re: st: binary mediation

["Ariel Linden, DrPH" <ariel.linden@gmail.com>]

First off, individuals posting to the Statalist are requested to provide
information about the program they are using and where they can be found for
download (http://www.stata.com/support/faqs/resources/statalist-faq/). In
this case, you are referring to -binary_mediation-, a user-written program
(findit binary_mediation).

In regards to your question: no, the program is not intended to be used with
a multiple-categorical independent variable, and you can very easily have a
misspecified model on your hands if you use it as such.

I suggest you consider -khb-, a user-written program (net sj 13-1 st0236_2)
and an accompanying Stata Journal article
(http://www.stata-journal.com/article.html?article=st0236)

-khb- can be used with a multitude of variable types for Treatment, Mediator
and Outcome variables.

Ariel 

Date: Tue, 30 Apr 2013 13:45:07 +0300
From: Sharon Sznitman <sznitmans@gmail.com>
Subject: st: binary mediation

I have a question regarding binary_mediation:
My independent variable (IV) ethnicity has 3 categories (1=a,
2=b, 3= c). My mediation variable (MV) is a binary variable
(0=no, 1=yes) and my dependent variable is binary (0=no binge, 1=binge).

I think that stata's program binary_mediation will allow me to do a
mediation analysis with a binary DV and lets me calculate the proportion
mediated, and whether the mediation is significant. However, can the
command binary_mediation deal with the fact that my  IV has
3 categories?

 If binary_mediation does not allow for a three category  IV, do you have
suggestions for another way of doing the mediation analysis with the
types of variables I have?

Thanks,
Sharon


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/