[Mediation-information] mediate function with cluster

[Kosuke Imai]

Dear Philip,

I'm ccing my collaborators who know more about the details of the package so that they can respond to this. 

Kosuke

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

On Nov 16, 2012, at 6:09 AM, Philip Parker <Philip.Parker at uws.edu.au> wrote:

> Dear Professor Imai,
> 
> I am currently working with your mediation package in R which is fantastic. However, when I utalise the cluster augment I get the following error:
> 
> Error in tapply(x, cluster, sum) : arguments must have same length
> 
> I have tried referencing the cluster variable directly and as a charater (i.e. how treat and mediator are called in the function) as follows
> 
> > effects<- mediate(Model2_SingleLevel, Model3_SingleLevel,
> +                   sims=100, treat="Zses", mediator="Ach1", 
> +                   cluster=AchData$SCHOOLID)
> 
> and
> 
> > effects<- mediate(Model2_SingleLevel, Model3_SingleLevel,
> +                   sims=100, treat="Zses", mediator="Ach1", 
> +                   cluster="SCHOOLID")
> 
> I assume it is something very simple that I am foolishly missing but would be greatful for any advice. 
> 
> In passing I also noted that the sandwich package either did not load correctly or was not installed as a dependency with your mediation package. 
> 
> Kind Regards,
> Phil
> 
> Dr. Philip Parker
> Centre for Positive Psychology and Education
> School of Education |University of Western Sydney
> Locked Bag 1797, Penrith, NSW, 2751
> Tel: (612) 9772 6108| Fax: (612) 9772 6432
> http://www.uws.edu.au/centre_for_educational_research/cppe