Genelleştirilmiş Tahmin Denklemleri ve Regresyon Yöntemi İle Başarı Kestiriminde Karşılaştırılması

Abstract

In this research, the purpose is to compare the efficiency of generalized estimating equations (GEE) with fixed effect logistic regression (FELR) method for the analysis of clustered binary data variable in terms of different types of independent variables (continuous and categorical) different sample sizes (10x100; 20x50; 100x10), intraclass correlation coefficients (0,3; 0,5; 0,8) by using obtained estimated regression coefficients, standart errors and probability of type II error.Besides, within this research the relationship of the science achievements of the countries in PISA 2003, 2006 and 2009 projects with the countries gross enrollment ratios (ger), pupil-teacher ratios (ptr), ratio of educational expenditures to the GDP (gdp) and expected educational year (lee) is analysed by both methods. It is found that the regression model is not significant for both analysis methods.