Calculate BCa Confidence Limits

DESCRIPTION:

Calculates BCa (bootstrap bias-correct, adjusted) confidence limits.

USAGE:

limits.bca(boot.obj, probs=c(0.025, 0.05, 0.95, 0.975),  
           details=F, z0=NULL, acceleration=NULL,  
           group.size=NULL, frame.eval.jack=sys.parent(1)) 

REQUIRED ARGUMENTS:

boot.obj
object of class bootstrap.

OPTIONAL ARGUMENTS:

probs
probabilities for one-sided confidence limits; e.g. c(.025, .975) gives a two-sided 95% confidence interval. Note that values are undefined at 0 and 1.
details
logical flag indicating whether to return the values of z0, acceleration, group.size, and the empirical probability levels for each BCa confidence limit along with the BCa limits. If detail=F, only the BCa confidence limits are returned.
z0
vector of bias-corrections which is usually calculated within limits.bca.
acceleration
vector of acceleration values which is usually calculated within bca.percent. Note that jackknifing is performed to calculate this value, so a statistic must allow samples of reduced size.
group.size
number of observations to jackknife from each sample when calculating acceleration. It is time consuming to do a full delete-1 jackknife, so the default is to use a group size of floor(n/20), which yields roughly 20 groups of points, and hence 20 jackknife replicates. For the full jackknife, use group.size=1.
frame.eval.jack
frame in which to evaluate jackknife. By default, the frame from which limits.bca was called. This is where S-PLUS will look for the data used to construct the bootstrap object.

VALUE:

if detail=F, returns a matrix of BCa confidence limits with one row per parameter of interest. Otherwise, returns a list with these limits as the limits.bca component, plus the empirical probability levels, z0 , acceleration, and group.size.

REFERENCES:

Efron, B. and Tibshirani, R. J. (1993). An Introduction to the Bootstrap. San Francisco: Chapman & Hall.

Shao, J. and Tu, D. (1995). The Jackknife and Bootstrap. New York: Springer-Verlag.

SEE ALSO:

, , .

EXAMPLES:

temp <- bootstrap(stack.loss, var) 
limits.bca(temp) 
limits.bca(temp, detail=T)