Don’t mix roulette with poker – a recipe for better estimations

During a planning poker estimation we hide cards until everybody makes his mind and then we reveal them all at once. We do that to make the estimation more objective, unbiased.

One obvious bias that could skew our estimation is influence of more experienced, or just more vocal, team members. The pressure resulting from their self-confidence can impact our estimate, make it gravitate towards their opinion.

However, there is also another strong, but not so obvious bias: the Anchoring Effect.

First theorized by Amos Tversky and Daniel Kahneman, the Anchoring Effect is a psychological heuristic that impacts, among others, the way people estimate. When presented with an initial reference point (the “anchor”), people fixate on this reference, and estimate by making adjustments to the anchor, instead of coming up with their own, “objective” estimate. These adjustments are usually insufficient, resulting in a biased estimate, skewed in the direction of the anchor.

In effect, the following three questions: “How big is this user story?”, “Is this story bigger than 2 Story Points?” and “Is this story smaller than 13 Story Points?” will usually result in three different estimates of the same story.

What’s interesting, anchoring works even if the anchor is strikingly unrealistic. In one experiment, one group of participants was asked if Mahatma Gandhi died before or after age 9, the second group – if he died before or after age 140. Both anchors are blatantly incorrect, yet the Anchoring Effect they induced still resulted in 27 years of difference between the two groups’ average answers.

Even stranger – anchoring can still happen even if the anchor is unrelated to the subject being estimated. In an another experiment participants observed a roulette wheel, fixed by the researchers to stop on either 10 or 65. Participants were then asked to guess the percentage of the United Nations that were African nations. Average answer of participants observing the wheel stopping on 10 was 25%, while average answer of participants observing the wheel stopping on 65 was 45% – almost double as much!

The Anchoring Effect is dangerous, because it works on a subliminal level and it’s not only difficult to notice, but also practically impossible to resist, even when you’re aware of it.

What’s more, the Anchoring Effect isn’t limited to numerical values. It occurs also during brainstorming-like events (e.g. retrospectives), voting and in many other situations.

The only way to objectivize such events is to use external, pre-planned counter-measures, such as a blind voting (and to remove all roulette wheels from your workspace). If you’re responsible for running a meeting susceptible to anchoring, you must consciously design the form of such a meeting to eliminate opportunities for bias – otherwise you’ll get skewed results.

Can you think of any other situations susceptible to Anchoring Effect? Do you implement any counter-measures to un-bias them? Please share below in the comments!


What do you think?

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s