resamp
object.
The
quantile
function is used
to calculate the percentiles of each column of the
replicates
component of
x
.
limits.percentile(x, probs=c(0.025, 0.05, 0.95, 0.975), subset.statistic = T, narrow = F)
resamp
.
narrow=FALSE
(the default), then intervals are wider and coverage probabilities
are usually more accurate than with
narrow=TRUE
.
When bootstrapping, quantiles produced by
limits.percentile
are known as bootstrap percentile confidence intervals.
They are not particularly accurate; under general
conditions they are
first-order accurate (coverage errors O(1/sqrt(n))),
while BCa and tilting limits are second-order accurate (O(1/n)).
temp <- bootstrap(stack.loss,var) limits.percentile(temp)