[R] RSA and lmer

[Ben Bolker]

Pablo Ignacio Escribano Pinto <p.escribano <at> gmail.com> writes:

> 
> Hi all!
> I am currently working on a paper where I conduct a 
> surface analysis based on subordinate and supervisor responses.
> So far I have been using the RSA package and it works great.
> However, because several subordinates share a common supervisor,
> the RSA object that I need to estimate
> needs to include random a random intercept and slopes (2)–which
> implies a mixed model.
> Normally I use lmer for estimating mixed models, but I have 
> no clue how to integrate this to RSA in order to
> conduct a surface analysis.
> 
  Questions that have to do with mixed models will generally get
better answers on the r-sig-mixed-models at r-project.org mailing list.
I don't know the field of advanced response surface analysis very
well.  It sounds like you might need to contact the RSA package 
maintainer, and that you are likely to have to do a lot of re-engineering
from scratch.  Maybe as a shortcut you could figure out a way to
get robust (Huber-White/sandwich) estimates of the standard errors?

  Ben Bolker