[R-sig-ME] Problem with profile and glmer

[Emmanuel Curis]

It works with replacing

 thedata <- data.frame(herd,incidence,size,period)

with 

 thedata <<- data.frame(herd,incidence,size,period)

 so it seems that when searching for thedata, glmer searches in the
 parent environnement of testprofile() and not in the local
 environnement of testprofile().

I guess it has to do with the line

glmod <- eval(mc, parent.frame(1L))

but my knowledge stops here...

Hope that will help as a workaround until developpers find a fix for
that...

On Mon, Sep 15, 2014 at 02:21:37PM +1000, Ken Beath wrote:
« With the following code
« 
« library(lme4)
« data(cbpp)
« 
« testprofile <- function(herd,incidence,size,period) {
«   thedata <- data.frame(herd,incidence,size,period)
«   gm1 <- glmer(cbind(incidence, size - incidence) ~ period + (1 | herd),
«               data = thedata, family = binomial)
«   pr4 <- profile(gm1)
« }
« 
« testprofile(cbpp$incidence,cbpp$incidence,cbpp$size,cbpp$period)
« 
« I get the error in profile " Error: 'data' not found, but variables found
« in environment of formula: try specifying 'formula' as a formula rather
« than a string in the original model"
« 
« I have tried with as.formula for the formula. The profile and glmer work
« fine when they are called outside the function.
« -- 
« 
« *Ken Beath*
« Lecturer
« Statistics Department
« MACQUARIE UNIVERSITY NSW 2109, Australia
« 
« Phone: +61 (0)2 9850 8516
« 
« Building E4A, room 526
« http://stat.mq.edu.au/our_staff/staff_-_alphabetical/staff/beath,_ken/
« 
« CRICOS Provider No 00002J
« This message is intended for the addressee named and may...{{dropped:9}}
« 
« _______________________________________________
« R-sig-mixed-models at r-project.org mailing list
« https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models

-- 
                                Emmanuel CURIS
                                emmanuel.curis at parisdescartes.fr

Page WWW: http://emmanuel.curis.online.fr/index.html