# Forming Workshop Groups: Free Choice, Random Assignment, or Criteria?

A fictional example for choosing how to form workshop groups, without assuming one method is always better.

- Canonical URL: https://www.harmate.com/en/blog/the-end-of-random-group-formation-social-psychology-and-algorithms
- Author: Harmate Team
- Published: 2025-12-08
- Updated: 2026-10-18T00:43:30.477378+00:00
- Language: en

## Content

# Forming Workshop Groups: Free Choice, Random Assignment, or Criteria?

“Please form groups of three.” The instruction does not yet say how people should be assigned. The right process depends on what the activity needs to achieve, which constraints matter, and what information is actually available. Free choice, random assignment, and criteria-based grouping serve different purposes; none is best in every situation.

## Three methods, three use cases

| Method | Consider it when… | Check… |
| --- | --- | --- |
| **Free choice** | Autonomy, meeting people, or choosing partners is part of the activity. | Choice may not account for constraints or the composition the task requires. Check whether everyone can join a group and whether the activity remains workable. |
| **Random assignment** | Places are interchangeable and a simple, announced process is enough. | Randomness does not guarantee a particular mix of skills or meet specific constraints. Check group sizes and exceptions before using the result. |
| **Explicit criteria** | The exercise requires a defined composition, such as including different functions in each group. | Criteria should fit the task, be understandable, and rely on suitable information. A responsible person should review the proposal and be able to change it. |

These are procedural choices, not judgments about what participants are like or what they can contribute. They help you choose a process for the activity and then review its effects.

## A fictional example to adapt

Imagine a practice workshop with three equivalent stations, each with room for four people. Participants can choose a station while places are available. If more people request one station than it can hold, the facilitator could use a draw among those participants, explaining the rule before applying it. An availability constraint still comes first: no one should be assigned to a station they cannot attend.

The goal here is not to create a particular group profile. Free choice supports autonomy when places are available; a draw can resolve equivalent requests when capacity is exceeded; availability is a constraint to check whatever method is used. The draw does not guarantee any distribution beyond the announced procedure. If capacities or availability change, the facilitator reviews assignments before sharing the groups.

To compare group compositions—homogeneous, heterogeneous, or complementary—see the [decision matrix](/en/blog/homogeneous-vs-complementary-groups-the-decision-matrix). For the workflow from questionnaire answers to forming training groups, see [how to build groups from usable answers](/en/blog/how-to-build-training-groups-from-answers-you-can-actually-use).

## A proposal still needs review

Criteria-based grouping makes the chosen criteria visible; it does not prove that they fit the task or that the result will work for everyone. Before sharing groups, review the input, exceptions, and proposed composition against the activity’s goal. Explain which criteria were used and keep the option to adjust the proposal.

[Harmate’s automatic grouping](/en/product/automatic-grouping) can propose groups from criteria you choose. You can compare options, check exceptions, and review or adjust the composition before sharing it. This supports organizing a distribution; it does not decide which grouping method is pedagogically appropriate.

## Start with the activity

Write down what the groups need to do, then note the constraints that genuinely affect the assignment. If several methods can serve the goal, choose the one that is easiest to explain to participants. If you use criteria, say why they matter and who will check exceptions.

The useful question is not “randomness or an algorithm?” but “which process serves this activity, and how will we review its result?”