Dear Statalisters,

I am performing a mediation analysis using path models (using sem or gsem) where I am investigating how locus of control (continuous mediator) mediates the impact of socioeconomic status (income or education) on obesity (binary outcome). Studies on similar questions model this constellation of continuous mediator-binary outcome using weighted least-squares regression (WLSMV) using the software Mplus. I read the Stata Manuals, where I found information regarding the modelling of binary outcomes in path analysis/sem using logit or probit, but no information on how to carry out a mediation analysis where part of the indirect effect is represented by the effect of the exposure (income, continuous) on mediator (locus of control, continuous). Furthermore I found no explanation on how to implement a WLSMV in Stata. My questions therefore are:

1. do you have any advice on how I should model my mediational model with continuous mediator and binary outcome in Stata?
2. do you have any advice on how I could implement a WLSMV in Stata?

Thank you very much in advance for any help!
Sarah

PS: I am new on this forum, any advice on how I should improve my post is appreciated!