The Generalized Jackknife: Finite Samples and Subsample Sizes
作者:
Trevor Sharot,
期刊:
Journal of the American Statistical Association
(Taylor Available online 1976)
卷期:
Volume 71,
issue 354
页码: 451-454
ISSN:0162-1459
年代: 1976
DOI:10.1080/01621459.1976.10480367
出版商: Taylor & Francis Group
数据来源: Taylor
摘要:
This paper is concerned with bias-reduction properties of that generalized jackknife which pertains to a single-estimator, single-sample situation. Jackknifing is a technique for reducing bias by exploiting the dependence of the bias on sample size. In practice, this is carried out by reestimating the unknown parameter(s) with only part of the sample and combining the new and original estimates suitably weighted to produce cancellation in their biases. The size of the subsample affects both the bias and variance of the jackknife. While few results are available, “minimal data omission” seems sensible on grounds of variance. This paper shows the same is always true on grounds of bias.
点击下载:
PDF (283KB)
返 回