Improving the accuracy of project schedules

Abstract

How to avoid project failures driven by overoptimistic schedules? Managers often attempt to mitigate the duration underestimation and improve the accuracy of project schedules by providing their planners with excessively detailed project specifications. While this traditional approach may be intuitive, solely providing more detailed information has proven to have a limited effect on eliminating behavioral biases. We experimentally test the effectiveness of providing detailed specification and compare it to an alternative intervention of providing historical information about the average duration of similar projects in the past. We find that both interventions mitigate the underestimation bias. However, since providing detailed project specification results in high variance of estimation errors due to sizable over- and underestimates, only the provision of historical information leads to more accurate project duration estimates. We also test whether it is more effective to anchor planners by providing historical information simultaneously with the project specification or to provide the historical information only after beliefs regarding the project duration are formed, in which case planners can regress their initial estimates towards the historical average. We find that the timing of disclosing information does not play a role as the estimation bias is mitigated and the accuracy is improved in both conditions. Finally, we observe that the subjective confidence in the accuracy of duration estimates does not vary across the interventions, suggesting that the confidence is neither a function of the amount nor the detail of available information.