generalised structural equation modelling

[Adeola Kolade]

Dear all,

I will please like to find out how to include controls in mediation analysis using generalised structural equation modelling (GSEM). I use GSEM because my dependent variable is categorical and I am using an ordered logistic regression analysis. If my model is specified as this: {educational attainment_i} = β₀+ β₁parent divorce_i + β₂social class_i + β₃nbook_i+ β₄chealth_i + β₅moteduc_i+ β₆fateduc_i + β₇age_i + β₈gender_i + ε_{i .}

and my mediating variable is "depression" what command will I give Stata in order to perform mediation analysis (direct, indirect and total effect) and how will I draw this for path analysis. Please note that parents divorce is the independent variable as others are controls. I

will really appreciate your reply as it will be very helpful.

[Phil Bromiley]

If you google sem mediation Stata, you'll probably find some software that handles this problem. You will also find an example in the SEM documentation that does this.