Previous: Background -- concept maps  Main page  Next: Competency mapping

Aims The aim of this project is to use concept mapping principles to develop a tool that can be used to improve formative assessment and assist in course evaluation for introductory computer programming without adding significantly to the course load on students and without undue cost to the university, either in labour or in materials. A student-derived concept map of introductory computer programming would certainly help with assessment and evaluation. By showing the degree of relatedness of parts of the course as the students see it, it would expose any "missing links": parts of the course which should be related, but which are not close together on the student-derived concept map. Such missing links may indicate that students are failing to construct part of the course correctly, because they are mentally pigeonholing concepts that should be unified. The syllabus can then be refined, to make the connections between these topics more explicit. In effect, course designers can use the results of concept mapping to provide assembly instructions for the construction of new knowledge. Such a concept map could be generated by the students according to Trochim's method. However, organising the brainstorming sessions that would be required would be painful: if the activity is not summative, the students are unlikely to do it; yet it is difficult to see how it can be made summative. Moreover, any activity needs to be considered carefully before being added to the course. If students feel pressured, their learning tends to suffer as they forsake deep learning strategies for surface learning (Thomson and Falchikov, 1998). Furthermore, overseeing the collation, validation and entry of several hundred sets of grouping information would be a daunting and labour-intensive task. It seems that Trochim's method would not achieve the feasibility goal. But universities already collect and store a great deal of information about how students conceptualise their subjects: student results. Obviously, if two exam questions are testing the same basic concept then the marks for those questions should show a positive correlation, unless something has gone drastically wrong with the delivery of the question. This brings us to my key insight: it should be possible to work backwards and infer conceptual relatedness from strength of correlation. Multidimensional scaling and cluster analysis could be applied to correlation matrix calculated from students' marks, producing an empirically-derived concept map.

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: Background -- concept maps  Main page  Next: Competency mapping