Skip to content
Prev 15149 / 20628 Next

force lmer/glmer to use known random effects

Fitting a random term with known covariance matrix G is equivalent to 
fitting random term Zu where ZZ' = G and cov(u) = I. Eg Z' is the 
Choleski root of G. So is theoretically possible with lmer but may not 
be practicable (depends on size of G).
On 01/07/2017 07:29 PM, Alexia Jolicoeur-Martineau wrote: