Exploring Cost Estimation Inaccuracy - Why do practitioners still fail to predict the actuals?
Licentiate thesis, 2008
The purpose of this research is to further explore why practitioners’ ability to predict actual results is still not satisfactory. This problem was investigated from two perspectives. First, the practitioners view on causes of inaccurate cost estimates was explored in a quantitative multiple case study. Second, in order to understand why neither practitioners nor the researchers have been able to solve the problem of inaccurate estimates, a review was performed of the cost and effort estimation research. The results of the review were analyzed using Argyris’ organizational learning theory.
The empirical studies revealed a set of underlying causes of inaccurate cost estimates. It is hard for the practitioners to capitalize on the lessons learned due to the lack of estimation standards development and record-keeping of estimates and actual results. Also, due to the lack of results record keeping, the estimates (and thereby the estimation errors) include an outcome error.
The uncertainty of the requirements is expected not to be mirrored by an equally uncertain estimate and the disturbances in other projects are propagated through the common pool of resources. Finally, in the minds of practitioners; the process of estimation is intertwined with the negotiation process.
This research has also pointed at the need to apply an organizational learning perspective on cost and effort estimation indicating two main problems. First, it is probable that the practitioners and organizations failing to correctly estimate the actual results of their projects are likely to fail both in identifying the underlying causes of the estimation inaccuracy, and in correcting those causes. Also, the defensive reasoning of practitioners in such organizations makes it likely that they would, consciously or not, provide the researchers with invalid data. Second, the data collection methods frequently used in the studies researching the causes of inaccurate cost estimates may not allow causes previously unknown to the researcher to surface. Lastly, the already known issues are possibly not interpreted to the extent that would allow understanding of the underlying issues that cause the frustration of practitioners expressed in the studies.
Software Development Cost Estimation
EA-salen, EDIT-huset, Rännvägen 6, Chalmers
Opponent: Prof. Magne Jørgensen, Simula Research Laboratory, Norway