Hi! Thanks for the explanation video. Can the mediate command calculate partial mediation? Further, can we also include multilevel effects in here, as is possible in the gsem command?
Thank you! 1) To test for partial mediation in Stata's mediate command: > Check direct effect: If significant after accounting for the mediator, it suggests partial mediation. > Compare direct & indirect effects: Significant direct effect and/or substantial difference in their magnitudes indicate partial mediation. 2) The standard mediate command doesn't handle multilevel data: > Instead you could try the user-written ml_mediation command (type "help ml_mediation") > For further information see: stats.oarc.ucla.edu/stata/faq/how-can-i-perform-mediation-with-multilevel-data-method-1/ I hope this helps!
Currently, the "mediate" command is limited to cross-sectional data. Perhaps the following discussion will provide some further background to your panel data question: www.statalist.org/forums/forum/general-stata-discussion/general/1659266-mediation-analysis-panel-data
Thank you, sir, for your useful video and your great instruction. If we have two mediators with 3 and 4 categories. Please kindly give me some advice on how we can proceed with the mediate command?
You can use factor variables with the mediate command (type "help mediate" in Stata's command line). The simplest solution is to use "i." in front of categorical variables (e.g., i.var_name) so that Stata knows that the variable is not continuous.
Hi! Thanks for the explanation video. Can the mediate command calculate partial mediation? Further, can we also include multilevel effects in here, as is possible in the gsem command?
Thank you!
1) To test for partial mediation in Stata's mediate command:
> Check direct effect: If significant after accounting for the mediator, it suggests partial mediation.
> Compare direct & indirect effects: Significant direct effect and/or substantial difference in their magnitudes indicate partial mediation.
2) The standard mediate command doesn't handle multilevel data:
> Instead you could try the user-written ml_mediation command (type "help ml_mediation")
> For further information see: stats.oarc.ucla.edu/stata/faq/how-can-i-perform-mediation-with-multilevel-data-method-1/
I hope this helps!
@@statisticsmadeeasy5458 Thanks, super helpful!
Is this mediate command designed for cross sectional data only? How about panel data?
Currently, the "mediate" command is limited to cross-sectional data. Perhaps the following discussion will provide some further background to your panel data question: www.statalist.org/forums/forum/general-stata-discussion/general/1659266-mediation-analysis-panel-data
Thank you for making this useful video!
Thank you, sir, for your useful video and your great instruction. If we have two mediators with 3 and 4 categories. Please kindly give me some advice on how we can proceed with the mediate command?
You can use factor variables with the mediate command (type "help mediate" in Stata's command line). The simplest solution is to use "i." in front of categorical variables (e.g., i.var_name) so that Stata knows that the variable is not continuous.
Thank you sir for the explanation video!
Could you tell me the references explanation for my citation on my thesis, sir? Thank you so much for help!
You are welcome. Please have a look at the references in the video description. You can also refer to the official Stata help file.