Calculate Percentiles of Resampling Distributions

DESCRIPTION:

Calculates percentiles for the resampled statistics in a resamp object. The quantile function is used to calculate the percentiles of each column of the replicates component of x.

USAGE:

limits.percentile(x, probs=c(0.025, 0.05, 0.95, 0.975), 
                  subset.statistic = T, narrow = F) 

REQUIRED ARGUMENTS:

x
object of class resamp.

OPTIONAL ARGUMENTS:

probs
probabilities at which to calculate percentiles.
subset.statistic
subscript expression; if the statistic that was bootstrapped has length greater than 1, use this to request intervals for only some elements (parameters) of the statistic.
narrow
logical, determines how quantiles are computed. If narrow=FALSE (the default), then intervals are wider and coverage probabilities are usually more accurate than with narrow=TRUE.

VALUE:

matrix of percentiles with one row per parameter.

DETAILS:

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)).

SEE ALSO:

, , , , , , , .

EXAMPLES:

temp <- bootstrap(stack.loss,var) 
limits.percentile(temp)