Kappa katsayısı ve lojit model parametrelerinin EÇOB tahminlerinin araştırılması üzerine bir çalışma

---

Thesis Type: Doctorate

Institution Of The Thesis: Gazi University, Fen Bilimleri Enstitüsü, İSTATİSTİK ANABİLİM DALI, Turkey

Approval Date: 2005

Thesis Language: Turkish

Student: MELTEM EKİZ

Supervisor: HAMZA GAMGAM

Abstract:

IV A STUDY ON THE INVESTIGATION OF MAXIMUM LIKELIHOOD ESTIMATORS OF THE KAPPA COEFFICIENT AND LOGIT MODEL PARAMETERS (Dr. Thesis) Meltem EKİZ GAZI UNIVERSITY INSTITUDE OF SCIENCE AND TECHNOLOGY July 2005 ABSTRACT Judges based on sample units frequencies or probabilities is summarized in a 2x2 table, then the agreement between these rater judges is investigated by the Kappa coefficient Espescially in medical and social studies, it is possible to come across with classification problems such that raters classified each sample unit to the category "succesfully(l)" with distinct probabilities «y. These ny classification probabilities can be estimate from logit models which contains covariates of raters and/or units features. In this study, the maximum of the likelihood function, containing k and logit model parameters is obtained and so the maximum likelihood estimators are founded. Furthermore, the simulation method is used in obtaining the estimates for different sample sizes of each with twenty replications with the benefit of written two programs on Matlab packet programming. Science Code : 212 Key Words : Kappa coefficient, Logit model, Likelihood function Page Number : 103 Adviser : Prof. Dr. Hamza GAMGAM