Simultaneous small-sample multivariate Bernoulli confidence intervals.

Research output: Contribution to journalArticle

12 Scopus citations

Abstract

A technique based on the bootstrap is presented for assessing the simultaneous confidence level of k small-sample confidence intervals for multivariate Bernoulli marginal frequencies. The small-sample intervals used are those of Clopper and Pearson (1934, Biometrika 26, 404-413) and require iterative computation. To estimate the simultaneous confidence level, the multivariate Bernoulli vectors are resampled via the bootstrap and the Clopper-Pearson intervals recomputed on each pseudosample. The bootstrap estimate is then the proportion of times (computed via Monte Carlo) that all the k intervals computed by resampling contain the original sample frequencies. The technique is applied to single-sample HLA data.

Original languageEnglish
Pages (from-to)1001-1013
Number of pages13
JournalBiometrics
Volume41
Issue number4
DOIs
StatePublished - Dec 1985

Fingerprint Dive into the research topics of 'Simultaneous small-sample multivariate Bernoulli confidence intervals.'. Together they form a unique fingerprint.

  • Cite this