SPLIT–PLOT DESIGN: MODEL AND ANALYSIS FOR COUNT DATA
作者:
J. Engel,
期刊:
Statistica Neerlandica
(WILEY Available online 1986)
卷期:
Volume 40,
issue 1
页码: 21-33
ISSN:0039-0402
年代: 1986
DOI:10.1111/j.1467-9574.1986.tb01161.x
出版商: Blackwell Publishing Ltd
关键词: Split–plot design;count response data;loglinear model;Dirichlet multinomial distribution;negative binomial distribution
数据来源: WILEY
摘要:
AbstractThe analysis of count response data from designed experiments is well–known for independent response variables having the Poisson distribution. For experimental designs where responses are dependent, no general results seem to be available. An example of this type of design is the split–plot design, where sub–plot responses are essentially dependent within whole plots.In this paper, a model will be proposed for split–plot count data and a separate analysis for whole plot and sub–plot data will be presented. It is interesting to note, that the same model is used in the quite different context of consumer buying behaviour. It is derived by Goodhardt, Ehrenberg and Chatfield and it was called the ‘Dirichlet model’, as the Dirichlet multinomial distribution, together with the negative binomial distribution, build
点击下载:
PDF
(490KB)
返 回