When to use it Use it to explore an issue with a distributed group of people. Use it to move a group of people towards consensus. How to use it Define the problem Identify the problem that you want to work on, writing it down in a clear way that is easy to understand. This can be in various forms, from a questionnaire to a broad and open question. You can work on one problem and you can work on several problems at once. The constraint is usually the bandwidth and expertize of the people in the Delphi group. Give everyone the problem Recruit people to the Delphi group. This includes anyone who has been selected to contribute thinking on this project. There is seldom a meeting needed for Delphi work, making it ideal for virtual teams. Delphi thinking can be done with a small group and it can be done with hundreds of people. Around 20 people is a fairly common size. Send the problem or problems to everyone who is in the group and ask them respond. You will have to handle a lot of feedback, so asking for short bullet-points will make things much easier to deal with than rambling text. Collate the responses Take the responses that people send back to you and collate these into a single anonymous list or sets of lists. Make this as easy as possible for the people to read when you send it back out again, but be aware of causing inappropriate bias. For example you may group responses into appropriate headings, but with the caution that this might presuppose particular thinking. On the other hand, if you are seeking creative ideas you may deliberately mix up the answers. Give everyone the collation Send the collation back out to everyone with the request to score each item on a given scale (typically 1 to 5). You may also allow them to add further items as appropriate. Remember to include the original problem at the top of the page, along with instructions on what to do. You can also make responding easier by putting the items in a table with space for the score. Repeat as necessary The process may now be repeated as many times as is deemed appropriate. If you are seeking consensus and there was a wide range of responses, then this may require several iterations. In particular at least a second round to see how others have scored can be very useful. In analysing the scores, one method used in Delphi analysis for smaller groups and especially when percentage scores are used is to use the formula: (lowest score + highest score + 4 x average score) / 6. This gives more weight to the average whilst also allowing some influence from outliers. Example Given a problem around valve leakage, a community of engineers decide to use a Delphi group to explore ideas. A part of the table they come up looks like this. This could be done in a simpler table, but hey, this is engineers! | Comment | Score (1 is low, 5 is high) [previous score in square brackets] (number is how many people scored in this column) | | 1 | 2 | 3 | 4 | 5 | Test pressure either side of valve.
+ Not easy to do.
+ Look at lateral expansion. | | 1[3] | 5[3] | 1[2] | 2[1] | | Check for axial alignment | 1[2] | 4[3] | 2[1] | 1[2] | | | Weather effects? | 1[1] | 5[5] | 2[2] | | | How it works The power of the Delphi method in gaining consensus is first that comments are anonymous, which helps people avoid defending their original position. It also allows them to be gradually swayed by the majority but without undue direct social pressure. Being able to see what others have said or scored and then rethinking your own position can act as a very effective influence. See also Voting Original Location: http://creatingminds.org/tools/delphi.htm |