As-sessing the Modifiability of two Object-oriented Design Alternatives - a Controlled Experiment Replication
Ioannis Deligiannis, Panagiotis Katsaros, Ioannis G. Stamelos, Panagiotis Sfetsos, Elena De Angelis, A. Xatzigeorgiou · 2004
This paper presents a replication study of a controlled experiment, investigating the impact of many design characteristics on one of the most desirable quality factors, modifiability. Two alternative design structures were used; a responsibility-driven (RD) versus a control-oriented “mainframe ” (MF) design. Two groups of undergraduate students participated, each performing on one of the two designs. The subjects designed, implemented in Java, and tested a set of three maintenance tasks in order to assess the degree of their understanding, effort, and performance. The results indicate that the RD version due to its delocalised structure, exhibited higher correctness, better extensibility, and design stability, than the MF version. In order to provide an objective assessment of the differences between the two versions, a considerable number of metrics were used on the delivered solutions, quantifying separately each produced design’s characteristics.