POSITION PAPERS 197 The Study of Software Maintenance Organizations and Processes CAROLYN B. SEAMAN Inst. for Advanced Computer Studies, University of Maryland, College Park, MD VICTOR R. BASILI Inst. for Advanced Computer Studies, University of Maryland, College Park, MD In order to improve the quality of software products, it is necessary to enhance the quality of the software processes used to develop them. This recognition has led to a proliferation of work, both empirical and non-empirical, on software development processes, some of which has dealt specifically with maintenance. One important but often overlooked component of both development and maintenance processes is the organizational context in which they are enacted. Our position is that this is an even more pertinent issue in maintenance than in development because maintainers are not generally the original developers of a system. The software has a history in which a number of people have been involved, and the experience of those people becomes relevant during maintenance. Thus an efficient flow of information is crucial, and this flow can be hindered or facilitated by the organizational relationships between maintainers and other parts of the organization. It is the network of these relationships (reporting relationships, physical proximity, past working relationships, etc.) that constitute organizational structure. Furthermore, a variety of methods must be employed to study organizational issues in maintenance, including both quantitative and qualitative approaches. The authors have participated in two studies of software maintenance organizations which illustrate two different approaches to studying organizational issues in software mainte- nance. The first (Seaman, 1996) is a mostly quantitative study in which variables were defined and evaluated, allowing for statistical analysis. The goal of that study was to dis- cover possible relationships between the study variables, which measured organizational structure, communication, and productivity. The second study (Briand, et al., 1995) was a qualitative study which produced a rich descriptive model of the maintenance organization and process. The goal of this study was to highlight anomalies and possible problems in both the organization and process. In (Seaman, 1996), we examined information flow and organizational structure among a group of software maintainers. The dependent variable in this study was communication effort, defined as the total effort expended to share some type of information. The choice of dependent variable followed directly from the goal of the study. The choice of indepen- dent variables was more difficult. They were designed with care, but must be considered exploratory. All the independent variables were measures of the organizational structure, defined as the network of relationships between members of the software development or- ganization. The types of relationships upon which these measures were based are official reporting relationships, physical proximity, and past and present working relationships. The study took place at IBM Software Solutions Laboratory in Toronto, Canada. The