الملخص الإنجليزي
This study investigates some popular methods used to analyze and interpret Likert-type scale data.
This study has covered three main methods and their statistical properties. These include logistic regression methods, the use of indices approach, and paired comparison method. It also touches briefly on some other methods developed to address this type of data. Among these are Bayesian methods, Rough Set method (RS) and Factor analysis (FA).
The use of indices in surveys is a new way to measure the concentration on positive or negative responses based on observed proportions of answers. Extreme responses are made to carry more weights to reflect the strength of positive or negative attitudes. We have also suggested the use of natural logarithm as weights, such that extreme responses should be balanced on both sides of the neutral category. We have proposed an extension of weighted indices to be used to analyze sets of concepts, each consisting of a set of questions. Among the three methods that have been discussed the use of indices is the easiest. It needs little assumptions, less computations, and is easy to implement and interpret. On the other hand more study is required to investigate some of its characteristics such as the range of the indices and to suggest new weights. Furthermore, future studies could be focused on the use of the Rough Set method (RS) as another new and promising way to address the ordinal data.
