Approximate confidence intervals for the parameters in the linear
model represented by
object are obtained, using
a normal approximation to the distribution of the (restricted)
maximum likelihood estimators (the estimators are assumed to have a
normal distribution centered at the true parameter values and with
covariance matrix equal to the negative inverse Hessian matrix of the
(restricted) log-likelihood evaluated at the estimated parameters).
Confidence intervals are obtained in an unconstrained scale first,
using the normal approximation, and, if necessary, transformed to the
constrained scale, unless the control parameter
natUnconstrained is
set to
FALSE (see the documentation on
glsControl).
USAGE:
intervals(object, level, which)
REQUIRED ARGUMENTS:
object
an object inheriting from class
gls, representing
a generalized least squares fitted linear model.
OPTIONAL ARGUMENTS:
level
an optional numeric value with the confidence level for
the intervals. Defaults to 0.95.
which
an optional character string specifying the subset
of parameters for which to construct the confidence
intervals. Possible values are
"all" for all parameters,
"var-cov" for the variance-covariance parameters only, and
"coef" for the linear model coefficients only. Defaults to
"all".
VALUE:
a list with components given by data frames with rows corresponding to
parameters and columns
lower,
est., and
upper
representing respectively lower confidence limits, the estimated
values, and upper confidence limits for the parameters. Possible
components are:
coef
linear model coefficients, only present when
which
is not equal to
"var-cov".
corStruct
correlation parameters, only present when
which
is not equal to
"coef" and a
correlation structure is used in
object.
varFunc
variance function parameters, only present when
which
is not equal to
"coef" and a variance function
structure is used in
object.