Intervention grouping from assessment results

Use one assessment table to produce proposed groups that you can inspect before changing any pupil’s support. This works for teachers and intervention coordinators who need to match targeted teaching to named skills, rather than group pupils by a single total score.

The result is a short grouping note: who is proposed for each group, the evidence behind it, the skill to teach, and the point at which you will review the placement. You remain responsible for approving groups and checking that the assessment supports the decision.

1. Prepare a clean assessment table

Create one row per pupil. Use initials or approved pupil identifiers if you are putting information into an external tool. Follow your school’s policy for pupil data, and remove names, dates of birth, attendance notes and other details that are not needed for grouping.

Add columns for:

• Pupil ID
• Assessment date
• Skill 1: inference
• Skill 2: vocabulary in context
• Skill 3: retrieval
• Skill 4: sentence punctuation
• Total score, if used locally
• Teacher notes, limited to relevant learning evidence

Record each skill in a consistent form. Percentages, marks out of a fixed total, or a simple secure/developing/not yet secure judgement can work. Do not mix formats in the same column without explaining them.

For example, a reading assessment might show this:

Pupil ID	Inference, /6	Vocabulary, /4	Retrieval, /5	Note
A01	2	4	4	Gives literal answers
A02	1	2	4	Misses implied meaning
A03	5	1	5	Needs word meaning support
A04	2	4	2	Rushes evidence selection

Keep the source assessment beside the table. You will need it to check whether an apparently weak skill is actually based on too few questions, an absent pupil, or a misread item.

2. Set the decision rules before asking for groups

Write a short rule for what counts as a likely intervention need. This prevents the assistant from inventing a threshold after seeing the data.

For instance:

• Propose a group where a pupil scores below 50% on a skill with at least four assessed items.
• Flag, rather than group, any pupil whose score comes from fewer than four items.
• Keep groups between three and six pupils where possible.
• Prioritise the skill with the clearest repeated weakness.
• Use teacher notes only as supporting evidence, not as the sole reason for placement.

Choose a review point that is linked to the teaching cycle, such as after six taught sessions and a short re-check of the same skill. Do not promise that every pupil will be secure by that date. The review is a decision point, not a prediction.

If your available features, file handling or usage conditions vary, check the xAI documentation overview before uploading assessment material.

3. Ask for proposed groups in a fixed format

Paste the rules first, then the table. Ask for proposals, not final placements. The following prompt keeps the task narrow and makes the reasoning visible:

You are helping me prepare proposed intervention groups from an assessment.

Use only the assessment evidence and teacher notes below. Do not make final placement decisions. Do not infer needs that are not shown in the table.

Decision rules:
- Propose a group for pupils below 50% on a skill assessed by at least four items.
- Flag insufficient evidence where a skill has fewer than four assessed items.
- Group pupils by the main skill to teach, not total score.
- Aim for groups of three to six where possible.
- Teacher notes support, but do not replace, assessment evidence.
- Review after six sessions using a short assessment of the target skill.

For each proposed group, provide:
1. group name based on the skill
2. proposed pupil IDs
3. evidence for each pupil, including score and relevant note
4. the one skill to address
5. a short teaching focus
6. review point
7. pupils who need teacher review rather than a group, and why

Assessment table:
[paste table here]

Ask for one main target per group. A group called Reading support is too broad to plan from. Inference: selecting evidence for implied meaning tells you what the adult should teach.

4. Check every proposed placement against the source

Read the output with the original assessment open. Start with the membership list, because a plausible group title can hide a poor placement.

Use this check before approving anything:

If you see this	Check	What to do
A pupil appears in two groups	Whether one weakness is clearly more urgent	Choose one current target, or mark the second skill for later review
A pupil is grouped from one weak item	Number and quality of items assessing that skill	Remove from the group and gather more evidence
Notes contradict the score	The marked responses and assessment conditions	Reassess, moderate, or retain for teacher review
A group has seven or more pupils	Whether their error patterns are genuinely alike	Split only if each smaller group has a distinct teaching focus
The proposed skill is broad	The exact question types pupils missed	Rewrite it as a teachable sub-skill

Check for false precision. The assistant may describe a pupil as having a particular misconception when the table only shows a low score. Unless the assessment response or your note demonstrates the misconception, change this to a neutral statement such as requires further checking of inference strategy.

Also check the pupils not selected for a group. A process can miss a pupil with steady low performance across several skills because no single score falls below your threshold. Add a teacher-review list for these cases rather than forcing them into an ill-fitting group.

5. Turn approved proposals into a repeatable record

After you approve placements, save a brief intervention record. Use the same fields each cycle so that your review meeting compares like with like:

• assessment name and date
• target skill
• approved pupil IDs
• starting evidence
• planned number of sessions
• teaching focus
• review assessment and date
• decision after review: continue, adjust, exit, or gather more evidence

This record separates the initial proposal from your professional judgement. It also means you can explain why a pupil entered a group and what evidence will be used to decide whether the support is still needed.

When the grouping does not work

If the output is vague, reduce the task. Provide fewer skills, define the threshold again, and ask for one group at a time. If the evidence statements do not match the table, ask the assistant to quote the exact score or note supporting each proposed placement, then reject any placement without that evidence.

If groups are too large or too mixed, do not ask for a neater list. Re-check the underlying items and split by error pattern only where the assessment supports it. If the assessment cannot show a clear pattern, collect a short follow-up check before changing provision.