Software quality revisted

This work presents an exploratory study of the differences in software quality perceptions between different groups of people involved with the software development process. Survey respondents ranked each of thirteen generally accepted attributes of software quality on a scale of one to seven according to their perceived importance for the piece of software most vital to that individual's work: correctness, maintainability, usability, testability, flexibility, portability, reusability, interoperability, integrity, accuracy, robustness, augmentability, efficiency. Differences in the priorities assigned to these attributes were explored using a number of different statistical techniques. The main category of analysis is termed here the "stakeholder role" of the individual with respect to the chosen software, and consists of two dimensions: managerial/non-managerial and development/use. The four primary categories were thus termed developer, development manager, user and user manager. Several additional ways of grouping the individual respondents were explored. Firstly, the organizational sector in which the respondent works (education, military, etc.) was used. Secondly, the class of software (custom developed corporate application, standard mass market package, etc.) was analyzed. Thirdly, the overall view of the nature of quality, according to the taxonomy established by Garvin, was examined. Finally, other personal characteristics of the respondent, including their general interest in software quality issues and years of experience with the software development process, were explored. This study, which including the pilot data examined the responses of almost seven hundred individuals, found that ideas of what constituted software quality were broadly shared between different stakeholder groups. This reassuring finding suggests that software which can achieve high levels of quality according to the attributes valued by all groups, such as correctness and accuracy, will be viewed as satisfactory by all stakeholder groups. The survey did reveal important differences between the priorities assigned to a number of attributes by holders of different roles, particularly usability (favored by users) and testability and maintainability (favored by development staff).