Assessing Uncertainty of Software Development Effort Estimates: The Learning from Outcome Feedback

• Authors: Tanja Gruschke;Magne Jorgensen
• Venue: METRICS '05 Proceedings of the 11th IEEE International Software Metrics Symposium
• Year: 2005

To enable properly sized software projects budgets and plans it is important to be able to assess the uncertainty of the estimates of most likely effort required to complete the projects. Previous studies show that software professionals tend to be too optimistic about the uncertainty of their effort estimates. This paper reports the results from a preliminary study on the role of outcome feedback in the learning process on effort estimation uncertainty assessment. Software developers were given repeated and immediate outcome feedback, i.e. feedback about the discrepancy between the estimated most likely effort and the actual effort, for the purpose of investigating how much, and how, they improve (learn). We found that a necessary condition for improvement of uncertainty assessments of effort estimates may be the use of explicitly formulated uncertainty assessment strategies. By contrast, intuition-based uncertainty assessment strategies may lead to no or little learning.