Moderated Mediation with PROCESS Model 8 (SPSS)

Do you want to run a moderated mediation with a moderated a-path and a moderated direct c'-path using Hayes's SPSS macro (PROCESS Version 3)?

Check out my STATISTICS CONSULTING services here:

This tutorial will show you:
- How to run the analysis from the dialog box
- How to interpret the output (e.g. the index of moderated mediation and the follow up tests)

This tutorial is based on model 8 in PROCESS version 3.4.

How to report a moderated mediation:

The following video-tutorial will show you how to test regression assumptions for your model:

And this text-based tutorial will show you how to use the main PROCESS syntax commands:

0:00 Start
0:18 Which parameters?
2:00 Output interpretation
2:19 Index of moderated mediation
3:52 Conditional indirect effects
4:30 Conditional direct effects
5:20 a-path
6:35 b-path and c'-path

[RegorzStatistik]

How to REPORT a MODERATED MEDIATION (Tutorial):

[stasrieznik]

Thank you very much for a good explanation.
It was really helpful.

[Gabriel.Glacial]

This was incredibly helpful, thank you so much!

[emmavanarsch]

Thanks for your video and explanation! Why is it better to use model 8 instead of testing interaction effects directly? What is the advantage of model 8 compared to the test of Region of Significance and Proportion of Interaction and the test of crossover point?

[khaloud]

Thank You, I have question my IV is a group and I code them as group 1 =0 and group2=1 however when I got the result for model 8 it shows like this Group- medation- dependent variable . how I can know which group has more effect ?

[gabriellamacraejones]

this was very helpful, thank you! I have some questions if you may be able to help: My mediator significantly predicts my outcome, but no other pathways were significant(p > 0.05). 1. I am using a categorical moderator (2 levels, 0 = male 1 = female) The moderated mediation index is not significant but the conditional indirect effects are as follows :
0.0000 -0.0257 0.0466 -0.1128 0.0751
1.0000 0.0924 0.0478 0.0088 0.1980
so the conditional direct effect for females looks significant.. why then might the pathway not be moderated?
2. The bootstrap results for my outcome variable also say that my moderator and mediator both significantly predict the outcome, though this is not the case for the first output for my outcome variable (bio sex marginally above sig level, p= 0.06). Which result do I report?
if you have the time to help, I would really appreciate it (I'm new to statistics!)
thanks again :)

[marcobuttner]

Thank you for your video! However, I still do not fully understand whether I can use model8 if I also want to investigate the moderated effect of X on M. For example, I do have two hypothesis, a) concerns the moderated mediation effect and b) the moderated effect of X on M only. Hence, the question is as follows: Can I use Model 8 to test both hypothesis or do I have to conduct a second analysis with (e.g., Model 1) to test for the moderated effect of X on M?

Thank you in advance.

[cailawrance]

Thank you so much for your sharing!
Meanwhile, do you know how to do the sensitivity analysis and find the effect size for moderated mediation?
Thank you so much!

[sabrina]

Hi, danke für das Video, sehr anschaulich und einfach erklärt!
Mein Problem ist nur, das jetzt alles in einer schönen und anschaulichen Tabelle richtig darzustellen. Und vor allem das alles in Worten wissenschaftlich richtig wiederzugeben. Gibt es denn ein Video oder eine Studie, in der ich sehen kann, wie ich das alles richtig mache?

[FarsadDaPlayaaa]

Hi, eine Frage hätte ich, wenn ich so vorgehe, wie Sie im Video also auch das Model 8 verwende und "Bootstrap inference for model coefficents" auswähle und auch die anderen Optionen so wie Sie im Video wähle, muss ich dann dennoch die Annahmen für Regression per Hand checken, bevor ich die moderierte Mediation ausführen darf?

[ivory]

Thanks! Is it still meaningful to explain the main effects (IV/moderator to DV in your video) when the interaction effect is significant in Model 8 and 10?

[Mila-kkyq]

Thank you for your tutorial! Could you please help me with one question? I have a multi-categorical IV with 4 categories, a moderator with 3 (low, average, high) categories and continuous DV. Can I still use Process macro model 1 if I indicate that X and W are multicategorical? Thank you in advance for your help.

[smndysk]

Danke für das Video! Wie muss ich vorgehen, wenn der direkte Effekt nicht signifikant ist, aber die Moderation davon schon?

[ninico]

Thanks a lot for this explanation! I have a short question: I plan to include three moderators instead of only one. How should I best proceed here? Should I conduct three analyses, each time with a different moderator and the other two moderating variables as controls?

[xiaohanhu]

Thank you so much for this video! I have a question on understanding and interpreting results from model 8. I have significant interaction effect (between X and W) on both the a and c path; also significant main effect of X on a and c path (X predicting M, X predicting Y); M is also significantly predicting Y (significant b path). But the indirect effects on different levels of W are not significant, nor is the index of moderated mediation. So it is apparent that there is no mediation or moderated mediation here. But can I still report the significant relationships in path a, b, and c? What might be some explanations to the non-significant mediation while both a and b paths are significant? Would like to know if you have any suggestions. Thank you so much!!

[jacksonnjoroge]

Hi thanks for the video. For my analysis, I find a non-significant index of moderated mediation as well as a non-significant interaction term on the indirect effect path (path a). However, the interaction term on the direct effect (path c prime) is significant and since I am using dummy coding for the moderator, the direct effect of one level is significant while the other is not. How would you interpret that? Thanks!

[서울대지박령]

Thanks for your kind explanation! I was wondering whether the mediated moderation in repeated measures(within-subject design) from model 8 can also be tested. Thanks.

[amomentwithzoe]

what happens if the direct effect mediation didn't work?