USING GENERALIZED ESTIMATING EQUATION TO ANALYZE DATA ON QUALITY EDUCATION

Abstract

The method of Generalized Estimating Equation has been of great use in the research world of analysis of correlated data. Data sets that are collected for a single subject over several time frames exhibit this property which makes the use of certain statistical methods non-suitable for analysis. One of the major problems encountered in such data sets is the modeling of the within-subject correlation that is responsible for the structure of correlation that exists between observations belonging to a particular subject, when such information is not taken into considerations, it leads to gross invalidity of inferences. In this research work, the study of continuous response variable was showed by using the GEE to estimate the parameters in a fitted model. It is an application to a real life educational data. The data sets were obtained from Osogbo Grammar School in Osun State of Nigeria between 2014 and 2015. It comprises the students’ performance records through their performance sheet in three consecutive terms under study together with the students’ response of their economic and environmental status, this was done through the administration of marked questionnaires given to each student. This was administered to mainly SS2 students. The environmental variables were reduced to just nine through the use of factor analysis and these reduced variables were combined with the economic status as the independent variables on their performance through the use of GEE. The four mostly used ‘working correlation structure’ (independent, Autoregressive (AR1), Exchangeable and Unstructured) are used to fit four different models using R programming language, where the model fitted using Exchangeable ‘Working Correlation’ was adjudged the best structure. The effect of subject, house type and income all have significant effects on the performance of the students. However, some recommendations were also made based on the outcome of the research work.