Hi Murthy-<div><br></div><div>I recommend using the medsens function on output from mediate, not mediations. This will make things simpler for you.</div><div><br></div><div>best,</div><div>Dustin</div><div><br clear="all">
Dustin Tingley<br>Government Department<br>Harvard University<br><a href="http://scholar.harvard.edu/dtingley" target="_blank">http://scholar.harvard.edu/dtingley</a><div>Director, Program on Experience Based Learning in the Social Sciences<br>
<div><a href="http://projects.iq.harvard.edu/peblss/" target="_blank">http://projects.iq.harvard.edu/peblss/</a>
</div><div>Contribute to the Research Reservoir: <a href="http://projects.iq.harvard.edu/undergradscholars/pages/research-resources" target="_blank">http://projects.iq.harvard.edu/undergradscholars/pages/research-resources</a></div>
</div><br>
<br><br><div class="gmail_quote">On Tue, Aug 28, 2012 at 12:14 AM, Kosuke Imai <span dir="ltr"><<a href="mailto:kimai@princeton.edu" target="_blank">kimai@princeton.edu</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Thanks for your message. I'm ccing my coauthors who know the details of the software and might be able to give you some suggestions.<br>
<br>
Kosuke<br>
<br>
Department of Politics<br>
Princeton University<br>
<a href="http://imai.princeton.edu" target="_blank">http://imai.princeton.edu</a><br>
<br>
On Aug 27, 2012, at 8:26 PM, Murthy Mittinty wrote:<br>
<br>
> Dear Prof. K Imai,<br>
><br>
> Myself am Murthy. N. Mittinty, working as post doctoral fellow at the School of Population Health, in University of Adelaide. In my research I had found that use of Causal mediation analysis answers the research question that I am interested in. It is in this regard I tried using the Causal mediation analysis software developed by you and your colleagues in R. I have understood the output and the maths behind the programs through the references. I thank you very much for developing this software. However, I have come across one output which I could not get the program to spit out. This is in regard to multiple mediators. I used the mediations and did my analysis for multiple mediators. Using the plot I get the simple graphs (the ACME, direct effect and total effect) but could not perform the sensitivity analysis. Things that I tried and observed,<br>
><br>
> X<-mediations(....)<br>
> names(X)<br>
> dliter.SR4.CSB (first mediator)<br>
> X1<-X$dliter.SR4.CSB<br>
> Then to do the sensitivity analysis<br>
> X1sens<-medsens(X1,<a href="http://rho.by" target="_blank">rho.by</a>=0.05)<br>
> Get an Error statement<br>
> I looked at the names of X1 and this is exactly same as the names if I did with one mediator.<br>
><br>
> However. The problem is in the extract of model.y and model.m.<br>
> X1$model.y<br>
> NULL<br>
> X1$model.m<br>
> NULL<br>
><br>
> These outputs are not available because the Mediations uses a glm method and it does not give me the output which is similar to the one that was produced using the lm procedure. If you can please help me get the sensitivity analysis plots with multiple mediators, this is highly appreciated and acknowledged appropriately in my research. I even looked at plot.mediations but this gives the same plot as plot(X) two graphs of the ACME, direct effect and total effect. This is not what I am after but actually the sensitivity plots.<br>
><br>
> Thanks for all your help and time.<br>
><br>
> Kind regards.<br>
> I hope I made my point clear for you to provide your valuable guidance.<br>
> Murthy.N Mittinty.<br>
><br>
<br>
</blockquote></div><br></div>