mediate command, POmeans and predictive margins

[ronny manos]

Hello
I run a mediate command in Stata with the option: all aequations.
I also run the logit command with an interaction, after which I run the margin comand to get the predictive margins.
The predictive margins are similar but not identical to the POmeans.
Which should I be using?
It is OK to run the mediate but report the predictive margins?
Attached are the commands and the results.command with results.docx

[Rich Goldstein]

your description is not very clear but I believe that these two are not the same and, thus, differences would not be surprising; however, you can make this easier on everyone by following the advice in the FAQ (which you should read) and presenting exactly what command you gave and exactly what results Stata provided - all in CODE blocks so what you paste is easy to read