Critical success factors for outsourced software development projects from a vendor's perspective: A structural equation modelling analysis of traditional plan-based and agile methodologies

摘要

There are many factors proposed as to why software projects fail, one of them is the inappropriate choice of a project management methodology. Although there is an increased range of available management choices, project managers do not frequently consider their alternatives. They tend to narrowly tailor project categorisation systems and use categorisation criteria that are not logically linked with objectives. To address this, this study develops and tests an integrative contingency fit model for contrasting perspectives of traditional plan-based and agile methodologies specifically for outsourced software development projects. In addition, it takes a vendor‘s perspective, rather than the client perspective that is mostly used. Overall, the research seeks to answer these questions: (RQ1) what are the Critical Success Factors (CSFs) for outsourced software development projects from a vendor‟s perspective? (RQ2) What are the differences in these CSFs for traditional plan-driven and agile methodologies towards project success from a vendor‟s perspective?The IT literature reveals two major distinct categories of methodologies: traditional plan-based and agile. Previous research has identified CSFs with respect to project success with mixed findings. The recent increase in popularity of methodologies has shifted the debate, interest and controversy to CSFs that are the factors which are most important to make a methodology successful. While there is an increasing diversity of project types, project contexts and methodologies, the frameworks or theories connecting these are limited. To date software development projects studies have addressed generally one methodology per study and perceived candidate CSFs as a form of reasons of success amidst a wide range of project success criteria. Although contingency theory has been previously argued for outsourced software development projects, empirical models have frequently not fully incorporated contingency as fit or fit as moderation (i.e. traditional vs. agile). This study sought to fill this research gap.Cross-sectional data from 984 senior vendor project managers and team leaders was collected by a global web-based survey. Structural Equation Modelling (SEM) (a multivariate statistical technique, in which parameters are estimated by minimizing the discrepancy between the model-implied covariance matrix and the observed covariance matrix) was used for data analysis. SEM results provide support for several contingency hypotheses theorizing relationships between candidate CSFs and project success. Project management methodology was found to moderate the effects of various CSFs on project success, and in different ways for various success measures. Similarly, the results show the level of project uncertainty moderates the impact of various CSFs on project success, and in different ways for various success measures. Together these findings provide empirical support for contingency as fit and more fully incorporate fit as moderation.The study contributes towards understanding the differences between traditional plan-based and agile project management based on the perceptions of vendor respondents with regard to their client organizations, and also to understanding what are the most significant antecedents of success (the CSFs) in different project contexts. The study also examines the indirect and interaction effects, and the findings contribute towards understanding of the contingency perspective as a framework to be used by project managers and organizations. Practical implications of these results suggest that project managers should tailor project management methodologies according to various project types, which is likely to improve current project success rates.