Vol: 30 Page: 179

Authors:
Andrzej Zieliński

Title:
 ML estimation of power transformation parameter lambda in a two-way split-plot model

Language: English

Keywords:
split-plot model
power transformation
profile loglikelihood

Summary:
In the paper estimation of power transformation parameter λ is carried out under assumption that transformed data form a two-way split-plot model with the vector of residuals following normal distribution. Maximum likelihood estimate is a value of λ that maximizes profile log-likelihood i.e. function with maximum likelihood estimates of parameters and variance components instead of parameters and variance components themselves. The form of the profile log-likelihood is presented in the paper. The theory is illustrated with data concerning savouriness of some potato varieties.