| Message/Author |
|
|
| Anh-Thuy Le posted on Saturday, June 08, 2019 - 4:36 pm
|
|
|
Hi all,
I am very new to these analyses and was wondering, if I am trying to do multigroup modeling rather than moderated mediation, should I still include the interaction term in the multigroup commands? Or would it simply be the syntax for a mediation model using the "Grouping" command? Thank you! |
|
|
|
|
| The interaction is taken care of in the multigroup analysis in that the groups differ in their regression coefficients. See also our RMA book on multiple-group analysis of regression and mediation models. |
|
|
|
|
Hi!
I am conducting a moderated mediation analysis (categorical moderator) and wonder whether the multi-group analysis (examination of mediation in four different groups = categories of the moderator) has less power than the interaction approach. Is there a difference in power with these approaches? How could the interpretation look like if a moderated mediation is significant, but the mediation in none of the four groups is significant? Do you have any references for this?
Thank you!
Best wishes
Sabrina |
|
|
|
|
| Edit: In the moderated mediation analysis, the moderator is dichotomous. In the multi-group analysis we further splitted the grouping variable into four categories. |
|
|
|
|
| I think the 2 approaches should have the same power. I have no references; perhaps SEMNET does. If mediation is not significant in any of the groups, so despite taking moderation into account, you have to just say that there is no mediation. |
|
|
| Back to top |
