تحليل داده هاي رتبه اي و همبسته پزشكي به كمك معادلات برآوردگر تعميم يافته

The analysis of correlated ordered medical data is usually more complicated than continuous and binary responses. When we wish to model correlated ordinal responses, using common approaches, such as MLE methodology, for estimating regression parameters is usually time-consuming and requires complicated computer programs. In this paper, we utilize a marginal proportional odds regression model for analyzing correlated ordered categorical data. In addition, we use Generalized Estimating Equations methodology in order to estimate the regression parameters of this model. This theoretically simple method is constructed based on quasi likelihood approach. We also introduce different structures to account for the correlation between response data and present the estimating method for each of these structures. Finally, the described model is used to analyze data from an epidemiologic study relating to periodontal condition of students aged 15-19 years in Tehran high schools.