Sıralanabilir Karesel Olumsallık Tablolarında Uyum Parametresi İçeren İlişki Modelleri

Abstract

Kappa coefficient is the most popular measure for summarizing degree of agreement between variables in square tables. However, kappa has some unsufficient features for summarizing the tables by a single number. Hence, the agreement models are suggested to be used in square contingency tables with nominal categories. In recent studies, the square contingency tables having ordered categories are frequently used in many fields, such as medicine, sociology and behavioral sciences. The single way to apply agreement models for such tables is to ignore the ordered structure of the variables. However, this will lead loss of information. In order to analyze the agreement for square contingency tables having ordered categories, the association models with agreement parameter which analyze agreement and association together are suggested. The aim of this study is to introduce the association models with agreement parameter for square contingency tables with ordered categories. Because the variables are ordinal, the effects of different score values on the model fit and parameter estimates are investigated. The models suggested in recent studies are introduced. Additionally, in this study, the association models with agreement parameter are suggested. In two-way and multiway contingency tables, different score equalities are applied on the models through an application with two different data sets and the results are discussed. By means of the degrees of distinguishability which are calculated for the proper models, it is discussed whether the scaling of the categories is assigned accurately or not.