summary.crq {crq} | R Documentation |
Confidence intervals and standard error computations for censored quantile regression.
## S3 method for class 'crq': summary(object, taus = 1:4/5, alpha = .1, se = "boot", ...)
object |
An object of class crq . |
taus |
A vector of quantiles at which to evaluate estimates, their confidence intervals and standard errors. |
alpha |
|
se |
A method for estimating standard errors. |
... |
Other arguments to be passed to boot.crq , etc. |
Only one method is (so far) available for confidence interval and standard error computations; it uses the bootstrap to compute rescaled interquartile range estimates. Two flavors of the bootstrap are available: by default one obtains the conventional xy-pair bootstrap, if bmethod = "Bose" a weighted resampling method with standard exponential weights is used. Confidence intervals are computed by rescaling the interquartile range, of the bootstrap realizations as are the standard errors as suggested in Portnoy (2003).
A list of class summary.crqs each member consists of a tau component and a coefficient component. The latter component is a matrix containing the point, interval and standard error estimates.
Stephen Portnoy & Tereza Neocleous
boot.crq
, crq