| The software maintenance process is the last phase in the software life cycle. Studies have shown that, on average, 70% of MIS budgets are devoted to the maintenance process. There is considerable interest in changing software development and maintenance practices to "free" budget dollars for "development" projects and look for ways of making current practices less expensive. In a number of organizations, MIS management has turned to software engineering tools which are designed to support software maintenance as a potential solution to this problem.; In this dissertation a model based on the Technology Acceptance Model of Davis (1989) and the Task-Technology Fit model of Goodhue (1988, 1992), was developed to explain the factors which lead to the use of this class of software tool. The nature of the fit between software tool functionality and maintenance task demands is examined specifically. A model of the software maintenance process was developed based on the software debugging literature and program understanding literature. A model of software maintenance tool functionality was developed based on the Functional CASE Technology Model of Henderson & Cooprider (1990). New instruments to measure maintenance task characteristics and the functionality of software maintenance support tools were developed.; Data were collected from 36 programmers and their managers on 74 maintenance projects. This sample was drawn from the software development and maintenance organizations of three large firms. The data were analyzed using factor analysis, regression modeling and MANOVA.; The principal finding of the dissertation is that Task-Technology fit is a strong and significant predictor of software maintenance tool use. In addition, it was found that higher task complexity and higher levels of experience with software maintenance tools are associated with higher levels of tool use. Experience with the software being maintained did not result in lower levels of use. Finally, no difference can be demonstrated between debugging and enhancement projects on tool use or task activities.; These results increase our understanding of how programmers actually perform software maintenance. Further, these findings lead to recommendations for the improvement of maintenance support and maintenance software design. |
