Analysing Ordered Categorical Data: Various Methods to Attempt and Discrepancies in Interpretations to Accept

Abstract

This study concerns with a number of methods that are used to analyze ordered categorical data, the data which are related to variables having a set of finite and distinctive classifications. This type of data is used extensively in all fields, especially in sociology, education, psychology, medicine, and biology. In most situations the categories of these data are natural and sometimes are ordered. A number of methods have been reviewed and applied in this context, ranging from the chi-squared test of independence to log-linear and logit models. In addition to compare the theoretical background of these techniques, the study compares the results given by these methods explaining areas where they lead to similar conclusions and where they lead to different interpretations. The study gives empirical estimates of the effects between the variables, explains estimates corresponding to the ordering nature of the variables, and, more importantly, takes into account whether one variable is to be treated as a response. The data for this research are collected from a random sample of 300 students at the Teachers College of Makkah and the concern is devoted to the academic performance of students and associated factors. The data of the study are analysed by SPSS (Statistical Package for the Social Sciences).