Abstract
The thesis is a study of the techniques used for the analysis of student responses to multiple choice type objective examinations. Essentially two methods of analysis are discussed, the First one, known as the traditional method, is the one in most common usage today. The second involves the fitting of a mathematical function to the item characteristic curve and is consequently known as the curve fitting method. The mathematical function used here is the integrated normal function although the possible use of other functions is discussed. The full details and characteristics of both the methods are included and in the curve fitting method particular stress has been laid on the sample free nature of the difficulty index. The sample free nature of this index has been investigated empirically. Results are also included which show the optimum working conditions of the normal ogive model and its use in two novel situations. The thesis concludes with a discussion of the item selection techniques which are used with the traditional methods of analysis and sets out a strategy for item selection using the curve fitting methods.