Using a quality model approach to define and evaluate metrics for object-oriented programming systems

The use of the object-oriented paradigm has increased rapidly in the last decade. Since it was introduced many claims have been made about its benefits over traditional methods. These claims have been largely based on anecdotal evidence, with little theoretical or empirical work quoted.This thesis assesses the current state of work in the object oriented metrics field. Existing object oriented design metrics and quality models are analyzed. Theoretical and empirical validation is performed on existing metrics suites. The lessons learned from this validation are used as a basis for the development of a new set of metrics: Metrics for Object Oriented Programming Systems (MOOPS). Issues raised by current quality models such as a lack of definition at lower levels, are addressed by the MOOPS Quality Model. The MOOPS Quality Model and Metrics are used as the basis of an approach which combines existing work in the three areas of theoretical validation, empirical validation and quality models to achieve the overall goal:To improve the prediction of quality from a developer's viewpoint.A theoretical validation framework is constructed and used to validate the MOOPS Metrics and an empirical validation is performed using both industrially and laboratory based software systems.