st: Alternative to sobel/Goodman test ?

["Lars E. Kroll" <mail@lkroll.de>]

Dear all,

I'm searching for an alternative way of testing the indirect effect in my stcox modell.

My Hypothesis:

      MV 
     /  \
   a/    \b
IV /_c/c'_\ Mortality

I want to test the indirect effect of my IV (Education) via my MV(Income) on mortality. This is usually done by the sobel Test (cf. Baron & Kenny JPSP 1986 pp.1173 & sgmediation.ado ).

This way is not applicable in my case, because the relationship between IV & MV can be described using OLS whereas MV-Mort and IV-Mort are specified using a survival model like stcox.

I'm searching for a formula for the standard deviation (sd_indirect) of my indirect effect (b_indirect= c-c') to conduct a t-test.

I would appreciate any hints!

Yours,

Lars
__________________________________
Lars Eric Kroll
Robert Koch-Institut
- FG 24 Abteilung 2 -
D-13302 Berlin
Phone: +49 (0)30 18754-3909
Fax: +49 (0)30 18754-3513
Web: www.lkroll.de
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/