# Sociogram: Worked Example, Matrix, and Interpretation

See a worked sociogram example for adults: read the prompt, matrix, reciprocal and one-way choices, explicit zero, missing response, and interpretation limits.

- Canonical URL: https://www.harmate.com/en/blog/sociogram-worked-example-matrix-and-interpretation
- Author: Harmate Team
- Published: 2026-10-08
- Updated: 2026-10-08T02:03:07.304039+00:00
- Language: en

## Content

# Sociogram: Worked Example, Matrix, and Interpretation

A sociogram is useful only when you know exactly which relationship was asked about, of whom, and at what moment. This worked example follows a fictional adult training group from a nomination question to its matrix and graph. You can reproduce it with paper and pencil while keeping nonresponses and the limits of the arrows visible.

## A sociogram represents one specific relational question

The Office québécois de la langue française defines a sociogram as a diagram of psychosocial relationships among members of a group. Depending on the question, arrows might represent a choice, a request for advice, past cooperation, or another stated relationship. It is not a complete map of a team’s social life: it represents a defined tie, within a defined population, at a particular time. [OQLF, “sociogramme”](https://vitrinelinguistique.oqlf.gouv.qc.ca/fiche-gdt/fiche/8462624/sociogramme)

In the 1930s, Jacob L. Moreno and Helen Hall Jennings developed sociometry to study relationships within groups; Moreno used the sociogram as a visual representation of ties. A diagram of points and arrows can still make a structure easier to examine than a list. It does not turn situated answers into a general truth about the people represented. [University of Michigan, Research Center for Group Dynamics](https://rcgd.isr.umich.edu/news-events/in-the-news/the-research-center-for-group-dynamics-the-origins-of-network-analysis/)

The matrix and drawing answer complementary needs. The matrix preserves who nominated whom and makes the entries checkable. The sociogram makes reciprocal or one-way ties easier to see. Start with the answers; do not let layout or node size invent a hierarchy.

## Before collecting or sharing these answers

Before running the exercise, ask what concrete decision the map is meant to inform and whether nominations are actually needed for it. Define who belongs to the voting group, who may see identifiable answers, how long the data will be kept, and what happens if someone does not wish to participate. Voluntary participation is an important safeguard, but it does not by itself establish which legal basis applies to processing: roles and obligations depend on context. If in doubt, have the proposed process reviewed by the appropriate people before collecting anything.

Do not run the exercise as designed if the result could be used to publicly label someone, discipline a person, or allocate opportunities on its own. A nomination diagram cannot justify those uses by itself.

The CNIL highlights principles including purpose limitation, data minimization, transparency, people’s rights, limited retention, and security. Do not collect information “just in case”: explain the purpose, recipients, planned retention period, and handling process, then restrict access. Concrete obligations depend on context; this example is not legal advice. [CNIL, the six main GDPR principles](https://cnil.fr/fr/comprendre-le-rgpd/les-six-grands-principes-du-rgpd)

Replacing names with A–F in a working copy reduces direct exposure, but **pseudonymization is not anonymization** if someone can recover the mapping or recognize the relationships. An arrow also describes a relationship with another person. Do not project identifiable answers or discuss an individual case in front of the group. Adapt safeguards and seek advice if the stakes go beyond the exercise.

## Worked example: choosing partners to prepare an activity

Six people, coded A to F, are attending a professional training workshop. The facilitator is preparing a 15-minute exercise on giving feedback. She asks one question: **“To prepare this exercise with two other people before the next session, whom in this group would you choose? You may name zero, one, or two people.”** The timeframe, reference group, and activity are explicit. The codes are fictional and do not represent real results.

Each person may make at most two positive nominations. There is no question asking whom to avoid or whom someone would not want to work with, so the absence of an arrow will never be drawn as opposition. The recorded answers are:

| Respondent | Stated positive nominations | Response status |
| --- | --- | --- |
| A | B, C | Complete response |
| B | A, C | Complete response |
| C | B, D | Complete response |
| D | C, F | Complete response |
| E | No one | Complete response: explicitly stated zero choices |
| F | — | No response received |

That distinction matters: E explicitly nominated no one for this question; for F, we do not know what the answer would have contained. A blank value must not be used to collapse these two states into one.

## Turn the answers into a matrix

In the matrix below, the row shows who is choosing and the column shows who is chosen. `1` means a direct nomination was stated; `0` means a person who completed the answer did not nominate that target; `NA` means no answer is available for that row; `—` marks the diagonal, because self-nominations were not requested. The A–F order is kept in every row and column.

| From \ To | A | B | C | D | E | F |
| --- | ---: | ---: | ---: | ---: | ---: | ---: |
| A | — | 1 | 1 | 0 | 0 | 0 |
| B | 1 | — | 1 | 0 | 0 | 0 |
| C | 0 | 1 | — | 1 | 0 | 0 |
| D | 0 | 0 | 1 | — | 0 | 1 |
| E | 0 | 0 | 0 | 0 | — | 0 |
| F | NA | NA | NA | NA | NA | — |

The equivalent edge list is: **A → B, A → C, B → A, B → C, C → B, C → D, D → C, D → F**. E has no outgoing edge because E explicitly stated zero nominations. F has no interpretable row because F’s answer is missing. In this example, F remains in the collection population and no answer has been received; this is not a declared withdrawal. Before collection, the protocol must define how refusal or withdrawal affects the map, including incoming nominations. This list is a transcription model to copy manually; it is not a download, a software screenshot, or a Harmate result.

## Read the arrows without overinterpreting people

The choices A → B and B → A form a reciprocal tie **for this preparation task**. B ↔ C and C ↔ D are reciprocal too. This indicates that both people named each other in response to this question; it does not prove friendship, competence, or successful past collaboration. The prompt did not ask anyone to assess performance.

A → C is an observed nonreciprocal choice: C answered and used both nominations for B and D. This does not establish rejection of A. By contrast, D → F is a one-way choice in the available answers. Since F did not respond, we do not know whether F would have named D, someone else, or no one. Drawing an incoming arrow to F is faithful to the data; adding a return arrow would invent information. Missing a response is not a refusal.

E explicitly nominated no one, while no response is available from F. These states are not interchangeable. Even though E received no nominations in this example, the diagram does not justify labeling E “isolated.” The question concerned a future activity and capped nominations; it did not measure overall social standing or relationships outside the workshop.

The observed incoming nominations from complete responses can be counted: A receives 1, B receives 2, C receives 3, D receives 1, E receives 0, and F receives 1. With F’s response missing, A to E can receive nominations from at most four respondents, while F can receive them from five. These unequal denominators rule out a naive comparison of counts. These are descriptive counts in this small data set, not popularity or worth scores. They should not be compared across groups without examining the prompt, response count, nomination limit, and who was eligible to vote.

With a two-choice cap, other preferences may not appear: **not being chosen is not rejection**. A study of middle-school peer nominations found that long-roster order and the eligible voting group influenced observed nominations. That finding is specific to the school context: this article does not measure that effect or extend it to adult workshops; it reinforces the need to document the protocol. [Poulin & Dishion, 2008](https://www.labopoulin.com/_files/ugd/74ebc4_ea707b02e2a9498b998f6086bece1221.pdf)

## Turn the table into a drawing

![Fictional six-person sociogram showing the matrix’s eight arrows: three reciprocal pairs and two one-way choices. F’s dashed outline marks a missing response; E has no observed tie.](https://api.harmate.com/media/blog/body/db674d2c-f09b-4abc-816c-adf6030e4c08.webp)

Drawing of the fictional example, not a Harmate output. Top row, left to right: A, B, E; bottom row: C, D, F. Paired arrows show A ↔ B, B ↔ C and C ↔ D; single arrows show A → C and D → F. F’s dashed outline means missing response, not refusal. E explicitly nominated no one and receives no available nomination. Positions do not represent a ranking.

For the graph, draw six same-size nodes labeled A to F, then eight directed edges exactly matching the edge list (five lines if each reciprocal pair is drawn as one double-headed arrow). Use a double arrow only for two reciprocal nominations—or two opposing single arrows, depending on your chosen convention. Keep D → F one-way; retain F even though its row is unknown, and add a clear legend for that response status. Keep E visible: a declared zero outgoing nominations is not a reason to remove the node.

Avoid scaling nodes by nominations received, using “leader / rejected” colors, or arranging people to imply a ranking. If you use colors, map them to a documented property rather than a judgment: for instance, gray for missing responses and a neutral outline for complete responses. Explain the convention in the legend. Place nodes manually to reduce crossing lines, then check every arrow against the matrix. A prettier drawing is not a more valid analysis.

After reviewing the map, a voluntary conversation or interview may help check what the answers mean and what has changed. This follow-up does not turn the map into a diagnosis; it lets you ask about context instead of assigning an intention to an arrow. To place this relational input in a group-composition decision, compare it with [a decision matrix for homogeneous or complementary groups](/en/blog/homogeneous-vs-complementary-groups-the-decision-matrix). If your goal is to prepare training groups around expressed needs, see [the guide to building training groups](/en/blog/how-to-build-training-groups-from-answers-you-can-actually-use).

## What a sociogram cannot decide for you

A sociogram is a partial view, useful for opening an inquiry or preparing a careful conversation. It does not automatically identify the best teams, the people who are influential in every context, or those who need help. It should not become a ranking handed to managers or the sole basis for allocating opportunities. If the map raises a question, phrase it as a hypothesis to check with care: “What support would help you prepare this activity?” rather than “Why does nobody choose you?”

Harmate works with open-ended questionnaires and the context in people’s answers to help prepare groups. The product does not generate nomination-based sociograms or measure friendship, isolation, or social status. To see what is actually offered, visit [open-ended questions in Harmate](/en/product/open-questions). An open answer can shed light on expressed needs; it does not by itself guarantee a valid interpretation.

## Frequently asked questions

### What is the difference between a sociometric matrix and a sociogram?

The matrix is the source table recording, for each respondent, whom they nominated under a defined rule. A sociogram is one possible visual representation of it. Keeping the matrix beside the drawing makes it easier to check arrows and the legend.

### Is someone who receives no nominations isolated?

Not on the basis of this result alone. The person may have been absent or less familiar to voters, may have been left off because of the nomination cap, or may relate to others in a way the prompt did not ask about. “No incoming choice observed for this question” describes the result more accurately than a personal label.

### What should I do if someone did not respond?

If the person remains in the collection population and has not declared a withdrawal, keep them in the map, mark the row as unknown, and do not code nonresponse as zero. If participation is voluntary, do not force an answer just to complete the diagram. Document the number and status of available responses when presenting results.

### Can I use a sociogram to form groups?

It can provide a limited relational input if the question is relevant, understood, and appropriate to the purpose. It is not enough to decide group composition on its own: learning needs, work constraints, stated preferences, and possible consequences also need review. Explain the decision and check that it remains fair to the people involved.