Determining Delivered Functional Error Content Based on the Complexity of CASE Specifications

Despite problems of definition, software complexity has long been recognised as a significant determinant of software product error content. Consequently a large number of empirical studies have attempted to quantify this relationship, using code and design structure measures as complexity indicators. With the increasing use of automated development tools in business software development, however, assessment of the complexity of functional specifications is likely to be more appropriate. This paper therefore describes relevant assessment methods for both complexity and functional errors in this environment. Statistical analysis of significant relationships produced encouraging results based on the development of an error prediction equation.