[R-sig-ME] Binomial problem for lme4_0.999375-1: 0,1 vs xbar n

[Parsons, Van L. (CDC/CCHIS/NCHS)]

 The suggestion of Prof. Bates to try the (success, failure)
representation
in lmer for the condensed Bernoulli data gave results consistent with
the full data. 

  For the example discussed, this code worked: 

  lmer( n1*cbind( p1,(1- p1))~ 1 + (1| Subject)  ,
         data=sleep1c,  family=binomial )

 Van