On the Influence of Task Size and Template Provision on Solution Similarity

Abstract

In most cases of programming education, there is not a single correct answer to a given task. Instead, the same problem can be solved by two or more pieces of program code that look very different. At the same time, two or more pieces of program code that look very similar may actually solve very different problems. It is thus not easy to foresee which degree of similarity one can expect for all or at least the correct submissions to a given programming task. Since several applications may benefit from some kind of prediction of the similarity, this paper presents first, preliminary results from research on that topic. In particular, it presents results from an empirical study on the influence of exercise size and template provision. Results indicate that both factors are not suitable as simple predictors and that other factors have to be taken into account as well. Nevertheless, the results help to generate hypothesis for more detailed subsequent studies.