On Test Sizes in Linear Models for Transformed Variables
作者:
Camil Fuchs,
期刊:
Technometrics
(Taylor Available online 1978)
卷期:
Volume 20,
issue 3
页码: 291-299
ISSN:0040-1706
年代: 1978
DOI:10.1080/00401706.1978.10489674
出版商: Taylor & Francis Group
关键词: Criteria of transformation;Linear models;Test sizes
数据来源: Taylor
摘要:
The effect of the transformation on the subsequent analyses as expressed by ANOVA tests is emphasized here in contrast to the search for a “correct” transformation. New criteria of transformation based on half-normal standard deviates of statistics which test homogeneity of variances and additivity are presented. These criteria, the maximum likelihood criterion of Box and Cox [8] and the criterion of Schlesselman [22] are compared by Monte Carlo simulations for some linear models.
点击下载:
PDF (867KB)
返 回