The other day I needed to divide my audience into smaller groups several times within the same meeting.  The meeting leader wanted the groups to be different each time.  For the initial split, we simply counted off by the number of groups we wanted.  However, given that people were sitting in roughly the same places throughout the meeting, this technique was only going to work once.

I started using other simple criteria to sort the participants into groups (birth day, birth month, etc). Figuring out the sorting rules is pretty simple.  However, it’s one of those tasks that you really shouldn’t spend much time on since it can be easily codified. So, that’s what I did.

The following table provides different ways to randomly divide an audience into subgroups. Each row represents a different sorting criteria.  Each column represents the number of groups that you are trying to create.  The individual cells show the sorting rules. 

In some cases, you’ll see that the criteria don’t appear to be evenly divided. This is done to accommodate differences in their actual distributions in real life. For example, there are a lot fewer people whose first name begins with “X” than “A”.  However, those types of adjustments weren’t always necessary (surprisingly birth month, birth day are actually pretty evenly distributed).  These rules are all based on US data.  Some of the criteria, especially those associated with names, could be quite different in other countries.

No model is going to be perfect.  You still may have to do some adjusting on the back end. However, these sorting rules should give you a good head start. And remember, to fully get the benefit of the statistical distributions, your audience needs to be large enough.  The more groups you are creating, the larger that the audience needs to be.   Dividing sixteen people into two groups using the rules for months will probably work well.  However, dividing those same sixteen people into six groups using the rules for months might not.  You need a larger group to offset the random variation that exists.

You may need to click on the image to see the full table. For a PDF version of this table, click here.

———————————–

Brad Kolar is an executive consultant, speaker, and author with Avail Advisors LLC.  He can be reached at brad.kolar@avail.advisors.com.