ldecomp mediation analysis

[Ayesha Siddiqui]

Hi I am new to mediation analysis and had a few questions about the ldecomp command:

1. Can the mediator be categorical and not binary? I have a hard time conceptually interpreting the results in the discussion of my study when using a non-binary mediator.

2.. Can the control variables be categorical and not binary?

3. How did you go about deciding on/setting the fixed values of the control variables? I understand the default (if you dont specify) in stata is to set it as a median/mean.

Thanks for any guidance at all!
Best regards,
Ayesha

[Ayesha Siddiqui]

Anyone have thoughts? I am kind of stuck at this point of my analysis and am having a hard time finding reliable resources to help me. Thanks in advance!

[Jennifer Castaneda]

You may find this material useful: https://www.stata.com/meeting/italy1...t13_grotta.pdf