When is your current project going to be finished?
Are you being optimistic about the completion time of your current project although you know that majority of projects in the past took you longer to finish than expected? If yes, don’t worry, you are not alone. Many of us keep on repeating the same mistake. We make plans only to find out later that they were too ambitious, yet again.
My research strives to shed light on the process of performance time estimation. I investigate the conditions for accurate estimates, as well as potential sources of bias. For example, I am testing the relative power of task experience against numerical anchor in repetitive estimating of time needed for the same task. I also look at the influence of historical data on prediction process and willingness to obtain such data before the estimation. Finally, I work with different incentive structures to test the ability to move away from poor estimates.
The story of Sydney Opera House
Why projects fail?
Traditionally, it has been estimated, that up to 50-90 percent of business projects are not delivered within pre-defined time, cost or quality, or are not finished at all. The enormous failure rate led to establishment of formal processes for project management in 1960s. The methodology has been continuously developed since. However, after 50 years of worldwide project management education and training, most project management offices in companies barely reach the 50 percent threshold of successful projects.
The project failure is usually driven by multiple causes, as the complex nature of projects brings various sources of interference with the plan. However, one phenomenon seems to be reoccurring – unrealistic schedules stemming from inaccurate estimates of future project activities duration. Over-optimistically planned projects then run late, which triggers budget extensions, hinders customer satisfaction and binds company resources.
My research is focused on the performance time prediction process in the context of project management. The goal of my research is to investigate the causes of systematic bias in predictions and determine ways to mitigate it.
My first research project focuses on the role of numerical values (anchors) on duration estimates. In the context of project management, the anchor can appear, for example, in the form of:
- Initial wild guess (How long? Three months, maybe?)
- Suggestion (Do you think two weeks are enough for you to get it done?)
- Customer expectations (We would really like to introduce the product to the market before summer season.)
- Tentative deadline (The management expressed the their beliefs that the project could be finished by the end of the year).
The customers or management certainly want to get their projects done as soon as possible. As a result, their suggestions or expectations can often be driven by wishful thinking and thus, anchors are prone to be rather optimistic. If those anchors actually influence duration estimates produced by project team, the projects are doomed to fail right from the start.
Moreover, the effect of anchor can persist over time and influence subsequent estimates of the same or similar activities. This may be the case for instance during the planning phase of the next project and especially plausible in the absence of feedback regarding the actual duration of the first activity. Even when there is no arbitrary anchor before the first estimate available, the first estimate itself can serve as an anchor for future estimates and cause systematic bias.
The experiment consisted of three rounds of estimating the completion time of a simple number comparison task and then performing it.
A total of 93 participants were asked to estimate how long it will take them to provide 400 correct evaluations. They were financially incentivized both for performance speed and estimation accuracy. They were randomly divided into three treatments:
- Control treatment
- Low Anchor treatment
- High Anchor treatment
Before first round only, participants in low (high) anchor treatment were asked to consider, whether it will take them less or more than 3 (20) minutes to complete the first round.
Results (in seconds)
We found extremely statistically significant differences in estimates across treatments for all three rounds. No difference in completion times were found.
The planning fallacy
Misestimation of project duration can be attributed to an optimism bias and is not necessarily a fallacy per se. The planning fallacy is defined as an underestimation of time needed to complete a project only when planners know of same or similar projects from the past that run late and do not update their beliefs towards future projects and continue to plan them over-optimistically. To look for potential cures, I plan to investigate:
1) Inside vs. Outside view
The relative power of task/project description (inside view) on its duration estimate vs. the reference class information of past task/projects completion times (outside view). I will not only investigate the influence of reference class information on final estimates but also willingness of subjects to obtain such data before estimation.
2) Accuracy incentive structures
I also plan to test different incentive structures on project duration estimates. For example, the final estimate should be different under hard deadline incentive structure when being late is not acceptable vs. pure accuracy incentive structure when the under and overestimation have the same consequences. However, in order to be aligned with the incentive structure, it is necessary for estimator to move his estimate away from the most available (mode) estimate.