An Empirical Study to Validate Metrics for Class Diagrams Marcela Genero, Mario Piattini, Coral Calero Department of Computer Science University of Castilla-La Mancha Paseo de la Universidad, 4 13071, Ciudad Real (Spain) {marcela.genero, mario.piattini, coral.calero}@uclm.es Abstract One of the principal objectives of software engineering is to improve the quality of software products. Quality assurance must be guaranteed from the early stages of the software development life cycle, focusing on high-level design artifacts like class diagrams. Indeed, class diagrams constitute the backbone of object- oriented information systems (OOIS) so, their quality has a great impact on the quality of the product which is finally implemented. To assess class diagram quality, it is useful to have quantitative and objective measurement instruments. After having thoroughly reviewed existing OO measures applicable to class diagrams at a high-level design stage, we presented in a previous work [16, [19] a set of UML class diagram structural complexity metrics, a class diagram internal quality attribute, with the idea that it is related to the external quality of such diagrams. In order to gather empirical evidence that the proposed metrics could be early quality indicators of class diagrams, we carried out a controlled experiment. The aim of which was to investigate the relation between the structural complexity of class diagrams and their maintainability. Keywords. Software quality, structural complexity metrics, class diagram maintainability, empirical validation 1. Introduction. Nowadays, one of the principal objectives of software engineering is to improve the quality of software products. There is a common agreement that the quality assurance must be guaranteed from the early stages of the software development life cycle [4],[8],[9],[10] focusing on high-level design artifacts like class diagrams. In the development of OOIS the class diagram is a key early artifact that lays the foundation of all later design and implementation work. Hence, class diagram quality is a crucial issue that must be evaluated (and improved if necessary) in order to obtain quality OOIS, which is the main concern of present day software development organisations. The early focus on class diagram quality may help IS designers build better OOIS, without unnecessary rework at later stages of the development when changes are more expensive and more difficult to perform. It is in this arena where software measurement plays an important role, because the early availability of metrics contributes to class diagram quality evaluation in an objective way avoiding bias in the quality evaluation process. Moreover, metrics provide a valuable and objective insight into specific ways of enhancing each of the software quality characteristics. Given that maintenance was (and will continue to be) the major resource consumer of the whole software life cycle, maintainability has become one of the software product quality characteristics [21] that software development organisations are more concerned about. Maintainability is not restricted to code, it is an attribute of the different software products we hope to maintain [15], like, for example, class diagrams. However, we are aware that maintainability is an external quality attribute that can only be measured late in