Tanya McGill

User development of applications provides end users with an alternative to the traditional process of systems development by allowing them to solve job related problems by developing their own software applications. User developed applications (UDAs) support decision making and organisational processes in the majority of organisations, and the ability to develop small applications forms part of the job requirements for many positions. Despite its pervasiveness, there are many risks associated with user development of applications. These risks result primarily from decreases in application quality that arise when end users have had little training and do not follow system development methodologies. The primary aim of the research described in this thesis is to gain a better understanding of UDA success. In particular, the thesis considers the role of system quality in UDA success and the ability of end user developers to judge whether the applications they develop will have a positive impact on their performance of tasks. The research also investigates factors that might impact upon this ability. The research objectives were addressed through two empirical studies. Two possible models of UDA success provided the starting point for Study 1. The first model is DeLone and McLean's (1992) model of IS success, and the second model is a version of this model that was modified to address concerns about the DeLone and McLean model and to reflect current research about UDA success. The models were tested using data from a field study involving business people participating in a business policy simulation, where they developed spreadsheet applications to assist in decision making. Structural equation modelling was used to test the models. Neither of the models was well supported by the data. However, the analysis provided strong support for relationships between perceived system quality and user satisfaction, information quality and user satisfaction, user satisfaction and intended use, and user satisfaction and individual impact. It is notable that the model paths that were supported in Study 1 were primarily those that reflect user perceptions rather than objective measures. This study highlighted that user perceptions of information systems success play a significant role in the UDA domain. The results did, however, suggest that there might be a direct relationship between system quality and individual impact. Study 2 was a laboratory experiment and the participants were end users from a range of organisations. A revised research model was developed based on the findings of Study 1, and structural equation modelling was again used to test the model. The model paths that were supported suggest that for small to moderate applications, increases in spreadsheet development knowledge lead to increases in system quality and consequently the development of better quality spreadsheets. They also suggest that for these kinds of applications, end users have realistic perceptions of system quality and hence that user satisfaction may be an appropriate measure of UDA success. The results of Study 2 also provided insight into the role of user involvement in end user development, clarifying the process by which benefits are obtained. The study also provided insight into the importance of spreadsheet development knowledge for successful use (as well as development) of a spreadsheet application. The results described in this thesis have practical implications for the management of user development of applications. They highlight the need either to increase end user levels of development knowledge via training so that end users can cope with applications of greater complexity, or to provide other forms of support for development. The role of organisational standards and guidelines is also be considered in the thesis and it is suggested that there is a particular need for guidelines on what kinds of applications are suitable for end user development.