Predicting information technology project escalation: A neural network approach

Information technology (IT) projects can often spiral out of control to become runaway systems that far exceed their original budget and scheduled due date. The majority of these escalated projects are eventually abandoned or significantly redirected without delivering intended business value. Because of the strategic importance of IT projects and the large amount of resources involved in the development of IT projects, the ability to predict project escalation tendency is critical. In this study, we compare neural network and logistic regression models in building an effective early warning system to predict project escalation. Variable selection approaches are employed to identify the most important predictor variables from those derived from the project management literature and four behavioral theories. Results show that neural networks are able to predict considerably better than the traditional statistical approach – logistic regression. In addition, project management factors are found to be more critical than behavioral factors in accounting for the success of an IT project.