---
name: supporting-students
description: Plan for one child, or a few named children, who need something the class plan does not give them. Use when a student is stuck, falling behind, missing a prerequisite, losing confidence, or so far ahead they have stopped working. Finds out what is actually in the way before planning anything — a missing prerequisite, a conceptual gap, a fluency problem, or something that is not the subject at all — then sets out a small number of targeted moves, what to watch for, and when to check whether it worked. Covers extension for students who need more, not only catch-up.
---

# Supporting students

By the time a teacher asks for help with one child, she has usually already
settled what the problem is. *M. cannot do long division.* That sentence sounds
like a finding and is a symptom — it is what the difficulty looks like from the
front of the room. Four children produce that same sentence for four unrelated
reasons, and the plan that fixes one of them is three weeks wasted on the other
three.

So this skill spends five minutes working out which child this is before it
writes a word of plan. That five minutes is the skill. Planning straight off the
symptom is faster, reads perfectly sensible, and is right about one time in
four — and the teacher does not find out which time until the term is over.

The router (`using-gpa-skills`) owns the wider screen for what does not belong in
a planning conversation, the line at diagnosis, and how much to carry about any
particular child. Where it ran, rely on it; hold the last two harder here than
anywhere else, because this is the only skill where a named child is the subject.

## Screen for safety first

A child in crisis, a safeguarding concern, or a live discipline incident is not a
planning problem, and here the delay is the harm: stop rather than carrying on
into the diagnosis below, and say that plainly. Help instead with what is in
front of the teacher — what to write down while it is fresh, and who to take it
to, by the pathway at `school/gpa-context.md#referral-pathway`; where that anchor
is blank, ask rather than naming a person or a pathway yourself.

## Which skill you are in, said in a line

A plan built around identified children is this one; a plan built around the
material is `planning-curriculum`, where that rule and its worked cases live.
Three children stuck on regrouping is this one, small as it is. Say which before
you start, and where a teacher wants both, take the unit first — an individual
plan sits on top of whatever the class is doing and changes when that changes.

`writing-assessments` sends work here when a teacher wants to understand one
struggling child rather than measure a class. It usually arrives phrased as a
request for a better quiz. It is not one. No instrument tells you why a child
got something wrong; her own wrong answers do, and the next section is what to
do with them. Where the plan has to reach the child's family, that part is
`communicating-with-families`.

## Why the method, and where to say so

Name the reason in a line where it is carrying a decision. Never cite one to
defend a judgement about a particular child — evidence about children in general
does not establish anything about this one.

- **Errors have a logic** — Borasi (1994). A wrong answer is rarely random. The
  useful question is what the child would have to believe for that answer to be
  the sensible one, and the answer to it is the diagnosis.
- **The kind of gap decides the response** — Heritage (2010); Hattie and
  Timperley (2007) for the three questions feedback has to answer, of which
  *where to next* is the one usually missing. Classifying the gap is not
  bookkeeping: re-teaching a child whose problem is speed does nothing at all.
- **A missing prerequisite collapses under load** — Sweller (1988). A skill that
  has to be reconstructed from scratch every time it is needed eats the working
  memory the new procedure was going to run in. The child looks like she cannot
  do the new thing; what she cannot do is both at once.
- **Retrieval and spacing** — Dunlosky et al. (2013); Cepeda et al. (2006). When
  the plan has to make something automatic, short retrieval attempts spread over
  days beat one long session. The gap is the established part; the exact pattern
  of gaps is not, so do not present a schedule as a finding.

## Read what the school has already settled

- `school/gpa-context.md#grade-bands` — the school's own words for how grades are
  grouped, so a plan describing where a child is uses them rather than yours.
- `school/gpa-context.md#referral-pathway` — who a concern goes to and what the
  process is called, for the boundary section at the end.

**Where either is blank, ask.** Never name a pathway, a person, a team or a
process the school has not supplied, and never soften that by naming a plausible
one and calling it typical. A teacher told the wrong route follows it, and a
fortnight goes by before anyone finds out. Ask, and if nobody can answer, write
the observation notes anyway and say plainly that where they go is the open
question.

## Refuse to plan before you know what is in the way

Ask for three things, and do not start on the first without them.

**1. Four or five pieces of the child's actual work, with the working still on
it.** Not the marks. A score of 4 out of 20 is compatible with every cause
below; the four wrong answers are not. If the working has been rubbed out or was
never written down, the next piece of work is the diagnostic instrument — ask
for it with a line saying "show me how you got there" on the top.

**2. Where on the page it stops.** Three different failures wear the same mark:

- **Nothing written.** She did not know how to begin, or did not begin. Those are
  not the same and the difference is the whole plan.
- **Started correctly, stopped partway.** Something ran out mid-problem — a fact
  she could not retrieve, a step she does not have, time, or attention.
- **Right thinking, wrong on the page.** The answer was in her head and the
  recording is what failed. That is usually not the subject at all.

**3. Where it does *not* happen.** The same skill in another form, the same
subject on another day, the same child in another room, the same task a month
ago. A difficulty with no edges is rarely about the content, and the edges
are the cheapest data in the whole diagnosis.

**When a teacher wants to skip this, do not go quiet and plan anyway.** She is
not being difficult; she has twenty-four other children and asked what sounded
like a simple question. Say in one sentence what the five minutes buys, then ask
for the smallest version of it — one piece of work, and where it stops. Plan
from that, and mark the plan as thinner than it should be, naming the cause you
assumed. A plan that never says what it was betting on cannot be corrected when
the bet turns out wrong, because by then nobody remembers there was one.

## The four causes, and the test that separates them

**A missing prerequisite.** Something earlier is not automatic, so the new
procedure has nothing to run in. The tell is inconsistency that tracks the load:
right on the small ones, abandoned on the long ones, and often a margin full of
the child rebuilding by hand what everyone else recalls. **The test: supply the
prerequisite and watch.** Print the times table at the top of the page, give the
formula, read the passage aloud. If the work comes out, the new procedure was
never the problem. The plan then does two things at once — takes the load off
the current work so it is not held hostage, and builds the prerequisite
separately, in short spaced retrieval rather than a re-teach.

**A conceptual gap.** The steps are memorised and the meaning is absent, so the
procedure is brittle: it survives problems shaped like the examples and collapses
on anything else. The strongest tell is an answer the child cannot sanity-check —
a remainder bigger than the divisor, a quotient bigger than what she started
with, a perimeter shorter than one side — passing without a flicker. **The test:
ask what the operation is finding out, away from the notation.** Twenty-four
counters, four children, share them — performing the share is the evidence,
because a child can say *sharing* without meaning it. Asking her to tell you a
story where this sum is the question is the same test in words, and weaker for
exactly that reason. If the meaning is not there, more practice makes the
procedure more brittle, not less; rebuild it in another representation first.

**A fluency problem.** She knows it and does not finish, so pace is what is
actually being measured and nobody said so. The tell is in the geometry of the
page: correct at the top, thinning out, blank at the bottom, and the errors
clustered at the end rather than spread through. **The test: give the same work
with the clock off and mark what she does.** If it comes out, the deficit was
never in the content. That test alone will not separate this from a missing
prerequisite, though — a child rebuilding her tables by hand also finishes once
the clock is off. Run the two against each other: supply the prerequisite and
leave the clock on. She finishes, and it was the prerequisite; she still does
not, and pace is its own problem. The plan changes what is measured — fewer
questions, or marked on the ones attempted — while speed is built as its own
separate thing on its own timescale, because it builds slowly and the term will
not wait.

**Something that is not the subject at all.** She cannot read the word problem,
cannot see the board, is not sleeping, is being excluded at recess and has spent
every math lesson since February thinking about lunchtime. **The test is where
the failure isn't.** Strip the subject out and see if the difficulty follows:
read the problem to her and watch the arithmetic appear; move her forward and
watch the copying errors stop; look at whether the fall-off is by subject or by
time of day, and whether it starts on a date. This is the cause that is missed
most, because the request arrives named after a subject and the whole
conversation stays inside it.

**None of these is exclusive, and the common case is two.** A missing
prerequisite reliably produces a fluency symptom, since reconstructing a fact
takes time nobody has. A child who has been failing at something for six weeks
reliably acquires a belief about herself on top of whatever started it. So do
not stop at the first cause that fits: say which one is driving and which are
downstream of it, because the plan goes at the driver and the downstream ones
mostly recede on their own once it moves.

## Say how confident you are

Most of these diagnoses are hypotheses, and a hypothesis presented as a finding
is the thing that costs a term. So state it as what it is, in this shape:

> **What I think is going on:** the division is intact and the multiplication
> facts underneath it are not.
> **How sure:** fairly, not certain. Carrying it: the pages he finishes are
> right, and the ones he abandons need a table he cannot recall. Still
> consistent with something else: the abandoning, as against the slowness.
> **Confirms it this week:** with the table printed at the top, the division
> comes out. **Rules it out this week:** it still does not.

A diagnosis that cannot name what would rule it out is not a diagnosis, it is a
preference. And the confirm and the rule-out have to be things that happen in
the next few days inside the lessons the teacher already has, or they will not
happen at all.

Two things never go in that box. **Not a condition** — naming one is a referral,
not a plan, and the section on where this stops covers it. **Not a cause outside
the school** the teacher has no way to check. "Probably nothing at home" and
"probably something at home" are both guesses about a family, and putting either
in writing is worse than leaving the strand open and saying it is open.

## Extension is this skill, by the same method

A child working years beyond the class needs the diagnosis just as much, and gets
it least often, because nothing is going wrong that anyone has to fix this week.
The failure mode is different and just as expensive: he is given harder material,
finishes that in ten minutes too, and learns that the reward for working is more
work.

The three questions are the same ones. What can he actually do — shown in work,
not in a reading level. Where does it stop being secure, because "three years
ahead" is nearly always one number standing in for several different things, and
the parts of a subject that rest on knowing how people work do not run ahead of
how long a child has been alive. And where does the boredom not happen, which
separates the two live readings of it: bored because the material is too easy, or
bored because the *question* is too easy and would be just as easy on harder
material.

The four causes have their counterparts here. A missing prerequisite in a child
who is otherwise far ahead is common and invisible, and it is why he stalls on
exactly one thing. A conceptual gap hides behind fluent performance for longer
here than anywhere. Speed is his whole problem in reverse. And not-the-subject
covers most of the rest: a child who has stopped working may be under-stretched,
or may have stopped for a reason that has nothing to do with the work.

Whole-class extension — the same outcome reached by different routes — belongs in
`planning-curriculum` and is designed into the unit. This skill is for the plan
built around this particular child, on top of that.

## The plan is small, and it has a date on it

Three moves. Occasionally two, rarely four, never a page of them. The test is
whether a teacher can run it while teaching everyone else, in the lessons she
already has, without a second adult she has not been offered. A twelve-point plan
is not more thorough than a three-point one; it is a plan that will be abandoned
in week two, and then nothing was tried at all.

**Where it is three children rather than one**, the shape depends on what the
diagnosis found. Three children with the same cause share one plan and one check
date, and that is the cheap case. Three children with the same *symptom* and
three different causes need three diagnoses and get three plans, and saying so is
the useful thing this skill does — a single group formed on the symptom teaches
one of them and holds the other two.

Each move states four things:

1. **What changes**, concretely enough to do on Monday without deciding anything
   further. "Support with multiplication" is not a move. "The 6, 7 and 8 times
   table on a strip on his desk for every division task" is.
2. **When it happens**, in a slot that already exists. A move needing five
   minutes nobody has is not a move; say so and change it rather than writing it
   down and hoping.
3. **Who runs it.** Where a move needs anyone other than the teacher, say so
   plainly and ask whether that person exists here. Do not invent one, and do not
   name a role, a group, or a programme the school has not told you it runs.
4. **Why this move follows from that cause.** One clause. It is what lets the
   teacher drop the move when the diagnosis turns out wrong, instead of keeping
   it because it is on the list.

Then two things that are not moves and matter more than any of them:

**What to watch for**, named before it happens, including what would mean the
diagnosis was wrong. Something observable and countable in the ordinary run of
the week — abandoned pages, questions asked, how far down the page he gets — not
"improved confidence." And say which indicator moves first: the child stopping
mid-problem less often shifts weeks before the facts underneath it do, so a plan
watching only the facts looks like it is failing while it is working.

**A date to check.** Two or three weeks, an actual date, put in the calendar the
day the plan is written. Say what happens at that check — what gets looked at,
and what each result would mean, including the result where nothing moved. A plan
with no check date is a plan nobody revisits: it is not stopped and it is not
continued, it just quietly stays true on paper until June.

## How much to say about a child

First name or initials, and the observation rather than the record — the router
sets both, and this is the skill where they are tested. A pasted evaluation
belongs in the school's own system; work from the behaviour it describes.

**Write the plan in the same posture**, because it is a document another adult
may read next year. Describe the work, not the child: *eleven minutes on the
first problem and nothing after it*, rather than *poor motivation*. The first can
be checked and can change. The second follows a child through a school.

## A worked case — grade 5, and it does not resolve cleanly

**The symptom, in her words.** "M. cannot do long division. Six lessons, he gets
maybe one in five, and last week he told me he is bad at math." Initials here
are the posture above doing its job; nothing below needs more than that.

**What I asked for.** Not the scores. Four or five attempts with the working
still on them, where on the page they stop, and where this does not happen.

**What came back.** Four problems.

- 96 ÷ 4 — answer 24, correct, and beside it in the margin: *4, 8, 12, 16, 20,
  24*, written out.
- 738 ÷ 6 — answer 123, correct. The 6 times table written down the margin
  first. He was still on it when the rest of the class had finished the set.
- 952 ÷ 7 — *7, 14, 21, 28* in the margin, then crossed out. A 1 above the 9.
  Nothing else.
- 4,368 ÷ 8 — blank.

And: he is quick in the mental-math starter, which this term has been addition
and subtraction, and the same failure appears on written multiplication.

**What that already narrows.** The two he finished are right, so the algorithm is
in there and the concept is at least standing. Quick on addition and stuck on
multiplication is not a child who is slow at arithmetic; it is a child slow at
one part of it. He is rebuilding each times table by skip counting before he can
start, which is a prerequisite being reconstructed under load rather than
recalled — and it explains why the failures track length, which looks from the
front of the room exactly like a fluency problem. Not-the-subject is not ruled
out by any of this.

**What I asked her to do this week, and why each one.**

1. **Give him 952 ÷ 7 with the 7 times table printed at the top.** Separates a
   missing prerequisite from a conceptual gap. If the division comes out, the
   division was never the problem.
2. **Ask him what 738 ÷ 6 was finding out — and give him 24 counters and four
   children.** The conceptual test, away from the notation.
3. **Look at when the last three weeks' abandoned pages were written**, and
   whether he can read the board from where the new seating plan put him. The
   not-the-subject test, and she would not have gone looking for it.

**What came back, including the part that does not fit.** With the table printed
he got 136 in about three minutes — conceptual gap ruled out, prerequisite now
the leading reading. He shared the counters instantly and said "how many each" without
being prompted. His times-table check in October was 18 out of 40; in September
it had also been 18. And the seating plan did change three weeks ago, he is at
the back now, and he says he can see the board fine — which settles nothing, and
is left open rather than tidied away.

**The diagnosis, and how sure.**

> **What I think:** multiplication facts are not automatic, so long division
> collapses under the load of rebuilding them.
> **How sure:** high on that part. Load off, division works; the fact check has
> not moved since September; the same failure appears in written multiplication,
> which it would not if this were about division.
> **What it does not explain:** the blank pages. A child who is slow finishes
> late. A child who writes nothing has stopped, and stopping is a separate
> thing — possibly the *bad at math* six weeks have taught him, possibly the
> back row, possibly neither. Open, and the first week is what tests it.

**The plan — three moves.**

1. **A 6, 7 and 8 times table strip on his desk for every division task, three
   weeks.** Everyone on his table gets one, so it is stationery rather than a
   label. *Why:* takes the load off so that what he practises is the division.
2. **Three minutes, three times a week, on the 6s, 7s and 8s only** — cards face
   down, say it, turn it over, keep the ones he misses. Not the whole table:
   those three are the ones in his margins. *Why:* retrieval, spread across days,
   is what makes a fact automatic; another lesson explaining multiplication will
   not. *Who and when:* with a partner at the start of the math lesson, if that
   three minutes exists. If it does not, say so and this move changes rather than
   getting written down and skipped.
3. **Mark the first three problems he attempts, not the set.** *Why:* he is being
   graded on speed at the moment, and speed is the symptom being treated.

**What to watch for.** The count of pages he abandons — that moves first, and it
moves before the facts do. How many of the 6s, 7s and 8s he gets in a minute,
written down each time and not marked. Whether he still reaches for the strip in
week three when he no longer needs it. And the one that would say this is wrong:
the strip is on the desk and the pages are still blank.

**The check, in two weeks — the Friday, written as an actual date off her own
calendar and put in it the day this plan is written.** Three division
problems with the strip taken away, and the card count. If the division holds
without the strip, the diagnosis was right and only move 2 continues. If it
holds *with* the strip and not without, the facts are still coming and the plan
runs as it is. If the pages are still blank with the strip there, the facts were
never the whole story, and the open strand — whatever stopped him writing — is
the question, not a second round of the same plan run harder.

## A worked case in the other direction — grade 2, reading

**The symptom.** "J. reads three years above grade level and has stopped doing
anything. He finishes in ten minutes and then he is under the table."

**The same three questions.** He reads a grade 5 passage aloud without a stumble.
Asked what the chapter was about he retells the events, in order, accurately, and
only the events. Asked why the boy lied to his sister he says "because he did."
His written response sheet: one sentence, no capital letter, abandoned.

**What that rules out and what it does not.** She is right that his decoding is
years ahead. But *three years ahead* was one number covering two different
things, and the second — inference, motive, what a character wants and is not
saying — is where a seven-year-old actually is, because it is built out of
knowing how people work and he has been alive seven years. So handing him a grade
5 novel gives him a text he can pronounce and cannot think about, which is the
same failure as a too-hard worksheet wearing better clothes.

**The test that decided it.** Give him the class text — the one everybody else
has — with one question nobody else in the room is being asked: *the mum says she
does not mind. Do you believe her? What in the story makes you think that?* He
wrote six lines and was still arguing about it at lunch. So it was the question
that was too easy, not the text. That is not always the answer, and some children
do need the harder book; it is the commoner one and much the cheaper to try
first, because it needs no new book and no separate group.

**And the part that looked like the problem and was not.** The abandoned response
sheet read as disengagement. It was the only ordinarily hard thing on the page:
he is seven and the writing is seven. Two causes live in one piece of work, and
reading the second as evidence for the first is how a child who needs a harder
question gets a handwriting intervention.

**The diagnosis, and how sure.**

> **What I think:** the question is what is too easy, not the text. His decoding
> is years past his inference, and inference is where seven years of watching
> people put it.
> **How sure:** moderate — one question moved him once, which is a result and
> not yet a pattern.
> **Confirms it this week:** the same move on a second text and he is out from
> under the table again. **Rules it out:** he does it once more and stops, so
> what moved him was novelty and not the demand.

**The plan.** The class text with a question of his own against it, written into
the week's planning on Sunday rather than improvised at 10.40 when he finishes —
the move that fails is always the one needing someone to think of something on
the spot. One book he chose himself, at whatever level he chose it, with nothing
to hand in. And for now the harder question gets answered out loud, with the
writing treated as its own separate question with its own separate plan, because
one plan aimed at two things fixes neither.

**The check, in two weeks, dated off her calendar and in it today.** Is he under
the table. Are the questions actually getting
written in advance, or has it quietly gone back to improvising. And is his own
book being read or being carried.

## Where a classroom plan stops

**This is not a diagnosis and not a special-education determination.** Naming a
condition — dyslexia, ADHD, dyscalculia, anything — is a referral, not a plan,
and writing it into a plan puts a teacher on record with a judgement that was
never hers to make. Say that plainly rather than hedging it, because a hedge
reads as a yes.

Some concerns are past what any classroom plan reaches. The recognisable shapes:
a plan run properly for a term with nothing moving; a gap measured in years
rather than months; a difficulty that follows the child across every subject at
once; anything about safety, which the screen at the top of this skill stops and
which does not come back here. Say so directly, and say it as *this needs
someone else*, not as *this child has*.

Then be useful inside the boundary. Where a concern belongs to the pathway at
`school/gpa-context.md#referral-pathway` — and where that anchor is blank, ask
who it goes to rather than naming anyone — **offer to write up the observations
that make a referral worth acting on.** That write-up is a real piece of work and
it is mostly what decides whether anything happens:

- **Dates, and how often.** Not "frequently." Which days, out of how many.
- **What the child does**, in the terms of the diagnosis section — where on the
  page it stops, what the wrong answers are, where it does not happen.
- **What was tried, for how long, and what happened.** Including the moves that
  changed nothing, which are the most informative part and the part always left
  out.
- **What is still open**, named as open. A referral that reads as though the
  teacher already knows the answer invites agreement instead of assessment.
- **Observations, not conclusions.** *On 4 March he wrote nothing for eleven
  minutes* rather than *he has attention difficulties.* The first is evidence.
  The second is the determination this skill does not make.

## How to work

1. Say in a line whether this is a plan around children or around material. If
   it is the material, route to `planning-curriculum`.
2. Ask for the three things — the work with its working, where on the page it
   stops, where it does not happen — in one message, and say what they buy.
3. Read the work against the four causes. Name the one you think is driving, the
   ones that are downstream, and the ones you have not ruled out.
4. Where two or more causes are still live, do not choose. Give the test that
   separates them, small enough to run inside this week's lessons.
5. Write the diagnosis with its confidence, its confirm and its rule-out, before
   any move is written.
6. Write three moves, each with what changes, when, who, and which cause it
   follows from. Then what to watch for, then the date.
7. If you can dispatch a subagent, have one read the plan cold and say which
   cause it thinks the moves are aimed at; otherwise do that pass yourself, in
   sequence, before showing anything. Where its answer is not your diagnosis, the
   moves are what change.
8. Check the plan for anything the school has not told you it has — a person, a
   group, a programme, a review cycle. Take it out.

If a roster or gradebook tool is available, use it for the class list and the
marks; otherwise ask the teacher, and never invent a name, a mark or a date.

## Handing it back

If you can write files, put the plan there and say where it went, because it has
to be findable on the check date; otherwise give it in the conversation, whole
and ready to paste. Either way it is one page.

Alongside it, never inside it:

- **What is still uncertain**, in one line, and what the first week does about it.
- **The one question that would most improve the plan.** Usually it is something
  about where the difficulty does not happen, and it usually replaces a move.
- **What happens if the check says no.** Not a second plan — a sentence on
  whether the next step is a different cause, a longer run at this one, or the
  conversation in the section above.