Hostname: page-component-cd9895bd7-dzt6s Total loading time: 0 Render date: 2024-12-26T01:45:14.907Z Has data issue: false hasContentIssue false
  • English
  • Français

A note on coverage error of bootstrap confidence intervals for quantiles

Published online by Cambridge University Press:  24 October 2008

D. De Angelis ,
Peter Hall  and
G. A. Young
D. De Angelis
Affiliation:
University of Rome, ‘La Sapienza’
Peter Hall
Affiliation:
CMA, Australian National University
G. A. Young
Affiliation:
DPMMS, 16 Mill Lane, Cambridge
Get access Rights & Permissions [Opens in a new window]

Abstract

An interesting recent paper by Falk and Kaufmann[11] notes, with an element of surprise, that the percentile bootstrap applied to construct confidence intervals for quantiles produces two-sided intervals with coverage error of size n−½, where n denotes sample size. By way of contrast, the error would be O(n−1) for two-sided intervals in more classical problems, such as intervals for means or variances. In the present note we point out that the relatively poor performance in the case of quantiles is shared by a variety of related procedures. The coverage accuracy of two-sided bootstrap intervals may be improved to o(n−½) by smoothing the bootstrap. We show too that a normal approximation method, not involving the bootstrap but incorporating a density estimator as part of scale estimation, can have coverage error O(n−1+∈), for arbitrarily small ∈ > 0. Smoothed and unsmoothed versions of bootstrap percentile-t are also analysed.

Type
Research Article
Information
Mathematical Proceedings of the Cambridge Philosophical Society , Volume 114 , Issue 3 , November 1993 , pp. 517 - 531
Copyright
Copyright © Cambridge Philosophical Society 1993

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

REFERENCES

[1]Beran, R.. Prepivoting to reduce level error of confidence sets. Biometrika 74 (1987), 457–468.CrossRefGoogle Scholar
[2]Beran, R. and Hall, P.. Interpolated nonparametric prediction intervals and confidence intervals. J. Roy. Statist. Soc. Ser. B 55 (1993), 643–652.Google Scholar
[3]Bloch, D. A. and Gastwirth, J. L.. On a simple estimate of the reciprocal of the density function. Ann. Math. Statist. 39 (1968), 1083–1085.CrossRefGoogle Scholar
[4]Bowman, A. W. and Hall, P.. Empirical determination of smoothing for the bootstrap. Unpublished.Google Scholar
[5]Angelis, D. De. Bootstrap smoothing of the bootstrap. Unpublished.Google Scholar
[6]Angelis, D. De and Young, G. A.. Smoothing the bootstrap. Internat. Statist. Rev. 60 (1992), 45–56.CrossRefGoogle Scholar
[7]Angelis, D. De and Young, G. A.. Bootstrapping the correlation coefficient: a comparison of smoothing strategies. J. Statist. Comput. Simul. 40 (1992), 167–176.CrossRefGoogle Scholar
[8]Efron, B.. Bootstrap methods: another look at the jackknife. Ann. Statist. 7 (1979), 1–26.CrossRefGoogle Scholar
[9]Efron, B.. The Jackknife, the Bootstrap and Other Resampling Plans. SIAM, Philadelphia (1982).CrossRefGoogle Scholar
[10]Falk, M. and Janas, D.. Edgeworth expansions for studentized and prepivoted sample quantiles. Statist. Prob. Lett. 14 (1992), 13–24.CrossRefGoogle Scholar
[11]Falk, M. and Kaufmann, E.. Coverage probabilities of bootstrap-confidence intervals for quantiles. Ann. Statist. 19 (1991), 485–495.CrossRefGoogle Scholar
[12]Falk, M. and Reiss, R.-D.. Weak convergence and smoothed and nonsmoothed bootstrap quantile estimates. Ann. Prob. 17 (1989), 362–371.CrossRefGoogle Scholar
[13]Hall, P.. Theoretical comparison of bootstrap confidence intervals. (With discussion.) Ann. Statist. 16 (1988), 927–985.Google Scholar
[14]Hall, P.. Edgeworth expansions for nonparametric density estimation, with applications. Statistics 21 (1991), 215–232.CrossRefGoogle Scholar
[15]Hall, P., Diciccio, T. J. and Romano, J. P.. On smoothing and the bootstrap. Ann. Statist. 17 (1989), 692–704.CrossRefGoogle Scholar
[16]Hall, P. and Sheather, S. J.. On the distribution of a Studentised quantile. J. Roy. Statist. Soc. Ser. B 50 (1988), 381–391.Google Scholar
[17]Hettmansperger, T. P. and Sheather, S. J.. Confidence intervals based on interpolated order statistics. Statist. Prob. Lett. 4 (1986), 75–79.CrossRefGoogle Scholar
[18]Reiss, R.-D.. Asymptotic expansions for sample quantiles. Ann. Prob. 4 (1976), 249–258.CrossRefGoogle Scholar
[19]Sheather, S. J.. Assessing the accuracy of the sample median: estimated standard errors versus interpolated confidence intervals. In Statistical Data Analysis Based on the L1-Norm, ed. Dodge, Y., pp. 203–215 (North-Holland, Groningen 1987).Google Scholar
[20]Sheather, S. J. and McKean, J. W.. A comparison of testing and confidence interval methods for the median. Statist. Prob. Lett. 6 (1987), 31–36.CrossRefGoogle Scholar
[21]Silverman, B. W.. Density Estimation for Statistics and Data Analysis (Chapman and Hall, 1986).Google Scholar
[22]Silverman, B. W. and Young, G. A.. The bootstrap: to smooth or not to smooth. Biometrika 74 (1987), 469–479.CrossRefGoogle Scholar
[23]Wang, S.. On the bootstrap and smoothed bootstrap. Commun. Statist. Theory Meth. 18 (1989), 3949–3962.CrossRefGoogle Scholar
[24]Young, G. A.. A note on bootstrapping the correlation coefficient. Biometrika 75 (1988), 370–373.CrossRefGoogle Scholar
[25]Young, G. A.. Alternative smoothed bootstraps. J. Roy. Statist. Soc. Ser. B 52 (1990), 477–484.Google Scholar