Moderation and Interaction Effects in Mplus

This video aims to provide you with a basic overview of moderation and how to estimate such in Mplus. Three different approaches to Moderation in Mplus is presented: (a) Continuous Moderation with Three Latent Factors, (b) Multi-Group Approach with Two Latent Factors and an observed factor and (c) Continuous Moderation with three observed indicators.
The resources for this series of lectures (Slides, syntaxes, data) can be downloaded here: bit.ly/31iGzfc

[@Artnotforthesakeofart]

Thank you for making this video! Now I know how to try moderation in my model =)

[@DrFurb]

I really enjoyed this thank you. Great resource. Please consider doing more videos.

[@sheriabauman5260]

Dr. Zyl's videos are so helpful. His explanations and demonstrations are clear and understandable. I now rely his videos for all my analyses in MPlus; I have numerous books on these topics, but none are as useful as these videos.

[@LlewellynVanZyl]

You're too kind Sheri! Thanks so much for the complement. I try my best 🙂 you're amazing

[@jessicamurray2611]

You sir are a life saver!!!

[@Emmanuq]

Great stuff; very useful. Thank you so much.

[@shahab5737]

Many thanks for your assistance in our project

[@rosatoro9406]

Thank you for this example, in the case that your moderator is not binary, how does the syntax line change where you set the low and hi w values?

[@sanjagani]

This is really great stuff. Can you please explain two things that will be helpful to understand interaction better for the community? (1) How are Johnson-Neyman plots different from the loop plots that you have shown here? How to interpret them? (2) Can you please show and explain any one significant moderation using plots? Thank you.

[@amrnth]

Thank you so much for your videos! I was struggling with my coursework a lot, but you became my savior :) \ Glad to see you're still making content even though your videos are useful only for a narrow circle of specialists, thus unpopular. Is there any way to support your work with money?

[@angelvani8574]

Thank you very much for your videos, you are helping me enormously to understand how to use MPLUS. If I may, I would like to ask you two questions: 1) in the case of models with latent variables and latent moderators (no observed variables), if I want to use 3 or more moderators should I test their interaction variables (i.e. interaction effects) one at a time or all together at the same time? I tried entering all three at the same time but the results (model results section) are missing the p-value column 2) If the model is instead made up of observed variables and I use 3 or more moderators (they are also observed variables) does it make sense to look at the fit indices of the model or should the fit indices not be considered and is it sufficient to look only at the standardised coefficients of both the independent variables and the interaction variables? I ask you this question about the fit indices because I have two situations: either the model is saturated (so the fit indices are not estimated) or when the model is 'not' saturated the addition of just one interaction variable (not the moderator, the interaction variable) makes them bad. I thank you in advance for your valuable advice

Can you post code for a moderated mediation sem model. Would it be a whole lot different? Ive done serial mediation in mplus but not with moderation. Also what about adding control variables.

[@yokoanna7]

Thank you for your valuable content :)! May I ask for some advice dealing with a 3 way interaction between one latent variable x one continuous observed variable x 1 dummy (i.e., social support x socioeconomic status (ISEI) x gender)? Is there such thing as a combination of the approaches? Thank you!

[@andremeyer5346]

Thanks a lot for the really helpful video. Do you (or does anyone else) have any suggestions for literature on the second approach? Thanks a lot in advance. :)

[@weimiaozhou5267]

Thanks for sharing. I have a burning question: it seems the overall model fit results did not show CFI, TLI, RMSEA, etc. I run my own data which show significant interaction for standardized results but still yielded no model fit index such as CFI, TLI. Is this standard for testing moderation among latent variables?

[@LlewellynVanZyl]

Seems to me that your model didn't convergence.

[@andremeyer5346]

Hi, have you been able to resolve this issue? I'm having a similar problem and was wondering if this was by default.

[@willysunshine]

Why don't you center the predictors in the third example?

[@tegaxel]

I want to thank you for all the videos you have created, they are useful and help me a lot in understanding how to use Mplus. I would like to ask you a question about the syntax of moderation with latent variable (10:45 sec), I have a doubt: in the syntax the "standardize" command (within the DEFINE: section) seems to me that it only standardizes the items that saturate the independent variables (i.e. the items of the variables FIT, Happy and INTER) while the items of the dependent variable TASK are not standardized. Is it not necessary to standardize the items of the dependent variable TASK as well? Thank you very much

[@LlewellynVanZyl]

Hi Michelle, no you only standardise the IV and Moderator so that the interaction term is standardized. You don't standardize your IV. It's the same in all moderation approaches where you create an interaction term.

[@Tabiyaa561]

Thank you so much for this great video. I searched for such syntax on moderation effect with latent variables but I could not find it. Could you please tell me how where you had found that syntax? And if you have a further suggestion about online Mplus resources, could you please inform me?

[@MplusforDummies]

I updated the link in the description. Hope that helps

[@dielawn87]

Do you have any tutorials on moderated mediation?