPLS, and AMOS—and reflect more clearly on the methodological implications.

I truly appreciate your guidance throughout this discussion.

Yes, but I recommend you to compare AMOS (CB-SEM) with SmartPLS (CB-SEM) not only SmartPLS (PLS-SEM). This will allow you to control that results are not affected by differences in the algorithms of the two software. And, of course you must check what estimation method for the moderation is implemented in AMOS.

Thank you again, Professor Vasilica, for your continued support and insightful suggestions.

Based on your recommendations, I now plan to compare four analytical approaches to better understand the moderating effect:

1. Hierarchical regression in SPSS

2. PLS-SEM in SmartPLS

3. CB-SEM in SmartPLS

4. CB-SEM in AMOS

This multi-method comparison will help clarify how estimation methods and measurement models influence the significance of the moderator. I truly appreciate your thoughtful guidance—your input has been invaluable in shaping this next step in my analysis.

I found SMART PLS better for mediating and moderating modeling. In SPSS, we have to make a lot of models and calculations. However, SMART PLS offers multiple equations in a single model.

I have found SPSS difficult to use when analyzing

model

selecting the one that suits your model. Recent papers suggest the use of Smart PLS.