What is Anchoring Bias?

As any project manager will tell you, getting good estimates for tasks is an incredibly challenging proposition.  This part of project planning is critical for both traditional waterfall style projects and also when using Agile methods.  One of the biggest barriers to good estimates is a cognitive bias called “Anchoring”.  If you aren’t familiar with this term, I am certain that you are familiar with how it works.  Usually a project estimates session goes something like this:

Person A (1st) :  10 days
Person B:  12 days
Person C:  8 days
Person D:  11 days

What happens is that whoever gives the 1st answer ends up ‘anchoring’ the group and the rest of the answers will cluster around it – even if it’s completely wrong.  In the absence of clear information, people have a natural tendency to anchor to any available information.  Also, in a group situation such as a project meeting, this effect can be more amplified.  Consider if a senior technical person started with an answer of 10 days.  Would a junior staff member say 5 or 20?  In most cases the answer is no.

"The anchoring and adjustment heuristic was first theorized by Amos Tversky and Daniel Kahneman. In one of their first studies, the two showed that when asked to guess the percentage of African nations which are members of the United Nations, people who were first asked "Was it more or less than 10%?" guessed lower values (25% on average) than those who had been asked if it was more or less than 65% (45% on average)." (source: Wikipedia )

How do you fix it?

There is a well known technique for eliminating anchoring bias from estimates sessions called “Planning Poker”.  We use it for basically all of our estimates at Decision Resources.  We’ve even used it for things like giving feedback on our new web site design!  It’s very simple and here’s how it works:

1. All estimators are given some cards.  You can use special cards like we have or just use normal playing cards.  The cards we use are 0,1,2,3,5,8 - a fibonacci sequence.
2. The item to be estimated is discussed.  If you are using Scrum, then the product owner will probably describe a story.
3. Each person secretly selects a card and then all estimators simultaneously turn their card over.
4. If all people agree, then that number is the estimate.
5. If someone is significantly different, then they get a ‘soapbox’ where they explain why they gave the number that they gave.
6. The estimate process is then repeated until consensus is reached.

Why does it improve estimates?

It can seem incredible to some people that such a simple process can improve estimates so dramatically (as well as make your estimates sessions more fun!).  Here’s why I think it works:

People are forced to take a stand
Often in project meetings, only the loudest or most influential voice gets heard.  The quiet guy up the back (who may be in a better position to know something) has to take an independent position.

Consensus is real
If everyone independently comes up with the same number, then the consensus is much more solid.

Hidden information is discovered
This is the real power of planning poker.  The divergent opinion can reveal lots of things.  If the estimate is low, then maybe they know a simpler way of doing something.  Maybe they don’t understand the complexity involved.  If the estimate is high, then maybe they know of a problem that the rest of the group hasn’t yet seen.  This information discovery process is extremely valuable and is very difficult to achieve normally.


Of course, no estimate process is perfect - this included.  The whole point of this exercise is not to try and get super precise estimates, but rather get a feel for how difficult particular jobs are and discover any information within the group that is hidden. 

If you are having trouble with project estimates, I definitely recommend you give it a go.  If nothing else, it will make your estimates sessions more interesting.