How to Check Whether Learning Recovery Is Working

Before collecting another score, finish this sentence: “After this check, we need to decide whether to…” The ending might be continue the current explanation practice, change the examples, reduce a temporary prompt or ask the teacher what an answer must demonstrate. Each ending calls for slightly different evidence.

Suppose a learner has practised selecting information from tables. If the next decision concerns reading a row and column correctly, a short task containing several relevant tables may help. If the decision concerns explaining a conclusion from those tables, more information-finding questions alone cannot settle it. Both activities involve tables, but they ask the learner to supply different thinking.

This is why “Is recovery working?” needs a smaller companion question. Try: “Can Sam now justify a comparison from the data without being told which figures to use?” The question names an action and a support condition. It gives the observer something to look for in the work, and it makes a limited positive answer possible.

Choose a decision that matters soon. A learner preparing for next week’s class discussion may need to organise a spoken explanation. Someone who must complete an independent written task needs evidence about writing that explanation. Neither needs a test of every skill that might eventually become useful. A review becomes expensive when it collects information that nobody will act on.

Explain the purpose to the learner before starting. “We are checking whether the new way of selecting evidence is useful” feels different from an unexplained demand to prove improvement. Invite the learner to point out an important condition: perhaps the earlier questions included unfamiliar words or the adult had already discussed every answer.

The CMU guide to formative assessment describes using learning evidence to guide teaching and subsequent learning, with checks that can be low stakes.

The practical consequence here is simple: the review should produce a teaching decision. If every possible result leads to “do another worksheet”, the proposed check needs more thought. Decide what a particular error, an independently completed answer or an ambiguous result would change before asking the learner to spend time producing it.

Contents · Next chapter

A capability is easier to examine when it contains an action, some relevant material and a requirement. “Understand data” is broad. “Use the correct entries from a small table to justify which of two options meets a stated condition” gives the learner and observer a shared object of attention.

Now identify what would remain on the page or in a recorded explanation. Perhaps the answer names the relevant entries, compares them accurately and connects that comparison to the condition. A correct choice with no reason might satisfy one task but not another. Write the requirement before seeing the response so that the standard does not quietly rise after success.

The evidence need not always be a mark. A labelled diagram, an explanation of a choice or a short worked solution can reveal the action more directly. However, if you use a checklist, each entry should describe something observable. “Shows good understanding” is difficult to inspect. “Explains why the alternative fails the stated capacity requirement” is clearer.

Avoid dividing an answer into so many tiny points that the important relationship disappears. A student can mention several correct words without connecting them. If the capability is explaining, preserve a criterion for the connection itself. Spelling, presentation and speed may deserve attention elsewhere, but do not let them silently determine an evidence-selection score unless they belong to the stated purpose.

The CMU guidance on assessment alignment recommends matching the work used to assess learning with the capability the teaching intends to develop.

Apply that principle by trying your criterion on two invented responses. One should meet it; the other should be close but incomplete. Can you explain the difference without relying on whether the answer “sounds better”? If not, clarify the requirement before using it to judge progress.

For example, “Option B is better because it has more” leaves the important quantity and requirement uncertain. “Option B has 18 places, so it can accommodate all 17 participants” makes the connection inspectable. This does not establish a universal writing formula. It shows the particular relationship being checked. A different task may require a different kind of explanation.

Contents · Previous chapter · Next chapter

Two sheets can both be labelled “data interpretation” while containing very different work. One may mostly ask students to copy a number. Another may require comparisons, calculations and explanations. Similar headings, equal lengths and matching maximum scores do not establish similar demands.

Write a short description of the kinds of work in the check. Call it a task mixture or a blueprint, whichever is clearer to you. It should be detailed enough to expose a consequential change but short enough to use. Three relevant categories are often more informative than a long list of chapter names that says little about the thinking required.

In the fictional case used later, the target concerns reading small data tables. The observer distinguishes these demands:

The table describes this invented learning task. It is not an official examination classification. A real teacher may need other categories, such as interpreting a scale, handling missing information or distinguishing an observation from an inference.

Record how many opportunities each family receives. If the uncertain capability is justification, it needs enough opportunities to be visible. A sheet containing eleven locating questions and one justification question is mainly a locating check, even if the adult’s concern is explanation.

Keep other material features reasonably similar when comparison matters: reading load, number size, vocabulary, number of steps, visual arrangement and time allowance. You cannot usually prove that homemade questions have exactly equal difficulty. You can still document the obvious differences and avoid treating the resulting totals as a precise measurement instrument.

Use fresh questions within the same description. Repeating the identical answer may show useful recall, but it leaves open whether the learner can select the method again. Change the surface details while preserving the decision being checked. If you deliberately add a new representation or an extra step, label that as a separate extension.

Keep one earlier work sample beside the later sample. This allows the review to compare actual reasoning, not merely category names. “Three comparison questions” can conceal very different demands. The blueprint helps you notice a mismatch; inspecting the work tells you whether the mismatch matters for the decision you need to make.

Contents · Previous chapter · Next chapter

“With help” is often too vague to be useful. An adult may clarify an unfamiliar instruction, point to the relevant row, identify the operation or supply the sentence that completes the reasoning. These contributions leave very different amounts of the task for the learner to perform.

For an informal check, preserve the first attempt and note any later assistance. You might write, “Selected both figures independently; adult asked which was larger; learner corrected subtraction.” This keeps a valuable learning event visible without turning the corrected answer into an unsupported claim of independent comparison.

Do not invent a numerical penalty for hints. Subtracting an arbitrary half mark cannot tell you which decision the learner supplied. A short description is usually more informative. If a taught checklist is available, record whether the learner chose and used it or whether an adult directed each step.

Keep necessary access supports separate from temporary instructional prompts. An appropriate assistive tool, accessible format or agreed accommodation does not automatically make an answer less valuable. Specify the task and its permitted conditions. Independence means the learner supplies the relevant thinking within those conditions; it does not mean learning without every tool or support.

For example, if reading a small printed label is not the capability being examined, an accessible version of that label may make the evidence more relevant. Removing it merely to make the check “harder” changes what the task asks. When established arrangements apply, follow them and seek clarification from the responsible educator if their role in this particular check is uncertain.

Also record whether examples, notes or answers were visible, whether the learner had just rehearsed similar work and whether the available time changed. A complete transcript is unnecessary. Capture conditions that could plausibly alter the conclusion. “Quiet room” may add little; “stopped halfway for an unexpected interruption” matters substantially.

Discuss help openly with the learner. The record is not an investigation into whether they deserve credit. It is a way to choose teaching accurately. A successful answer after a precise hint can identify a useful next lesson. The companion guide on reducing help without losing progress develops how to test a smaller prompt while keeping a route back to support.

Contents · Previous chapter · Next chapter

A percentage answers a question about a particular whole. Six correct answers out of eight attempted questions is 75%. Six correct answers out of twelve assigned questions is 50%. Both calculations can be correct, but they describe different facts. Writing only “75%” hides the difference.

Use counts alongside percentages, especially with small sets. One changed answer in a four-question check moves the percentage by 25 points. That is a large movement on a chart and a small amount of work on a desk. The size of the numerical change does not tell you whether the evidence is broad or dependable.

Separate at least three situations when they occur: an attempted answer that misses the requirement, an answer that meets it and an opportunity that was not observed. If the learner never reached the final questions because the session was interrupted, those items do not show an incorrect method. If the learner chose to leave them blank during an otherwise completed timed task, that also needs recording, but the relevant question may involve pacing, understanding or both.

Do not automatically remove every blank from the denominator. That could make the result look stronger by excluding the hardest items. Instead report the assigned total, the attempted total and why any work is missing. Interpret each in relation to the purpose. Completion matters when completing the whole task is part of the capability being checked.

Distinguish percentage points from proportional change. Moving from six correct out of twelve to nine out of twelve means moving from 50% to 75%, an increase of 25 percentage points. It is also three additional successful responses. For a small learning review, that last description often carries the clearest meaning.

Do not average percentages from unequal sets unless there is a reason for that weighting. Three out of four and four out of eight combine to seven out of twelve, about 58.3%. Averaging 75% and 50% gives 62.5%, which assigns equal weight to two differently sized sets. Neither operation repairs differences in task demand or support.

The safest first record is plain: “Nine of twelve fresh items met the agreed requirements on the first attempt; all twelve were attempted; the table remained available; no instructional hints were given.” You can calculate a percentage afterwards. The sentence already protects several conditions the percentage cannot communicate.

Contents · Previous chapter · Next chapter

Sam’s following results, dates, questions and teaching events are entirely invented. They illustrate a review process; they are not student records, testimonials or evidence for the effectiveness of a programme. The sequence includes several checks to expose different comparison problems, not to prescribe that testing schedule for every learner.

Sam is practising how to support a conclusion using information from small tables. The fictional first check, A, contains twelve items: three locating, three comparing and six justifying. The tables remain available, all items are attempted and there are no instructional hints. Each item carries one point when its specific requirement is met.

Check B occurs after several learning sessions. It uses different questions, doubles the locating items and halves the justification items. During corrections, an adult points out the correct pair of numbers for one comparison and supplies the missing connection for one justification. Both answers then meet the requirements. No other answer changes.

The tempting headline is “Sam improved from 50% to about 66.7%.” It compares A’s first attempts with B’s final corrected work and overlooks the changed task mixture. The arithmetic describes the two totals, but the sentence does not establish the intended recovery.

A more useful reading begins with the target. Justification had two independent successes among six opportunities in A. B has no first-attempt successes among three opportunities, followed by one successful response with a supplied connection. That does not demonstrate improved independent justification. It also does not prove a general decline: the items differ, and there are fewer opportunities.

The first-attempt totals happen to be equal, six out of twelve. That equality is not proof that nothing changed either. The locating share has increased, while the justification share has decreased. The check now asks a different combination of questions. Neither the rising corrected total nor the unchanged first-attempt total cleanly answers the original review question.

Sam’s teacher keeps the corrected work because it shows what happened when help was added. Then she prepares a fresh check with the earlier mixture. This decision does not discard the learning sessions or punish Sam for accepting help. It repairs the comparison so that the next record can say something more useful about the capability they are trying to strengthen.

Contents · Previous chapter · Next chapter

The next fictional check, C, returns to three locating, three comparing and six justification items. It uses fresh tables with similar reading load and number demands. As in A, the tables remain visible, there are no instructional hints and Sam attempts all twelve items within the same agreed time allowance.

Sam now meets the requirements on three locating items, two comparison items and four justification items. The total is three plus two plus four: nine out of twelve, or 75%. Compared with A’s six out of twelve, that is three additional successful first attempts and a difference of 25 percentage points.

The component record matters. Locating changes from two of three to three of three. Comparison remains two of three. Justification changes from two of six to four of six. The target therefore contributes two of the three additional successes. The improved total is no longer explained entirely by work outside the target.

Now inspect one of the answers. A fictional table lists 14 available places for Workshop Cedar and 18 for Workshop Maple. The question asks which workshop can accommodate a group of 17 without dividing it. Sam writes, “Maple has 18 places, which is enough for 17. Cedar has only 14, so the whole group would not fit there.” This meets the stated requirement to connect both capacities to the group’s size.

Another item asks whether a table establishes that every participant arrived on time. The table gives only the number present at the start and the total registered. Sam repeats both figures but concludes that everyone was on time even though the start count is lower. The answer is attempted, but the conclusion does not follow. A high total should not erase this remaining difficulty.

The teacher can now make a narrower statement: “On this fresh check with the earlier mixture and no instructional hints, Sam produced more acceptable justifications. He still needs work on conclusions that the data do not support.” That sentence recognises progress without declaring every form of data reasoning secure.

Check C is fairer, not perfectly equivalent to A. The teacher has controlled several visible features, but has not statistically equated the questions. The useful next decision is therefore based on both the numerical pattern and the inspected answers. Continue appropriate explanation teaching, focus the next examples on unsupported conclusions and arrange a later opportunity to see whether the relevant thinking remains available.

Contents · Previous chapter · Next chapter

An immediate success answers a useful question: what can the learner do now, close to the teaching? A later attempt asks another: what remains available when the earlier explanation is less immediate? Record both the elapsed time and any relevant practice between them. “One week later” does not mean “without any further exposure” unless that is actually what happened.

The IES study guide recommends helping learners judge learning after a delay; it does not prescribe the intervals in this fictional example.

In the invented record, check D is planned seven days after C. Ordinary learning has continued, but the same questions have not been rehearsed. The blueprint again contains three locating, three comparing and six justification items. An unexpected interruption ends the session before the last two justification items are reached.

Sam succeeds on all three locating items, two of three comparison items and three of the four attempted justification items. That is eight successful answers among ten attempted items. Two planned justification opportunities are unobserved. The record should show all three facts.

Eight out of ten attempted is 80%. Eight successful responses out of twelve planned is about 66.7%. Neither percentage should be placed beside C’s 75% as an uncomplicated trend. The first excludes two questions; the second counts the planned opportunities even though the last two were never observed. Neither tells us what Sam would have written there.

Do not mark D “useless”. It contains evidence of successful later work on eight items and two unsuccessful attempts. Inspect them. However, the interruption prevents a complete comparison under the planned conditions. Preserve D as a partial observation and seek a modest fresh opportunity to resolve the missing question.

Three days later, check E uses the full original mixture with new content, the usual time allowance and no instructional hints. Sam meets three locating, two comparing and four justification requirements: nine out of twelve again. This is another encouraging sample under the stated conditions. It does not prove that forgetting never occurs or establish a universal interval for future review.

The Consolidate guide develops practice that makes partly successful learning more dependable. Here, the review’s job is narrower: separate the evidence of a later attempt from the effects of missing work, changed help or rehearsal, then state what decision that evidence supports.

Contents · Previous chapter · Next chapter

A matched comparison helps answer whether performance has changed under broadly similar demands. A changed task asks whether the capability travels to a nearby situation. Both questions matter, but mixing their scores can make each harder to interpret.

After E, Sam tries an invented planning task presented as two short notices rather than a table. One activity has 20 places but requires attendance on both days. Another has 16 places and allows attendance on either day. A group of 18 can attend only the second day. Sam must identify the relevant constraints, evaluate both options and explain whether either satisfies the complete request.

For this original task, the observer records three requirements: preserve the group size and attendance condition; evaluate the capacity and attendance rule for each option; and reach a conclusion that follows from both. Sam identifies the two constraints and correctly notes that the 16-place activity is too small. He recommends the 20-place activity without addressing its two-day requirement.

The answer meets the first requirement. It does not meet the second in full or the final requirement, because neither option satisfies the request as written. The useful observation is that Sam carries the capacity comparison into prose but misses a second constraint when making the overall decision. There is no need to convert one of three into a percentage and compare it with nine of twelve.

Choose the next teaching move from that boundary. A teacher might model how to list all requirements before checking each option, then invite Sam to try a different two-condition task. That is a response to the actual missing relationship. Repeating the old locating questions would not directly address it.

Do not describe the changed result as proof that the earlier progress was false. The earlier checks showed successful work with small tables under specified conditions. The new task introduces prose and combined constraints. It extends the question. Conversely, do not dismiss the difficulty merely because the task is new; it may be close to the real use the learner needs.

Label each record by its purpose: comparable follow-up, later check or changed-situation check. This small distinction prevents a chart from implying that every number belongs on one scale. Progress becomes clearer when you can say both what is now more available and where the current evidence stops.

Contents · Previous chapter · Next chapter

Sometimes the uncertainty lies partly in the requirement or the observer’s judgement. Two adults may read the same explanation differently. One accepts a reference as clear; another thinks it leaves the relevant quantity unspecified. Before treating the disagreement as evidence of a learner’s weakness, compare the criterion and the response together.

Use a small example. The requirement asks the learner to explain why a claim exceeds what a table shows. The response says, “It only gives the total.” Does that identify the missing information clearly enough? The answer depends on the exact question and the agreed expectation. If the table contains a total but no timing information, a fuller response may need to state that arrival times cannot be established.

Read the complete task rather than judging an isolated sentence. A pronoun can be clear in one context and ambiguous in another. Where a requirement is genuinely uncertain, record that uncertainty. Do not silently give the later work a stricter interpretation and then report a fall in capability.

For recurring informal reviews, keep one example that meets the requirement and one that does not, with a sentence explaining why. Use them to check your judgement. These examples are aids to consistency, not a validated scoring system. Replace them if they no longer represent the task, while making that change visible in the record.

If the question concerns an actual school assessment, ask the responsible teacher about the expected reasoning. Bring the prompt and the learner’s response. “Does this explanation establish the required link?” is easier to answer than “Why did the marks fall?” Avoid deriving a whole teaching programme from an uncertain interpretation of one mark.

Let the learner explain what they intended after the original response is preserved. Their explanation can identify a useful teaching need, such as making an implicit connection explicit. It does not retroactively turn the original written answer into a complete one. Record the two observations separately: what the writing conveyed and what the later conversation revealed.

Contents · Previous chapter · Next chapter

A sequence can show that performance changed after some teaching without proving that one particular teaching move caused the change. Sam also attended lessons, encountered related work and became familiar with the review procedure. The fictional records do not isolate every influence.

This does not make the observations unhelpful. The teacher needs to decide what to do next, not publish an experimental estimate. She can say that the present work supports continuing a useful line of teaching while refining a remaining difficulty. She should avoid saying that one worksheet format produced a measured amount of lasting recovery.

Be especially careful with tiny numerical changes. Moving from eight to nine successful answers in a short check may be encouraging, but different items and ordinary variation remain possible explanations. Inspect the added success. Does it show the relationship that teaching targeted, or is it an unrelated easier item?

The learner’s account can help identify possibilities. “I remembered to compare both conditions” points towards a decision worth examining in the work. “I guessed this one” changes how confidently that answer supports a capability claim. Neither statement should replace inspection; each is additional information about the attempt.

Avoid creating a universal rule such as “three improved checks mean recovery is complete”. The amount and kind of evidence needed depend on the decision, the capability and the consequences of being wrong. A low-stakes decision to try a slightly more demanding practice task requires a different level of assurance from a formal assessment judgement.

Keep the language proportional: “These attempts support…”, “The later sample suggests…” or “We still need to check…” are useful when followed by a precise statement. They should not become vague hedging around a hidden guarantee. Name the condition and the next action.

For further practice in separating events, observations and explanations, use the Recovery Casebook. Its cases help you question a persuasive story. This guide turns that caution into a repeatable review: define the decision, preserve the conditions, inspect the response and leave unsupported causes open.

Contents · Previous chapter · Next chapter

The review is complete when it produces a proportionate next action. Three useful directions are continue, change and clarify. They are practical choices, not grades for the learner or permanent labels for the teaching.

Continue when the evidence supports a useful direction and the remaining work is clear. In Sam’s case, fresh matched checks show more successful justification under the stated conditions. Continuing explanation practice is reasonable, while focusing on conclusions and combined constraints. Continuing does not require repeating every earlier activity at the same volume.

Change when the work reveals a specific mismatch. Perhaps the learner can select evidence but cannot explain its relevance. More selection-only questions would leave that relationship untaught. Change the modelling or practice to address the missing connection, and specify what a later attempt should reveal.

Clarify when the evidence cannot distinguish the possibilities that matter. A changed blueprint, uncertain scoring or unrecorded help may prevent a clean conclusion. The next move might be one short comparable task or a conversation about the requirement. Clarifying can be more useful than immediately adding lessons or increasing the workload.

These directions can coexist for different parts of a capability. Continue ordinary practice for locating information, change how unsupported conclusions are taught and clarify the role of an ambiguous instruction. Avoid letting one broad label conceal those separate decisions.

Include the cost of the next check. If five minutes of fresh work will settle the question, there may be no reason to assign another full paper. Use existing classroom evidence when it meets the need. Additional monitoring should earn its place by improving a decision.

If the learner is returning after a substantial interruption, first establish what current work requires and what knowledge remains available. The companion guide on returning to learning after a long break covers that re-entry problem. A progress comparison becomes more meaningful once there is a clear starting task and an achievable first learning cycle.

Contents · Previous chapter · Next chapter

This exercise uses fictional writing records, separate from Sam’s data tasks. All scores, support events and interruptions are invented. The task is to review evidence about writing useful instructions, not to apply an official English marking scheme.

Rosa is learning to write a short message that helps a recipient complete an activity. Each original practice prompt specifies eight requirements: two concerning purpose and audience, three concerning required actions and three concerning conditions or exceptions. Each requirement counts once in this informal checklist. An answer must communicate the relevant information accurately to meet it.

In record P, Rosa uses a fresh prompt and writes without instructional help. She meets both purpose-and-audience requirements, two of the three action requirements and one of the three condition requirements. She attempts the complete message. The prompt remains available and she has fifteen minutes.

Record Q follows teaching. A different prompt contains three purpose-and-audience requirements, four action requirements and one condition requirement. Rosa meets all three, all four and the one condition requirement in the final message. However, an adult supplies the sentence that states the condition. Before that contribution, seven of the eight requirements are met. The whole task is completed within fifteen minutes.

Record R, four days later, returns to the original two-three-three mixture. Rosa writes a fresh message with the prompt available, the same time allowance and no instructional help. She meets two purpose-and-audience, three action and two condition requirements. All eight are attempted. The missed condition concerns the recipient’s action if equipment is unavailable.

Record S is planned a week later with another fresh prompt and the original mixture. An interruption ends the session before Rosa writes the final condition. Of the seven requirements represented in what she has written, six are met: two purpose-and-audience, three action and one condition. One attempted condition is inaccurate and the remaining condition is unobserved. Do not assume what the unfinished sentence would have said.

Finally, Rosa tries a changed task. Instead of writing a message from a supplied prompt, she must revise this invented instruction after reading a new update:

Original instruction: “Collect your materials at Room 4 on Thursday before 14:00.”

Update: “Room 4 is unavailable on Thursday. Collection moves to Room 7. The deadline remains 14:00. Anyone unable to attend must email the organiser before Wednesday at 17:00.”

Rosa revises it to: “Collect your materials at Room 7 on Thursday before 14:00. If you cannot attend, email the organiser.” For this task, check three requirements: apply the location change, preserve the collection deadline and communicate the complete alternative arrangement including its earlier deadline.

Before reading the answer, write a short review:

1. Calculate P, Q and R’s first-attempt totals, and distinguish Q’s final supported total.
2. Explain why Q does not establish independent recovery in handling conditions.
3. Compare P with R, including the change in condition requirements met.
4. Give both useful denominators for S and explain the limitation of each percentage.
5. Assess the changed task against its three stated requirements.
6. Choose one continue, change or clarify decision, and describe evidence that could inform the next review.

Keep your proposed check small and relevant. The purpose is to resolve the remaining question about communicating conditions accurately. A longer spelling test or another completion tally would answer a different question.

Contents · Previous chapter · Next chapter

P meets two plus two plus one requirements: five out of eight, or 62.5%. Q meets seven out of eight before the adult’s contribution, or 87.5%, and eight out of eight afterwards, or 100%. R meets two plus three plus two: seven out of eight, or 87.5%, on a fresh independent attempt.

Q has a different mixture. It contains only one condition requirement, compared with three in P and R. Furthermore, the adult supplies the sentence that fulfils that one condition. The final full score therefore does not establish that Rosa can independently handle the conditions that the recovery work targets. The seven other successes remain useful observations about the work she supplied herself.

P and R offer a more relevant comparison because their broad mixture, time allowance and recorded support conditions match. The total rises from five of eight to seven of eight: two additional requirements met, a difference of 25 percentage points. Purpose and audience remain two of two. Actions improve from two of three to three of three. Conditions improve from one of three to two of three.

That pattern supports a limited positive statement about the sampled writing. It does not establish equal prompt difficulty, dependable performance on every exception or a cause for the change. The missing equipment condition identifies a useful boundary to examine in the actual response.

S meets six of seven observed requirements, about 85.7%. It also meets six of the eight planned requirements, or 75%. The seventh observed requirement is attempted but inaccurate; the eighth is unobserved because of the interruption. The first percentage excludes the missing opportunity. The second includes the planned whole but cannot tell us how Rosa would have completed it. Neither supports a simple claim of improvement or decline from R.

In the changed task, Rosa applies the location change and preserves Thursday’s 14:00 collection deadline. She states that someone unable to attend should email, but omits the earlier Wednesday 17:00 deadline. The complete alternative-arrangement requirement is therefore unmet. Two of the three stated requirements are met. Reporting that as another point on the earlier eight-requirement trend would mix different tasks.

A defensible decision is to continue message-writing practice while changing the examples to make alternative actions and their separate deadlines explicit. A short fresh prompt with a main instruction and an earlier exception deadline could reveal whether Rosa communicates both without a supplied sentence. Record the prompt, the first response and any help, then inspect the relationship. Another reasonable response is to clarify an ambiguous criterion before choosing that check. The decision earns its place through the uncertainty it resolves.

Contents · Previous chapter · Next chapter

Keep the review record compact enough to use again. A parent, teacher or learner should be able to read it and identify the capability, the conditions, the evidence and the next move. A folder full of percentages without those relationships is harder to use than one clearly described comparison.

Attach a small amount of actual work where possible. One successful explanation and one revealing incomplete answer can make the summary easier to interpret. Keep corrections distinguishable from original attempts. Do not rewrite the first record merely because the learner now understands the answer.

Use the review conversation to recognise a concrete achievement: “You supplied the comparison yourself on the new questions.” Then name the remaining condition: “We still need to see whether you check both constraints in prose.” This gives the learner an accurate account of progress without making a promise about every future task.

If the main difficulty has become interpreting personal expectations after a setback, use Rebuilding Confidence. If several subjects compete for attention, return to Academic Recovery to fit this evidence into the whole plan. A progress review should inform that plan without taking it over.

Choose one current learning question. Put the relevant earlier attempt beside a fresh, suitably comparable task. Preserve the conditions and the learner’s first response. Then make the smallest useful decision the evidence supports. The value of the review lies in what it helps you teach, practise or clarify next.

Contents · Previous chapter · Continue through the learning library