How to perform multilevel mediation analysis?

[Yoshito Kawakatsu]

Hi,
I tied to perform multilevel mediation analysis by using ml_mediation (http://www.ats.ucla.edu/stat/stata/faq/ml_mediation.htm).
However, I have a question how to perform the analysis using the following data.

My data is as follows;
DV; Continuous variable
IV; Categorical variable (3 scale)
DV; Categorical variable (3 scale)
cid; cluster ID

Is the following command correct ??
ml_mediation, dv(DV) iv(IV) mv(DV) l2id(cid)

Could you let me know what is the correct command?

Sincerely,

Yoshi

[E. David Aja]

I don't have a clear sense for what it is you're asking about--you may want to post some example data and the code you used to set it up. However, if your variables have the names shown above, it's unlikely that you're getting the result you want because your dependent variable and mediator variable have the same name.

[Yoshito Kawakatsu]

Thank you for your comment and sorry for my mistakes.
I wrote the wrong name of the variables.

I explain the detail of my study and data as follows;

In my study, employee's working performance is the outcome, which is continuous variable and named "performance"
I would like to know the direct and indirect effects of superviser's knowledge level, named "knowledge", which categorized into 3 scale, low, middle and high.
And the mediator variable is type of training, named "training". There were three type of training.

In sum,
DV is employee's working performance, "performance", which is continuous.
IV is superviser's knowledge, "knowledge", which is categorical variable ( 3 groups)
MV is type of training,"training", which is also categorical variable (3 groups)
cid is cluster ID

Now, I have two question.
1. Could you let me know what code should I use?

2. Is it possible to control other variables (ex. age, sex, and other factors influencing their performance) ?

Is it clear explanation?
If not, please let me know. I will explain more.

Thank you for your supports,
Sincerely,

Yoshi