resample (version 0.4)

CI.percentile: Bootstrap confidence intervals

Description

Bootstrap confidence intervals - percentile method or t interval.

Usage

CI.percentile(x, probs = c(0.025, 0.975), expand = TRUE, ...)
CI.t(x, probs = c(0.025, 0.975))
CI.bca(x, probs = c(0.025, 0.975), expand = TRUE, L = NULL, ...)
CI.bootstrapT(x, probs = c(0.025, 0.975))

Arguments

x

a bootstrap or bootstrap object.

probs

probability values, between 0 and 1. The default vector c(0.025, 0.975) gives a 95% two-sided interval.

expand

logical, if TRUE then use modified percentiles for better small-sample accuracy.

…

additional arguments to pass to quantile.resample and quantile.

L

vector of length n, empirical influence function values. If not supplied this is computed using jackknife.

Value

a matrix with one column for each value in probs and one row for each statistic.

Details

CI.bootstrapT assumes the first dimension of the statistic is an estimate, and the second is proportional to a SE for the estimate. E.g. for bootstrapping the mean, they could be the mean and s. This is subject to change.

CI.bca and CI.bootstrapT currently only support a single sample.

References

This discusses the expanded percentile interval: Hesterberg, Tim (2014), What Teachers Should Know about the Bootstrap: Resampling in the Undergraduate Statistics Curriculum, http://arxiv.org/abs/1411.5279.

See Also

bootstrap, bootstrap2, ExpandProbs (for the expanded intervals).

Examples

# NOT RUN {
# See full set of examples in resample-package, including different
# ways to call all four functions depending on the structure of the data.
data(Verizon)
CLEC <- with(Verizon, Time[Group == "CLEC"])
bootC <- bootstrap(CLEC, mean, seed = 0)
bootC2 <- bootstrap(CLEC, c(mean = mean(CLEC), sd = sd(CLEC)), seed = 0)
CI.percentile(bootC)
CI.t(bootC)
CI.bca(bootC)
CI.bootstrapT(bootC2)
# }