Previous: Main page  Main page  Next: Background -- education

	Introduction and Background It has always been difficult to teach computer science to beginners. Part of the problem is that there are few ways to assess a student's programming ability that are both practical and reliable. There seems to be broad agreement that new ways to assess programming skills are needed, but many of the proposed solutions have their own problems with fairness, reliability or cost-effectiveness. Concept mapping, a technique for representing conceptual structures as two-dimensional maps, can be used to represent students' understanding of course material, but the use of traditional concept mapping techniques would involve a substantial amount of organisation. Depending on the method chosen, it may also require a lot of labour to administer or to interpret, rendering it unsuitable. This thesis presents a technique for generating competency maps --- structures similar to concept maps, but covering practical abilities as well as theoretical background knowledge --- from the marks for the assessment tasks that students already undertake. The new technique uses well-understood statistical methods that are widely implemented in off-the-shelf software, but are not too difficult to reimplement should a dedicated tool be needed. It simply makes better use of the data universities already collect, without much overhead on staff and without any extra assessment burden on students.
Concept mapping for introductory programming
Thesis main page
Introduction and background
Background: education
Assessment
Background: concept maps
Aims
Competency mapping
Benefits
Method and results
Data sets
Method
Results
Random data
Analysis and conclusions
Factor analysis
Cluster analysis
Methodological problems
A better test
Conclusion
Appendix I: Datasets
Appendix II: Activities
Appendix III: MDS coordinates
Appendix IV: Data generation scripts
Bibliography

Previous: Main page  Main page  Next: Background -- education