www.phidot.org

It's possible that I haven't done all my homework on design matrices and the sin link function, but can anyone explain what this error is about ? I'm trying to fit a simple robust design model (with no movement) - constraining S, p and c with the sin link function. S varies by time and p varies by time and session. Additionally, p=c (so p and c share columns in the design matrix). I get this error:

Code: Select all

S.time.sin=list(formula=~-1 + time, link="sin")

p.time.session.sin=list(formula=~-1 + time:session, link="sin", share=TRUE)

N.session=list(formula=~session)

GammaDoublePrime.nomove=list(formula=~1,fixed=0)               
GammaPrime.nomove=list(formula=~1,fixed=0)

> mark(col.proc, col.ddl, model = "Robust",
+      model.parameters=list(S=S.time.sin,
+                            p=p.time.session.sin,
+                            N=N.session,
+                            GammaDoublePrime=GammaDoublePrime.nomove,GammaPrime=GammaPrime.nomove))
Error in sort.int(x, na.last = na.last, decreasing = decreasing, ...) :
  'x' must be atomic


If I use the logit link for p and the sin link for S things work out, but not vice versa, so there's something here about the parameters p and c and the sin link function ???

Any help is appreciated. Thanks.

Tommy

This coding error has been fixed in the 2.0.8 beta at https://github.com/downloads/jlaake/RMark/RMark_2.0.8.zip
See the Whatsnew help file for all of the changes/additions in 2.0.8. I'll release on CRAN when I return from the field in 2 weeks. --jeff