On a Class of Linear Estimators in Sampling with Varying Probabilities without Replacement
作者:
S.G.Prabhu Ajgaonkar,
期刊:
Journal of the American Statistical Association
(Taylor Available online 1965)
卷期:
Volume 60,
issue 310
页码: 637-642
ISSN:0162-1459
年代: 1965
DOI:10.1080/01621459.1965.10480819
出版商: Taylor & Francis Group
数据来源: Taylor
摘要:
Horvitz and Thompson [3] have suggested for the estimation of population characteristic certain classes of linear estimators depending on weights associated with the sample observations, when a sample of sizenis drawn without replacement using arbitrary probabilities of selection for each draw. However, it should be observed here, that there might not exist even a single linear unbiased estimator for the class which is termed an empty class. A situation arises where the minimum variance linear unbiased estimator for the non-empty class depends on the population values. In such a case, the criterion of the ‘necessary best estimator’ is proposed, in order to choose a serviceable estimator purely from the practical point of view. This is illustrated by one of the classes of linear estimators (T5).
点击下载:
PDF (318KB)
返 回