Moderated mediation when DV and MV are ordinal

[Dominik Bong]

Dear statalist community,

I am trying to conduct a moderated mediation analysis and find myself stuck on the following problem.

My DV is ordinal (integers ranging from -7 to +7, including 0),and my mediating variable is also ordinal (again, only integers from 0 to +8). Before considering the mediating variable, I have employed ordered logistic regression (ologit) to test my hypotheses and this worked fine. Now, however, I need to add the mediating variable and struggle to execute an adequate mediation analysis (if I have understood previous posts correctly, the normal sgmediation command is not appropriate for ordinal DVs). Could anyone please provide some help on how to best proceed?

Furthermore, in a final elaboration, I need to add to this model a continuous moderating variable (first stage moderation, i.e., influencing the effect of my IV on the mediation variable). Therefore, ideally I would need an approach for the mediation analysis that can be extended to include a moderator as well.

Your help is highly appreciated!

[Phil Bromiley]

Welcome to Stata list. You will increase your chances of useful answer by following the FAQ on asking questions-provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

This might be feasible in user written cmp but it is certainly doable in GSEM. You should be able to put in an interaction using factor variable notation in almost any Stata provided estimator. Factor variable notation is a good thing to use because it saves you making the interactions before the estimation, and makes it much easier to use margins to look at predicted values or derivatives after the estimation.