Jackknife-after-bootstrap is a technique for estimating the standard error of
some functional of the bootstrap distribution of parameter estimates. For
example, it may be used to estimate standard errors for the bootstrap
estimate of standard error.
Jackknife-after-bootstrap also calculates relative influences
reflecting the degree of influence each observation has upon the functional
under consideration.
functional of the bootstrap distribution for which standard
error is to be estimated.
May be
a character string specifying a column of
boot.obj$stats (i.e. "Bias",
"Mean", or "SE") or a function expecting a vector as its first argument
and returning a scalar, e.g.
mean.
threshold
observations with an absolute relative influence above this value will be
flagged as particularly influential.
...
other arguments passed to
functional.
frame.eval.boot
frame in which to evaluate the call to
bootstrap used in
resamp.get.indices
to regenerate the indices used in constructing the replicates.
By default,
the function is evaluated in the frame which calls
jack.after.boot.
VALUE:
object of class
jack.after.boot with components
call,
functional,
rel.influence
,
large.rel.influence,
values.functional,
dim.obs,
and
threshold.
WARNING:
Jackknife-after-bootstrap usually overestimates standard errors.
REFERENCES:
Efron, B. and Tibshirani, R. J. (1993).
An Introduction to the Bootstrap.
San Francisco: Chapman & Hall.