Skip to content
Prev 4789 / 20628 Next

glmer with/without intercept gave different results

Dear Ruby,

I do not think a REML solution exists for the Cell variance in this  
instance (it's infinity). I presume the data for each Cell all have  
the same Villus and Group?  If so you would be better off reducing  
your data to 250 binary data (i.e. One datum for each Cell) and  
removing the (1|Cell) term from the model.

Cheers,

Jarrod
On 18 Nov 2010, at 16:18, Chang, Yu-Mei wrote: