The Tutor Handbook Vol No.0110 | The Accommodation Baseline — How a Tutor Compares Progress Without Treating Legitimate Access Supports as Noise or Quietly Changing Them Between Checks

The Tutor Handbook · Volume 0110 · Series ID THB-0110

Return to The Tutor Handbook

A learner completes a task with a legitimate access support. A week later the tutor removes that support “to see what she can really do”. Performance drops. The tutor concludes that the earlier success was not independent.

That conclusion may be wrong.

Independence is not the same as performing without every support. Some supports are temporary instructional scaffolds that should eventually fade when the learner is meant to internalise the operation. Other supports exist so that the learner can access the task without an irrelevant barrier. If the support does not perform the target capability, removing it can change the measurement condition rather than reveal a purer version of the same capability.

The direct answer is this: when a legitimate access support is part of the learner’s ordinary educational condition, the tutor should treat that support as part of the baseline unless there is a clear educational reason to test a different condition. Progress comparisons should say what support was available, keep relevant conditions as stable as practical, and avoid interpreting a change in access as though it were a change in learning. If the tutor deliberately changes the support, the result should be read as a changed-condition probe, not silently folded into the same trend.

This article is not about deciding who “deserves” an accommodation

The Tutor Handbook is not a clinical service and a private tutor should not diagnose a medical, psychological, sensory or developmental condition. Formal access arrangements in schools, national examinations and other regulated settings are governed by the relevant institution and its rules. This article addresses a narrower educational problem: once a legitimate access support is already part of the learner’s working condition, how should a tutor interpret progress without accidentally changing that condition from one comparison to the next?

The existing Access Support Boundary asks whether assistance is preserving access or performing the target skill. The Accommodation Baseline begins after that distinction has been made. Its problem is comparability across time. If the support condition changes, what does a rise or fall in performance actually mean?

Access support, instructional scaffold and answer assistance are different jobs

Three things can look similar from a distance because all three make a task easier to engage with. Their educational jobs are different.

An access support reduces a barrier that is not the target of the task. Examples can include a permitted assistive technology, enlarged or reformatted material, a physical access adaptation, an established communication support, or another arrangement appropriate to the learner and context. The exact legitimacy and form of such supports depend on the educational setting; a tutor should not invent formal entitlements.

An instructional scaffold supports a capability while it is being learned. A worked example, planning frame, cue ladder or partially completed solution may deliberately carry some of the operation for a time. Scaffolds are often designed to be reduced as the learner becomes more capable.

Answer assistance supplies information that the present task is meant to elicit. If the target is selecting a method and the helper names the method, the result is assisted. If the target is retrieving a concept and the prompt contains the concept, the retrieval evidence is contaminated. That assistance can still be useful for teaching; it simply changes what the result can support.

The Accommodation Baseline works only if the tutor keeps these jobs separate. Otherwise “remove support” becomes a crude instruction that may strip away legitimate access together with temporary scaffolding.

Why stable conditions matter for progress claims

Suppose a learner scores 60 on one task and 75 on a later task. The tutor wants to know whether capability improved. That comparison is already imperfect because tasks differ. If the later task also uses a different access condition, the interpretation becomes harder. A score is not produced by capability alone; it is produced by capability interacting with the task and the conditions of performance.

This does not require a tutor to turn lessons into a laboratory. It requires intellectual housekeeping. Record the conditions that materially affect interpretation. If the learner used the same established access support on both occasions, the comparison is cleaner. If the support changed, note that the evidence now answers a different question. The tutor may still learn something valuable, but should not pretend the conditions were constant.

The principle is familiar in formal assessment research: score comparability depends partly on whether differences in administration or response conditions alter what scores mean. Private tuition is not formal psychometric equating, and a three-student tutorial cannot reproduce large-scale measurement methods. The useful lesson is modest: conditions matter, and silent condition changes weaken strong claims about progress.

The baseline is a condition description, not only a number

Tutors often think of a baseline as a starting score. For accommodation-sensitive work, the number is incomplete without the condition. “12 out of 20” says less than “12 out of 20 on this task, with the established text-access support available, no content prompts, ordinary lesson time and independent response after instructions were clarified once.” The second description is not automatically perfect evidence, but it preserves more of what a later comparison needs.

The support condition should be described only as specifically as the educational job requires. Do not collect unnecessary personal or medical detail. The Record Minimum remains important: keep enough information to interpret teaching and continuity, not a private dossier.

Four questions define an Accommodation Baseline

• What is the target? Which knowledge, reasoning, skill or performance is the task meant to reveal?
• What barrier is the support addressing? Describe the educational access function without inventing a diagnosis.
• Does the support perform the target? If yes, it is not neutral for this particular evidence claim.
• Will the support condition remain comparable? If it changes, label the next result as a changed condition rather than the same baseline stream.

These questions are intentionally simpler than a formal accommodation decision process. A tutor is not replacing the school, clinician or examination authority. The purpose is simply to avoid changing the ruler while claiming to measure the same change.

Composite case: Alicia and text access

The following learner case is fictional and composite. Alicia works with a text-access support that changes how a passage is presented to her. The support does not select evidence, infer meaning or formulate an answer. During several tuition sessions she uses the support consistently while practising inference questions. Her evidence selection and explanation improve.

A new tutor decides to remove the support for one session because “we need to know if she can really comprehend”. Alicia spends much more time decoding the presentation, loses place repeatedly and gives shorter answers. The tutor writes, “Comprehension has regressed under independent conditions.”

The problem is the condition label. The second task was not simply “more independent”. It changed access. If the support had not been performing inference, then removing it introduced an additional demand. The lower performance may reflect that changed demand, weaker inference, both, or something else. The tutor cannot recover the clean comparison by calling the support artificial.

A better next step is to restore the established access condition and check inference on a fresh passage. If the tutor has a legitimate educational reason to explore performance without the support, that can be recorded separately as a changed-condition observation. It should not silently replace the main baseline.

Composite case: Beatrice and the planning frame

Fictional composite case. Beatrice uses a detailed planning frame for composition. At first the tutor describes it as an access support because it helps her begin. But the frame supplies the purpose of each paragraph, reminds her where to place the conflict, suggests transitions and prompts a resolution. The target of the task includes independent planning.

Here the label does not settle the educational question. Whatever its history, the frame is currently performing part of the target operation. The tutor should not treat performance with the full frame as directly comparable to performance where planning is self-generated. During teaching, the frame may be entirely appropriate. For evidence about independent planning, the tutor needs either a reduced version or a different task condition, and the change should be explicit.

This is why the Accommodation Baseline must begin with the target. A support can preserve access for one task and carry the target in another. A spelling aid may be irrelevant when assessing argument structure but central when assessing unaided spelling. A calculator may be an ordinary tool in one mathematical task and inappropriate in another where calculation itself is the target. Educational function depends on what is being claimed.

Composite case: Ciara and timing

Fictional composite case. Ciara has an established timing condition for a particular kind of practice. Her tutor gradually shortens the time because the class is approaching a performance phase. Scores fall. The tutor reports that her knowledge is unstable under pressure.

That conclusion may eventually be useful, but the tutor has changed the job. Performance under a shorter time constraint is a changed-condition check. It can answer whether capability remains accessible under greater compression. It should not be merged with the earlier untimed or differently timed trend as though all scores came from the same condition.

For tutoring purposes, keeping two lanes can be more informative: capability under the stable access condition, and performance under a deliberately changed condition. The first asks whether learning is growing. The second asks how robustly that learning survives a different demand. Neither lane is morally superior. They answer different questions.

Do not treat support as statistical noise

A common mistake is to regard every support as an impurity around the learner’s “true ability”. That picture is too simple. Human performance always occurs under conditions: language, tools, instructions, time, layout, environment, prior exposure and social context. The relevant question is not whether conditions exist. It is whether the conditions are appropriate to the construct or capability the tutor is trying to understand.

If a support removes an irrelevant barrier while leaving the target operation with the learner, it can make the evidence more interpretable, not less. If a support supplies the target operation, it changes the evidence claim. The educational task is therefore classification, not purification.

Do not freeze every support forever

The opposite error is to assume that once a support exists, it must never change. Learners change, tasks change and instructional goals change. A support may be reviewed by the appropriate people or may no longer be needed in the same form. A temporary scaffold may be faded. A learner may become able to generate an organising aid independently. The Accommodation Baseline does not prohibit change. It prohibits silent change followed by overconfident comparison.

When a change is deliberate, document the transition. “Previous checks used X. From this session the learner is trialling Y for this educational reason. Treat the next two observations as a new condition until stability is understood.” That sentence is more honest than extending a single line through a condition break.

The two-baseline solution

Sometimes the tutor genuinely needs to understand two conditions. For example, one condition may reflect the learner’s established access support, while another explores whether a temporary instructional scaffold can be reduced. Rather than mixing the results, keep two baselines.

• Access-stable baseline: the target skill with legitimate access support held reasonably constant.
• Changed-condition baseline: the target skill under a deliberately altered support or performance condition, clearly labelled.

The goal is not to create double bureaucracy. Often a brief note is enough. The benefit is conceptual: improvement in one lane no longer has to be interpreted as improvement in the other. A learner may demonstrate strong conceptual growth under a stable support while still finding a reduced-scaffold condition difficult. That is useful information for teaching.

The difference between fading and withdrawing access

Fading is appropriate when the support is temporary and part of the target operation is meant to move inside the learner. Withdrawing access is different. If a support exists to make the task perceptible, physically usable or communicatively accessible without doing the target thinking, its disappearance may not represent educational progress at all.

The Scaffold Fading State already states the crucial boundary: legitimate accessibility support should not be faded merely to prove independence. The Accommodation Baseline extends that point into progress interpretation. Keep the access condition visible in the evidence record so a future tutor does not accidentally “test independence” by deleting part of the learner’s fair access route.

A three-student group makes condition tracking more important

In a three-learner tutorial, one student may use an access support that the others do not. The tutor must resist the temptation to treat uniformity as fairness. Same task, same time, same representation and same tool can look neat while creating different irrelevant barriers. At the same time, differential support must not become invisible. The tutor should know which learner evidence was produced under which condition.

Peer interaction adds another complication. One learner’s oral explanation may become an answer cue for another. A support tool may be shared unintentionally. The tutor does not need a laboratory protocol; they need orderly first attempts. Where independent evidence matters, each learner should have an opportunity to respond before peer discussion or answer-revealing assistance changes the condition.

What counts as “the same condition”?

Exact sameness is impossible and unnecessary. Learners are not machines and lessons are not repeated trials. The tutor is looking for material comparability, not perfect replication. Ask which condition differences are likely to alter interpretation of the target capability.

A different pencil probably does not matter for most algebra decisions. A different response mode might. A new room may not matter; severe interruption might. Different font size may be immaterial for one learner and central for another. A timing change may be irrelevant to an untimed knowledge check and central to a performance claim. The significance of the condition depends on the learner, task and target.

This is why accommodation-sensitive evidence cannot be reduced to a universal list. The tutor records only those conditions that plausibly matter to the educational interpretation.

A condition ledger that stays small

• Target: What capability was being checked?
• Access support: What educational support was available?
• Instructional scaffold: What temporary teaching support was available?
• Answer assistance: Did any cue carry the target operation?
• Performance condition: Was time, format or setting materially different?
• Result: What did the learner actually do?
• Claim: What does this result support—and what remains uncertain?

For most tutoring, this can be a few words in a lesson note, not a formal form. The condition ledger protects interpretation across handoffs and delayed reviews.

Research and policy boundary: AERO

The Australian Education Research Organisation’s practice guide on supporting diverse physical needs, published in 2025 and updated in July 2025, discusses reasonable adjustments to learning and assessment, environmental changes and adaptive equipment as ways of supporting access. It is Australian school guidance, not authority for Singapore tuition or examination arrangements. Its relevance here is conceptual: access changes can be legitimate features of educational participation rather than temporary hints that must always disappear.

AERO’s Monitor progress guidance also supports regular checks of what students know and can do so teaching can be adjusted. It does not prescribe the Accommodation Baseline proposed here. The article combines monitoring principles with a validity concern: if material access conditions change between checks, the tutor should make that condition change visible before drawing strong progress conclusions.

Research boundary: accommodations and validity

The US National Center on Educational Outcomes summarises accommodation research using a validity lens: an accommodation is intended to change materials or procedures in ways that improve access without compromising the interpretation of the target construct. Its accommodations review and later research syntheses show that effects vary by accommodation, population, subject and implementation. Evidence is not uniform enough to justify saying that a particular support is universally neutral or universally beneficial.

A 2023 NCEO research synthesis, available at Summary of Research on the Effects of Test Accommodations, likewise reports a mixed evidence landscape and highlights implementation. Older Institute of Education Sciences work on oral accommodation illustrates why validity questions matter: a support can alter performance in ways that differ across learners, so one cannot assume that every changed administration preserves the same interpretation.

These sources concern formal assessment and specific populations. They should not be imported mechanically into private tuition. They support a general caution: accommodations are part of the measurement condition, and their effects must be considered rather than ignored. They do not provide a formula for deciding any individual learner’s entitlement or diagnosis.

Do not convert an accommodation into a label for the learner

A support condition describes an educational arrangement, not the learner’s identity. Tutors should avoid language such as “she is an extra-time student” or “he is a reader-support child”. The relevant sentence is functional: “On this task, this support was available; it did not supply the target answer; performance should be interpreted under that condition.”

This keeps the learner larger than the accommodation and keeps the tutor focused on the actual capability being taught. It also makes it easier to revise the route later if the support condition legitimately changes.

Parent communication: what changed, not what is wrong

Parents need clear, bounded language. A useful update might say: “We are seeing stronger inference under the same text-access condition used across recent sessions. Today I also tried a different format and performance dropped, so I am treating that as a changed-condition observation rather than saying comprehension regressed.”

That sentence protects the learner from a false deficit story and protects the tutor from overclaiming. It also tells the parent what will happen next: maintain the stable access condition for the main learning trend, and investigate the changed condition only if it serves a legitimate educational purpose.

When the support itself may need review

A tutor may notice that a support appears to be carrying more of the target task than intended. The tutor should not unilaterally revoke a formal arrangement. They can separate the tutoring evidence and raise the educational question through the appropriate route. For a temporary tutor-created scaffold, the tutor can test a smaller version. For a formal or externally established access arrangement, the tutor should respect the governing process and avoid practising beyond their role.

The educational note can stay observational: “With support X, the learner generated the explanation independently. Without the additional tutor-created prompt Y, method selection was not yet stable.” This separates legitimate access from the extra instructional assistance that can be faded.

The delayed comparison

One good session is not a trend. After teaching, return under the stable access condition on a changed task. If the learner can still perform the target capability, confidence in the learning claim rises. If the tutor then tests a reduced instructional scaffold, keep that result in the changed-condition lane. Over time the record can show two meaningful developments: capability strengthening under fair access, and temporary scaffolds becoming less necessary.

That is a much richer story than “support went down, therefore independence went up”. Sometimes true independence means using an appropriate tool without adult management. Sometimes it means generating a strategy rather than waiting for a prompt. Sometimes it means maintaining performance under the access conditions the learner is legitimately expected to use. The target decides.

Where the Tutor Classification Model matters

The Tutor Classification Model helps only where function changes interpretation. A Class 3 Diagnostic Tutor needs to know which barriers are part of the task and which are irrelevant to the target. A Class 4 Route Designer may need to coordinate access conditions across tasks so progress evidence remains interpretable. A Class 5 Performance Coach must distinguish realistic performance conditions from artificial deprivation of support. A Class 6 Learning Architect may need to keep school, home and tuition from silently using different support assumptions. These are functions, not human rankings.

Repair, Alignment and Frontier remain the only tuition modes in this architecture. Accommodation-sensitive work can occur in any of them. Repair may require access before a weak link can be seen cleanly. Alignment may require school and tuition conditions to stop contradicting each other. Frontier work may increase intellectual difficulty while holding legitimate access steady. “Performance” does not need to become a fourth mode.

A compact Accommodation Baseline protocol

• Define the exact target capability.
• Identify the established access support relevant to the task.
• Separate that support from temporary instructional scaffolds.
• Check whether any support carries the target operation.
• Record only the condition details needed for educational interpretation.
• Keep the main progress checks reasonably comparable.
• If a condition changes, mark the break rather than extending the same trend silently.
• Use a fresh task after delay before claiming stable learning.
• Do not remove legitimate access merely to prove independence.
• Do not treat tuition observations as a clinical diagnosis or formal entitlement decision.

Failure mode: the tutor makes the test “fair” by making it easier

Fair access is not the same as lower demand. A tutor can accidentally remove the target while trying to be supportive. If a comprehension task is meant to test inferential reasoning, giving the inference in simpler words is not merely access. If a Mathematics task is meant to test method selection, restricting the learner to one named method changes the demand. If a Science task is meant to test retrieval of a concept, supplying the concept is teaching.

The Accommodation Baseline therefore protects rigour as well as accessibility. Preserve the target. Remove or reduce barriers that are not the target where appropriate. Keep the evidence claim proportional to the support actually present.

Failure mode: the tutor makes the test “real” by removing access

The mirror-image mistake is to decide that authentic performance means no support at all. Real educational environments always contain tools, conventions and permitted conditions. Authenticity means matching the intended context, not maximising deprivation. If the learner is ordinarily permitted or appropriately provided a legitimate access support, removing it may make the task less representative of the condition that matters.

Where the future performance environment is uncertain or governed by another institution, the tutor should not promise what will be permitted. Practise the knowledge and capability robustly, respect current authorised conditions, and verify formal rules from the relevant authority when a specific high-stakes claim is needed.

The final return

A fair progress story needs a stable question: what changed in the learner, and what changed in the conditions around the learner? Those two changes cannot always be separated perfectly. They can at least be named.

The Accommodation Baseline prevents a quiet form of measurement drift. The learner uses one access condition in September, another in October, a reduced one in November, and then the tutor draws a single progress line as though every point meant the same thing. That line may look scientific while hiding its largest interpretive change.

Keep legitimate access visible. Keep scaffolding visible. Keep answer assistance visible. Change conditions when there is a reason, not because independence has been confused with unsupported performance. Then let the learner’s capability be judged against the target under conditions that make the result worth interpreting.

Sources and evidence boundary

• Australian Education Research Organisation — Physical needs: supporting diverse needs: Australian school practice guidance on reasonable adjustments, access and participation.
• AERO — Monitor progress: current research-informed guidance on using evidence of student learning to adjust teaching.
• National Center on Educational Outcomes — accommodations research review: formal assessment research focused on accessibility and validity.
• NCEO — Summary of Research on the Effects of Test Accommodations, 2023: synthesis showing that implementation and effects vary across accommodation types and populations.
• Institute of Education Sciences — Validity of Oral Accommodation in Testing: older working paper illustrating that changed administration conditions can affect score interpretation.
• Stanford National Student Support Accelerator — Tutoring Quality Standards: tutoring evidence framework relevant to instructional practice, formative assessment, student progress and safety.

The Accommodation Baseline is an educational reasoning framework, not a validated assessment, clinical protocol, legal opinion or official accommodations policy. Much of the cited research concerns school systems and formal assessments outside Singapore. It informs the validity principles used here but cannot determine what any individual learner should receive in a school or national examination. All worked cases are fictional composites.