Date

2003

Author

Avcıoğlu, M Didem

Metadata

Show full item record

Item Usage Stats

140
views

0
downloads

Cite This

Experimental design methods have broad coverage of application areas. Usually, basic goal in experimental design methods is to compare the effect of controllable experimental factors on the response and locate the interest on the one, which is most effective. However, in many experimental situations, responses are not only affected by the controllable experimental factors, but also by uncontrollable variates, usually named as covariates. Main aim in these models is to relate the response to both the controllable and uncontrollable variates and investigate the effects of the variates on the response. Analysis of covariance (ANCOVA) is a frequently used method of analysis, especially in the biological sciences. Nevertheless, most of the work done on this model is based on the normality assumption of the error distribution. In practice, however, non-normal distributions occur more frequently. The aim of this study is to investigate this model under non-normality and to develop robust and efficient methods for estimation and hypothesis testing.

Subject Keywords

Experimental design, Analysis of covariance, Maximum Likelihood Estimators (MLE), odified Maximum Likelihood Estimators (MMLE), Robustness, Non-normality

URI

https://hdl.handle.net/11511/13638

Collections

Graduate School of Natural and Applied Sciences, Thesis