SPA Conference session: Experiments in Agile Estimation Techniques

One-line description:When to hold or fold in "planning poker"
 
Session format: Workshop (75 mins) [read about the different session types]
 
Abstract:In this workshop we will run an experiment to compare two different estimation techniques: "planning poker", and individual estimation. We will analyse the data gathered from the experiment and compare the results with our own estimation experience.

Objectives

* Introduce and provide experience with the group estimation technique planning poker.

* Provide experimental evidence to support a quantitative and qualitative comparison of planning poker to individual estimating, including how the two estimation techniques are affected by well known manipulations like anchoring and irrelevant information.

* Pass on our experience in agile estimation on multiple projects.

* Provide feedback on participant’s estimation bias.
 
Audience background:Any, especially those interested/involved in estimation.
 
Benefits of participating:In attending this workshop, participants will:

* learn about the group estimation technique "planning poker", and gain some practice in using the technique.

* learn about their own individual estimation bias.

* hear about real project estimation experiences

 
Materials provided:Stories to be estimated
 
Process:After a short introduction of the planning poker technique, the attendees will be divided into groups (each group should have about 6-8 people). We will prepare two stacks of “stories” to be estimated in two rounds. The attendees will estimate one stack individually, and one stack in a group using planning poker. Half the groups will estimate individually first, and the other half using planning poker first.

The stories will exist in multiple versions, where the different versions have been manipulated by including information that is well known to affect the estimates. Manipulations include anchors and irrelevant information.

After all the exercise rounds are complete, we will work on correlating the data gathered (some data entry into a pre-prepared analysis tool) whilst we talk about our own estimation experiences from Agile projects using planning poker in various settings.

Once we’ve entered the data, we will present the quantitative results:

* How many stories were estimated (how fast is this technique)?

* How do the mean and standard deviation compare? Did we observe any group polarization effects?

* How did the different manipulations affect the estimates? Did the manipulations affect the individual estimates more than the planning poker estimates?

* What was each individual’s personal estimation bias?

And we will lead a round-up/retrospective on the qualitative results:

* How did the groups feel about the estimation techniques?

* How involved did the participants feel?

* Did the participants agree with the group estimate?
 
Detailed timetable:
 
Outputs:The data produced from this session’s experiment will be summarized on a poster.
 
History:Proposed to XP Day London 2006
 
Presenters
1. Mike Hill
Mandu Limited
2. Nils C. Haugen
Objectnet
3.