- mean – 6.8 Months
- Median – 6 Months
- standard deviation – 3.7
control – no explicit anchoring
- mean – 8.3
- Median – 7
- standard deviation – 4.4
Condition for ’20 months’
- Mean – 17.4
- Median – 16
- standard deviation – 5.6
These results are consistent with my own experience. It is important to note that although they were supposedly estimating exactly the same software requirements, it is very possible that the 2 month’ group would have produced significantly different products than the 20 month group. This is some food for thought. What influences are creating anchors when you and your team put together estimates? There are many, and some of them are likely to be arbitrary. These may come from stakeholders, sponsors, project managers, team members, or even leaders. This is my favorite example. It leads to a poor estimate and creates an anchor for future estimates. What do you think?
Glen Alleman recently referred me to Jorge Aranda’s paper, “How Expectations Can Cause Project Estimates to Go awry” among other material about software estimation. It was well worth the time I spent reading Anchoring and Adjustment In Software Estimation. Andrew Stawarz via Flickr. To cut to the chase, subjects were asked to estimate software tasks in a controlled fashion, in three groups using different “anchoring” methods. The only thing that separated the groups was the expectation statement made by the manager prior to estimation. Group 1 (control – No explicit anchor given).?I would like to estimate this project myself, but I confess that I don’t have any experience in estimating. We will wait to see your calculations before we can give you an estimate. Group 2 (‘2 months’ condition).?I confess that I don’t have any experience with software projects but I think this will take approximately 2 months to complete. We’ll wait to see your calculations to get a better idea. Group 3 (’20 month’ condition)?I confess that I don’t have any experience with software projects but I think this will take around 20 months to complete. We’ll wait to see your calculations for a better estimate. To see the whole paper (and to determine whether it is relevant to you domain), I’d like to summarize some of my most striking findings. These were the results for all participants. There are also other slices of data that can be found in the paper, including those from experienced participants and those based on the estimation method used. Condition ‘2 months’