Extract the fixed effects, random effects, variance of the
fixed effects, or variance of the random effects from a linear mixed
effects model fit with lmekin.
Usage
# S3 method for lmekin
fixef(object, …)
# S3 method for lmekin
ranef(object, …)
# S3 method for lmekin
vcov(object, …)
# S3 method for lmekin
VarCorr(x, …)
# S3 method for lmekin
logLik(object, …)
Arguments
object
an object inheriting from class lmekin
representing the result of a mixed effects model.
x
an object inheriting from class lmekin
representing the result of a mixed effects model.
…
some methods for this generic require additional
arguments. None are used in this method.
Value
the fixed effects are a vector and vcov returns their
variance/covariance matrix. The random effects are a list with one
element for each random effect. The ranef component contains the
coefficients and VarCorr the estimated variance/covariance matrix.
The logLik method returns the loglikelihood along with its
degrees of freedom.
Details
For the random effects model \(y = X\beta + Zb + \epsilon\), let \(\sigma^2\) be the variance of the error term
\(\epsilon\).
Let \(A= \sigma^2 P\) be the variance of the random effects
\(b\). There is a computational advantage to solving the problem
in terms of \(P\) instead of \(A\), and that is what is
stored in the returned lmekin object.
The VarCorr function returns elements of \(P\); the print
and summary functions report values of \(A\).
Pinhiero and Bates call \(P\) the precision factor.
References
J Pinheiro and D Bates, Mixed-effects models in S and
S-Plus. Springer, 2000.