[Mediation-information] interaction terms in mediation package

[Kosuke Imai]

Hi Sun,

  It was great to meet with you at the SREE.  The direct effect under the control condition means that the mediator is held constant at the value that would be realized under the control condition, M(0).  Alternatively, you can define the direct effect while fixing the mediator to the value that would be realized under the treatment condition, M(1).  You may want to consult our psychological methods or American Political Science Review paper available at http://imai.princeton.edu/projects/mechanisms.html  Also, I'm ccing this message to our mediation email list so that others can benefit from your question.  It would be great if you could direct your future questions to that list.  See https://r-forge.r-project.org/projects/mediation/

Thanks,
Kosuke

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

On Apr 26, 2012, at 2:38 PM, Sun Yoon wrote:

>     Hi Imai,
> 
> I hope you are well. I am Sun Yoon from UW-Madison. I am not
> sure if you remember me or not though--I asked a question about multi level
> data structure in Mediation package in the SREE conference in March. I am
> trying to figure out what the direct effect under the control group means. I
> have another question, if you don't mind, about interaction terms in the direct
> effect. I understand that there should be different mediating effects under
> control and treatment, but I don't understand how the control
> (non-intervention) has an impact on the dependent variable. Shouldn’t be
> identical the direct effect of the intervention under the control and
> treatment? Thank you so much. And look forward to hearing from you soon.
> 
> 
> Best regards,
> Sun
> 
> --
> Sun Young Yoon 
> Ph.D. Student 
> Department of Educational Leadership and Policy Analysis 
> University of Wisconsin-Madison