Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: sobel goodman test


From   John Antonakis <John.Antonakis@unil.ch>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: sobel goodman test
Date   Sat, 21 Mar 2009 10:54:10 +0100

Sure, you can do this with xtivreg (RE or FE) or xtiverg2 (RE)....if you have more than two equations could you use reg3 (with FE only). There is the overid or xtoverid that you can download to test for overidentifying constraints. Also, if your model has fixed effects only, reg3 will do perfectly fine and then you can test the indirect path using the delta method (i.e., nlcom)--I don' know how this would be done in the context of xtivreg though (though I think that programs like Mplus can handle that--i.e., the indirect path using delta or bootstrap).

HTH,
John.

____________________________________________________

Prof. John Antonakis
Associate Dean Faculty of Business and Economics
University of Lausanne
Internef #618
CH-1015 Lausanne-Dorigny
Switzerland

Tel ++41 (0)21 692-3438
Fax ++41 (0)21 692-3305

e-mail: john.antonakis@unil.ch

____________________________________________________



On 20.03.2009 23:25, Philip Ender wrote:
Narasimhan Sowmyanarayanan wrote...

    Hello stata listers:

    I recently came across a user written command for doing the sobel
    goodman mediation test in the link below. The command is sgmediation

    http://www.ats.ucla.edu/stat/stata/ado/analysis/sgmediation.hlp.htm

    I believe that this relies on the Baron and Kenny procedure. My
    question was more on the lines that if one is working on panel data,
    how to accommodate a test for mediation. I assume that the sgmediation
    does not do the job anymore. Is there an appropriate way to test
    mediation using the "xtreg" or "xtmixed" framework. I apologize if the
    question sounds naive. I tried to look at several different references
    to the best of my ability before I decided to put this on the list.
    Hopefully I can gain some insights on this from some of you. Once
    again, thanks in advance for anyone who may respond.

    Regards

    Narasimhan

As the author of the -sgmediation- program, I can verify that it will
not work with panel data,
in fact, it will not work with any command other than -regress-.  I
don't know of any Stata
programs for mediation with panel data.  I suggest that you take a
look at David McKinnon's
book "Introduction to Statistical Mediation Analysis."  It has a
chapter on multilevel mediation
models.

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



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index