[Mediation-information] A Question on Causal Mediation Analysis

dustin tingley dtingley at gov.harvard.edu
Mon Oct 3 20:05:47 CEST 2011


Min-Young-

Thanks for your interest in using our package.

We have not yet implemented duration models into our package. In principle
it should be able to handle these but we have not yet done this. If you know
some R you can modify our code. Otherwise this will have to wait until a
future release.

One quick note, in our framework you need only two models (using your
notation, Y=M+X and M=X).

best,
Dustin

Dustin Tingley
Government Department
Harvard University
http://scholar.harvard.edu/dtingley



On Mon, Oct 3, 2011 at 1:48 PM, Min-Young Kim <mkim.uiuc at gmail.com> wrote:

> Dear Professor Dustin Tingley,****
>
> ** **
>
> I am writing this email to ask you a question on the “Causal Mediation
> Analysis Using R”.****
>
> ** **
>
> I am a PhD candidate at the business school of the University of Illinois
> at Urbana-Champaign. ****
>
> ** **
>
> I am currently working on a project in which I employ a mediation analysis.
> But I am facing some difficulties due to the nature of my mediating
> variable.****
>
> ** **
>
> The following is a brief outline of the mediation model.****
>
> ** **
>
> **1.      **X: a count variable (the number of countries)****
>
> **2.      **M: time (time taken to first forward citation in patent
> citation network)****
>
> **3.      **Y: a count variable (the number of citations a patent
> receives)****
>
> ** **
>
> Due to the various data types, I am estimating with:****
>
> **1.      **A negative binomial model for X->Y****
>
> **2.      **An accelerated-failure time (AFT) model (one of the survival
> models) for X->M****
>
> **3.      **A negative binomial model for M->Y****
>
> ** **
>
> As many of the current statistical models for mediating analyses are
> developed for continues mediating and/or dependent variables, I am
> experiencing difficult time in testing my hypotheses.****
>
> ** **
>
> In the meantime, I have found that “A Question on Causal Mediation Analysis
> Using R” could help me dealing with these issues.****
>
> ** **
>
> But when reading through the instruction manual, especially Table 1 in page
> 9 (of the version September 19, 2011), I found that that accelerated-failure
> time model is not included in the mediator model type.****
>
> ** **
>
> So, my question is:****
>
> **1.      **Whether it is possible to test my mediation hypotheses with
> your “mediate” program for R?****
>
> **2.      **if not, do you have any plan to include the AFT model in the
> “mediate” program?****
>
> ** **
>
> Any comments would be really appreciated.****
>
> ** **
>
> Thank you so much.****
>
> ** **
>
> Sincerely yours,****
>
> ** **
>
> Min-Young Kim****
>
> ** **
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/mediation-information/attachments/20111003/d99f70fe/attachment.htm>


More information about the Mediation-information mailing list