Managing multiple interdependencies in large scale software development projects

The dissertation research is structured as three papers, each of which explores a specific component of the interdependence-structure-performance relationship. The first paper, "Interdependency: Conceptual, Empirical, and Practical Issues," surveys the literature on interdependency across three distinct theoretical paradigms, information processing theories, resource-based theories, and theories of sense-making, and develops a conceptual framework that places these competing paradigms in perspective. It also documents how researchers have measured and evaluated interdependence and notes a decided uniformity in past empirical approaches to the topic. The second paper, "An Empirically-Derived Taxonomy of Multiple Interdependencies" attempts to open up the "black box" of task relationships. Drawing upon a large sample of examples of interdependency, it develops, via affinity diagram techniques, a taxonomy of multiple interdependencies. Given that model, the third paper, entitled "Webs of Interdependency: Strategies for Managing Multiple Interdependencies in New Product Development," examines the strategies teams use to manage multiple interdependencies and proposes implications for project performance, which are supported by internal and external project evaluation data. The analysis revealed three basic strategies: systemic manipulation, constrained systemic manipulation, and local reaction. In this study, the highest performing projects adopted a systemic manipulation approach in which they focused on the pattern of interdependencies as a design variable. These teams made critical up front design decisions in order to simplify task coordination, prioritized interdependencies, gained leverage off of central tasks, and were aware of the path dependent nature of their choices. Results suggested that interdependency management can be viewed as a strategic process variable insofar as interdependencies are manipulatable through strategy and organization design. This flexibility is constrained, however, due to the existence of certain technical and organizational legacy factors, which tend to embed tasks and task relationships in a prior history of structure. The research design of this study is that of multiple case studies (Yin, 1984) of six large scale software development projects at two leading companies, Microsoft and Lucent Technologies (formerly ATT Cusumano & Selby, 1995). Utilizing a multi-method approach, the study draws upon a combination of qualitative and quantitative data including 71 interviews with project team members, on-site observation at both firms, and various forms of project documentation such as schedules, design documents, and memos. The outcome variable (project performance) was assessed via a brief questionnaire administered to experts and upper management in each firm. Because the study was designed to be an inductive investigation, the above hypotheses await future testing. (Copies available exclusively from MIT Libraries, Rm. 14-0551, Cambridge, MA 02139-4307. Ph. 617-253-5668; Fax 617-253-1690.) (Abstract shortened by UMI.)