Abstract:
This thesis considers logistic regression and discriminant analysis. Because the ordinary least regression requires that a dependent variable in a study cannot be categorical, the logistic regression and the discriminant analysis techniques are two major techniques that are often used to handle categorical dependent variable problems. This study, therefore, seeks to getting answer to the question of whether using the logistic regression and the discriminant analysis techniques, on the same data set, would yield the same result or not.
Since the logistic regression and the discriminant analysis methods relates in many ways, extensive review of the theories behind them and comparison, in terms of similarities and differences, were necessary and have therefore been captured in the study. The graphical nature of the two techniques, interpretation of results, in using the two techniques, and significant tests of the various aspects of the two techniques are also not left out. At the end, empirical comparison of the binary logistic regression and the two-group discriminant analysis was made and the result for this comparison suggests that logistic regression gives a better result than two-group discriminant analysis when all requirements and assumptions of the two techniques are met.
Finally, summary of the findings of the research is also captured in the study and consideration also given to discussion and conclusion about the findings.
