A Test for the Simultaneous Detection of Two Outliers Among Extreme Values of Small Data Sets
作者:
ConstantinosE. Efstathiou,
期刊:
Analytical Letters
(Taylor Available online 1993)
卷期:
Volume 26,
issue 2
页码: 379-390
ISSN:0003-2719
年代: 1993
DOI:10.1080/00032719308017392
出版商: Taylor & Francis Group
关键词: Q-test;Detection of Outliers;Extreme Values
数据来源: Taylor
摘要:
A simple test (QP-test) for the simultaneous detection of two outliers among extreme values of small data sets is described. This test is analogous and complements Dixon's classic Q-test. It is based on the probability distribution of the product of two Qexpvalues (QP: quotient product). Each Qexpvalue is calculated from the data set considering the presence of one and ignoring the presence of the other suspect value. Critical QP-values, calculated by a stochastic (a Monte-Carlo) procedure, are given for data sets containing 5 to 16 observations (replicates) at 0.90, 0.95, and 0.99 confidence levels. Based on the same principle, tests for the simultaneous detection of more than two outliers can be readily designed.
点击下载:
PDF (438KB)
返 回