Hello dear specialists of STATA,
I'm going to test a mediate variable equalityrec in my model, and my model is a multi-level linear GSEM model, in which the dependent variable is ifredu and independent variable is eiscedrec, control variable is gndr. All variables are categorical variables beyond the dependent variable. All four variables are on the micro level, while the macro-level variable is "country". I have ran a syntax as:
gsem (i.eiscedrec i.gndr M1[cntry]-> equityrec) (i.equityrec i.gndr M1[cntry]-> ifredu) (i.eiscedrec i.gndr i.equityrec M1[cntry] ->ifredu )[pw=pweight*pspwght], nocapslatent latent(M1) vce(robust)
The attachment is the results present by STATA after running this line of code:
But I'm not sure how to interpret the results, especially how to confirm whether my MV plays a role. I have used the code "medsem" when using SEM model, which can directly present the direct/indirect effect of my models. Are there any code can do as medsem? Or how should I interpret these coefficient as evidence to say whether the mediation exist or not? Or even...if my codes are theoretically correct ? :-(
Thank you a lot in advance! Look forward to some kind answers!
*That's the first time I posting on this forum. If I have done anything wrong, please tell me and I will edit or delate.
0 Response to How to present (in-)direct effects when using multilevel linear GSEM
Post a Comment