Date
2009 - 2013

Research Degree Project

The thesis is concerned with evaluating quizzes when they are marked by a computer and explores the problems associated with quizzes when they are used for summative assessment. The types of quizzes maybe simple Yes/No, 4 and 5-option Multiple Choice quizzes (MCQs), Multiple Answer quizzes (MAQs), and more complex quizzes like Multiple Yes/No

Many tutors use quizzes because they believe that they are quick, reliable, valid and objective. Future work, will establish whether this is the case.

One of the main problems associated with quizzes is the guessing problem. Since this will cause marks to be inflated it inevitably means that some students are able to pass the test without sufficient knowledge of the domain. A number of solutions to this have been put forward but all of them have problems associated with them. This thesis examines these solutions.

The Research Question (RQ) is "What is the distribution of marks of quizzes?" and the sub-text is "What is the effect of tackling the guessing problem?

This is a theoretical study, and so 'real' people are not involved and so a) we do not have to gather any data from them b) the method of the study is mainly quantitative. Any data that we need is generated but it requires very large numbers and usually computer languages cannot handle very large numbers. For handling arbitrarily large integers we use a library called LEDA (written by a German firm called Algorithmic Solutions) and use the programming language C. To show that the C programs are right we validate them, where we can, by a) proofs that we develop b) hand calculations and c) using 'small' sets.

The people that should benefit from our study are any academics (not just at University) that use quizzes for assessment.

Project Supervisors

• Dr Keith Burley (Director of Study)
• Paul Crowther (Second Supervisor)

Researchers involved

Hugh Lafferty - Research Degree Student - Alumni (2009-2014)