Coupling and control flow measures in practice

This paper presents the results of applying a set of measures for assessing control flow and both the implicit and explicit coupling of hierarchical and network models of system designs. The measures were validated on three medium sized case studies against the dependent variables of development time and error rate. The first case study was part of a graphical user interface (GUI) developed for a commercial CASE tool. The second case study was the GUI element of the meta-CASE tool developed as part of the research presented here, as such this data-set is not truly independent. However, as the domain is similar to the first case study the results are presented here as illustrative. The third case study was an order processing system for an academic publishing company. To support the application of the developed measures a meta-CASE tool was developed, the architecture of which is presented. The control flow and export coupling measures have proved to be particularly accurate predictors of both development time and error rate. However, no signification correlation was found between the various imports coupling measures and either of the dependent variables.