Adaptive Testing with Granularity

In presenting this thesis in partial fullllment of the requirements for a Postgraduate degree from the University of Saskatchewan, I agree that the Libraries of this University may make it freely available for inspection. I further agree that permission for copying of this thesis in any manner, in whole or in part, for scholarly purposes may be granted by the professor or professors who supervised my thesis work or, in their absence, by the Head of the Department or the Dean of the College in which my thesis work was done. It is understood that any copying or publication or use of this thesis or parts thereof for nancial gain shall not be allowed without my written permission. It is also understood that due recognition shall be given to me and to the University of Saskatchewan in any scholarly use which may be made of any material in my thesis. Requests for permission to copy or to make other use of material in this thesis in whole or part should be addressed to: Abstract Adaptive testing attempts to determine the current level of a student's knowledge quickly and accurately by posing questions. Depending on the student's responses, an adaptive testing routine can estimate the student's level of knowledge to some precision. If more information is required to make a judgement, the routine can choose to pose more questions until the routine terminates with a nal assessment. Adaptive testing has been severely limited in the past since it was only able to assess a single learner trait. In realistic situations, many learner traits may need to be measured at one time. Recent work in student modelling has identiied many schemes for representing the multiple concepts present in any course and the relationships between them. This thesis presents a general scheme for using granularity hierarchies to represent a course hierarchy of multiple concepts and Bayesian networks to select test items and to update belief about a student's knowledge. The complexity of specifying such courses and their associated prior and conditional probabilities is reduced through the development of an authoring environment. An empirical study was performed to test the eeectiveness of this new scheme in terms of test length and test precision. The study compared these results against other adaptive testing routines and traditional, hand{crafted tests. ii Acknowledgements There are a great number of people to whom I am indebted and without whom …