Displaying 1 result from an estimated 1 matches for "pp101".
Did you mean:
ppp101
2011 Mar 08
0
nlme: Computing REML likelihood value from ML likelihood value
...I have a question concerning the computation of the value of the Restricted Maximum Likelihood (REML) function evaluated at a given set of parameter estimates from the Maximum likelihood (ML) value. Following the book of Fitzmaurice, Laird and Ware (2004) "Applied Longitudinal Analysis" pp101, the REML likelihood can be computed by multiplying the ML likleihood by the square root of the generalized variance of the fixed effects. When working on the log scale, this means:
log.REML = 0.5*(log(det(W))) + log.ML
where W is the covariance matrix of the fixed effects estimates.
However whe...