[Mediation-information] question on mediation sensitivity analysis

[Kosuke Imai]

Dear Kristina,

  Thanks for your email.  Unfortunately, right now, the mediation software does not incorporate the ordinal mediator for the sensitivity analysis.   This is on the to-do list but we have to figure out the exact details first.  For now, you might want to run the least squares for the first stage as an approximate sensitivity analysis.

Best,
Kosuke

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


On Nov 15, 2011, at 5:58 AM, Kristina Blennow wrote:

> Dear Kosuke,
> Thank you for your good work! I would like to use your mediation package in R. One problem that I have is that the mediator is ordinal and the outcome is binary. Is there a way that I could conduct sensitivity analysis on estimated mediation effects?
> Thank you in advance!
> Best wishes,
> Kristina Blennow
>  
> <image001.gif>
>  
> Kristina Blennow
> Professor of Landscape Planning
> Faculty of Landscape Planning, Horticulture and Agricultural Sciences
> SLU - Swedish University of Agricultural Sciences
> (Visitors: Sundsvägen 4-6)
> P.O. Box 58, SE-230 53 Alnarp, Sweden
> Tel. +46 40 415230
>