A model and method for measuring information system size

This thesis develops a measurement model and method that allows information system professionals to establish measures of information system size that are accurate and may be established early in the system development life cycle. It reports the results of an empirical investigation into the aspects of requirements and design metrics which lead to the production of source code. The theoretical foundations used to investigate this topic originate in systems theory, models of information systems and organizational theories of structure and complexity. Models of system development are reviewed as they play a key role in our notions of activities to perform during development. By drawing on existing estimating models from software engineering and other sources of practitioner literature, a model to predict development effort was synthesized. Three distinct constructs emerged: 1) system requirements, which drive effort; 2) personnel experience, which can mitigate effort; and 3) technology, which can also mitigate effort. System requirements was chosen to further define and operationalize, since they are the principal source of development complexity and hence system size. An Entity-Relationship and Event approach was taken to establish an early measure of System Requirements size. A theoretical framework of data and process complexity was developed which may be used to initially size a development project based on the information available at the requirements specification and design phases. It is argued that this new approach is more general than existing sizing techniques; namely, Function Points and Lines of Code. This new sizing approach is tested against 26 simple transaction based processing systems developed in FOCUS. An automated Code Analyser was developed to reverse engineer these systems back to their design measures. The personnel and technology variables were held constant for this initial test. For the thesis, only retrospective measurement occurred but it is expected that longitudinal measurement will eventually be possible. Two primary research contributions are seen emanating from this study. First is the development and application of theory to the problem of information system sizing. Second is a method for data collection and analysis which will help the software development industry move towards the goal of system development measurement and evaluation. This will improve planning for, and management of, information systems development.Business, Sauder School ofGraduat