[Mediation-information] Question about the R package of casual mediation analysis

[Kosuke Imai]

Please see below:

Kosuke Imai
Department of Politics
Princeton University
http://imai.princeton.edu

On Fri, Sep 25, 2015 at 3:41 PM, Xian Cao <Xian.Cao at business.uconn.edu>
wrote:

> Dear Dr. Imai,
>
> I am a third year PhD student at UCONN Management Department. I read your
> study "casual mediation analysis" - it is very interesting and useful. My
> advisor and I are working on a paper using your R package. I have a few
> questions that wonder whether I may have your answers.
>
> 1. Can I use this method when the X->M path is nonlinear. That means X is
> related to M curvelinearily.
>

Yes, that's exactly why this method is developed.   I encourage you to read
the following paper, which describes the methods implemented in this
package: http://imai.princeton.edu/research/files/BaronKenny.pdf


>
> 2. For the hypotheses testing part (med.fit and out.fit in the R package),
> can I use a bayesian method (mcmc) instead? Will this influence the
> subsequent casual mediation effect analysis?
>

Currently, we are not implementing Bayesian MCMC.  Our simulation method is
a (quasi-bayesian) monte carlo method, which is described in the
aforementioned paper.


>
> 3. Can I use this package to test the mediated moderation? That is,
> testing whether there is a mediator mediates the relationship between X and
> a moderator.
>

Yes, please have a look at this JSS paper:
http://imai.princeton.edu/research/files/mediationR2.pdf


>
> Thank you so much for your time,
>
> Xian Cao
>
> Management Department,
>
> University of Connecticut,
>
> Tel: 989-528-9566
>
> Email: *Xian.Cao at business.uconn.edu* <Xian.Cao at business.uconn.edu>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/mediation-information/attachments/20150927/8deedcc0/attachment.html>