An ABA most-to-least prompting worksheet should show the support scheduled and actually delivered on each opportunity, when that support was faded, and whether responding later occurred under the natural cue. High prompted accuracy can coexist with little natural-cue control, so one completion percentage is not enough.

Use this copyable ABA most to least prompting worksheet to review a case-specific arrangement while preserving access, assent or dissent, invalid opportunities, burden, and fidelity. The fictional example does not prescribe a starting prompt, fading step, or threshold.

Clinicians & ABA Professionals / Assessment and Treatment Planning.

Clinical, access, and contact boundary: Starting with substantial support does not authorize unnecessary touch, forced compliance, restraint, blocking, or removal of communication and sensory access. A qualified team must determine the target’s relevance, the assessed controlling support, competence and scope, consent requirements, assent or dissent, medical and motor safety, setting rules, and alternatives. Honor help, break, stop, refusal, pain, fatigue, freezing, pulling away, distress, and safety concerns. Current BACB ethics resources provide professional context; this worksheet is not authorization or a treatment plan.

Most-to-least begins with assessed support and a plan to transfer control

In a most-to-least arrangement, teaching begins with a support expected to produce the response, then systematically reduces that support across opportunities. The aim is not permanent prompted accuracy. The review asks whether responding transfers to less support and ultimately to the accessible natural cue.

“Most” and “least” are meaningful only within the written, case-specific arrangement. Verbal, visual, gestural, model, positional, and physical prompts cannot be ranked universally. A spoken instruction may create more dependence or burden than a familiar visual aid; a visual schedule or AAC display may be ordinary access that should remain.

The BACB test-content outlines and Sixth Edition BCBA Test Content Outline cover response prompting, fading, stimulus-control transfer, assessment, measurement, maintenance, and generalization. They do not prescribe a particular controlling prompt, fading direction, or mastery rule.

Purpose, assessment, and authorization card

Define the meaningful outcome before selecting the support. A controlling prompt should be assessed for this person and skill, not assumed from a staff label or inherited program.

FieldCase-specific entryPerson’s meaningful goal and description of the outcome. Describe.Participation in selecting, changing, or declining the target. Explain.Authorized assessment, treatment plan, and current version. Identify.Natural setting, cue, response, and consequence. Specify.Accepted accessible response forms. List.Baseline performance under the natural cue. Summarize.Evidence used to identify the planned starting support. Cite.Alternatives considered, including task or environmental changes. Compare.Communication, AAC, sensory, literacy, language, motor, and mobility access. Preserve.Consent requirements and how assent or dissent will be checked. Clarify.Contact allowed, prohibited, or requiring separate review. Bound.Qualified owner, interdisciplinary reviewers, dates, and version. Name.

Do not infer medical, motor, feeding, swallowing, communication, or accessibility safety from task completion. The Council of Autism Service Providers ABA Practice Guidelines provide broad service context, not a substitute for individualized assessment or another profession’s scope.

Define the natural cue and response first

The final cue and accessible response must be explicit. Otherwise, the team cannot tell whether prompt fading transfers control or simply changes how staff help.

ElementOperational definitionExampleExclusion or invalid conditionMeaningful natural cue. Define.Observable target response. Define.Accepted accessible alternative. Include.Initial response window. Time.Correct response. Distinguish.Incorrect response. Describe.No response/other. Bound.Valid opportunity. Define.Invalid opportunity. Define.Natural or programmed consequence. Specify.

Avoid defining success as eye contact, stillness, speech, posture, or speed unless that feature is necessary to the meaningful outcome and justified. A task completed after a prompt is correct under that level; it is not automatically independent or under natural-cue control.

Describe each support level as delivered

Record reproducible topographies rather than “full,” “partial,” “medium,” or “minimal.” If any level includes contact, specify the authorized body site, duration and force limits, release cue, and immediate stop rule in the governing plan.

LevelExact support and materialsWhat remains natural or availableScheduled useWithin-opportunity responseFade direction/stepContact or access boundaryLevel 3 / assessed starting supportLevel 2Level 1Natural cue / no programmed prompt

The numerical labels do not imply universal intrusiveness. Version the arrangement whenever topography, materials, natural cue, response, wait interval, or fading rule changes.

The ASHA AAC Practice Portal describes AAC as individualized and multimodal. Do not withdraw a communication system to create a “less prompted” condition. A device, visual scene, partner strategy, or aided language input may be ordinary access rather than a prompt slated for removal.

Prewrite advance, hold, rollback, and correction rules

Fading should follow prospective evidence and authorized decision rules. A calendar date, staff intuition, or a run of prompted correctness is not a sufficient rule by itself.

DecisionProspective ruleEvidence requiredExperience/access checkOwner and dateBegin at the assessed starting level.Advance to less support.Hold the current level.Roll back to more support.Present a natural-cue test.Correct an error within the opportunity.Reassess the support arrangement.Pause or stop.

A rollback can reflect an oversized fading step, an inaccessible cue, unfamiliar materials, poor integrity, pain, fatigue, or a changed context rather than learner failure. Document the hypothesis as a question and collect the evidence needed to evaluate it.

Keep scheduled and delivered levels separate

One row represents one planned opportunity. Record the scheduled level before the trial, then the level actually delivered, the initial response, any within-opportunity change, and the natural-cue evidence.

Date/timeTarget/contextVersionScheduled levelDelivered levelNatural cue present/accessibleValid? reasonInitial correct/incorrect/no responseWithin-opportunity added or reduced supportTerminal outcomeLatencyHelp/break/stop or dissentBurden/accessConsequence/correctionImplementerL3 / L2 / L1 / natural cueyes / no

If the scheduled and delivered levels differ, preserve both. An early prompt, wrong material, missing AAC, unavailable natural cue, unauthorized contact, interruption, or safety stop may make the opportunity invalid under the written rule. Retain the row and reason instead of counting it as learner error.

Level summary and natural-cue view

Scheduled levelPlannedInvalidValidInitial correctInitial incorrectNo response/otherDifferent level deliveredHelp/break/stopNotesLevel 3Level 2Level 1Natural cueTotal

Report the natural-cue row separately. Pooling strong responding at Level 3 with weaker responding under the natural cue can make a program look successful while obscuring incomplete transfer. Likewise, a decline in prompted accuracy after a fading step does not identify the cause.

Separate the denominators

Review questionNumeratorDenominatorResultDoes not establishValid coveragevalid opportunitiesall planned opportunitieslearner accuracyOverall initial correctnessinitial correct across levelsall valid opportunitiesnatural-cue controlLevel-specific correctnessinitial correct at one scheduled levelvalid opportunities at that levelperformance at another levelNatural-cue correctnessinitial correct under natural cuevalid natural-cue opportunitiesmaintenance, generalization, or causationPlanned correct yieldinitial correct across levelsall planned opportunitieswhy opportunities were invalidInvalid shareinvalid opportunitiesall planned opportunitieslearner error rateProcedure fidelitycorrectly implemented componentsobservable component opportunitiesclinical appropriateness or benefit

Carry counts beside percentages. Eight of eight at a substantial prompt level and six of eight under the natural cue answer different questions. Sequential level changes are not randomized comparisons; exposure, target difficulty, implementer, setting, and motivation may differ.

Fictional adaptive cooking-club setup example

Jordan, a fictional participant, chose an adaptive community cooking-club setup routine. The team defined the natural visual cue, accepted motor and AAC responses, ordinary adaptive equipment, an assessed starting support, three case-specific support levels, fading and rollback rules, and accessible help, break, stop, and invalidity criteria. This example tests arithmetic only.

The team planned 40 opportunities. Five were invalid: one used the wrong visual sequence, one lacked the adapted grip, one began without the AAC device available, one included an undocumented early support change, and one ended during an environmental interruption. There were 40 − 5 = 35 valid opportunities, or 35 / 40 × 100 = 87.5% valid coverage.

Eight valid opportunities were scheduled at Level 3, and all eight were initially correct. Ten were scheduled at Level 2, with nine correct. Nine were scheduled at Level 1, with eight correct. Eight were presented under the natural cue, with six correct.

Scheduled levelValidInitial correctIncorrect/no responseLevel-specific resultLevel 3880100.0% under the defined Level 3 supportLevel 2109190.0% under Level 2Level 198188.9% under Level 1Natural cue8626 / 8 × 100 = 75.0% under the natural cueTotal3531431 / 35 × 100 = 88.6% across unlike levels

Overall planned correct yield was 31 / 40 × 100 = 77.5%. The five invalid opportunities were 5 / 40 × 100 = 12.5% of planned delivery. The invalid share is not a learner error rate.

The 88.6% overall result does not mean 88.6% natural-cue control; the natural-cue evidence is six of eight valid opportunities. Nor do the eight correct Level 3 responses prove the support was necessary, acceptable, or safe. Review target mix, prior exposure, latency, errors, burden, access, and follow-up before changing the arrangement.

Fidelity record for fading and level delivery

ObservationCorrect versionScheduled level documentedNatural cue accessibleDelivered level exactInitial response coded firstFade/hold/rollback rule followedCorrection rule followedAAC/access preservedHelp/stop honoredContact within planCorrect componentsOpportunitiesNotes

In the fictional review, 15 observations across 10 components created 15 × 10 = 150 component opportunities. Reviewers scored 141 components correct and nine mismatches; 141 + 9 = 150. Descriptive fidelity was 141 / 150 × 100 = 94.0%.

Do not let the total erase a serious mismatch. An inaccessible natural cue changes the question. Unplanned contact, missing AAC, or failure to honor stop communication requires specific review. Fidelity describes whether the plan occurred; it does not establish assent, appropriateness, learning, or generalization.

Person experience and burden record

Review dateTarget/contextCommunication accessPerson’s report or observable communicationPrompt level experienced as useful, neutral, unwanted, or unclearHelp/break/stop honoredLatency/effortErrors or distressRequested changeFollow-up owner

Ask accessibly and allow delayed, nonvocal, device-based, or partner-supported communication. Correct responding and absence of resistance do not establish assent. Do not make food, water, bathroom access, communication, mobility, relationships, health care, safety, rest, or emergency help contingent on performance.

What the research supports, and what it does not

An early comparison of most-to-least and least-to-most prompting found procedure-specific differences in a small, defined teaching context. Research involving positional prompts and most-to-least fading and imitation teaching with a most-to-least hierarchy likewise demonstrates concrete arrangements, not universal orders.

A comparison of prompt-fading strategies and work on learner-specific prompt assessment support evaluating the actual person, target, and implementation rather than assuming one procedure is best. A study of receptive vocabulary taught with least-to-most and most-to-least prompting reinforces that results must stay attached to the procedure and population studied.

A conceptual review of gradual-change procedures helps clarify why small procedural details and measurement matter. It does not set a fading step for an individual. A small study of parent preference among prompting procedures shows that acceptability is worth measuring. Direct client input, individual evaluation, and applicable consent remain separate evidence.

The Ethics Code for Behavior Analysts addresses rights, competence, informed consent, assent when applicable, documentation, confidentiality, and data use. These sources support cautious, qualified review. They do not authorize contact, define a mastery threshold, or establish expected effectiveness.

Maintenance, generalization, and final decision record

Review fieldEntryTarget, natural cue, response, support levels, and versions reviewed.Planned, invalid, and valid opportunities by scheduled and delivered level.Initial correct, incorrect, and no-response counts by level.Natural-cue performance with raw count and denominator.Person’s communication, assent/dissent, burden, and requested changes.Access, health, sensory, motor, and interdisciplinary considerations.Fidelity mismatches and interpretive limits.Maintenance interval and intervening teaching exposure.Generalization people, settings, materials, and natural cues.Retain / advance / hold / roll back / redesign / pause / stop question.Qualified owner, rationale, effective date, and next review.

Reopen the plan when prompted accuracy remains high but natural-cue performance does not improve, errors cluster after one fading step, staff deliver different levels, burden rises, access changes, the person asks for a change, or maintenance and generalization do not support the meaningful outcome. The team may adjust materials or step size, reassess the controlling support, change teaching arrangements, seek interdisciplinary input, pause, or stop.

Preserve the prior version and its evidence. A new starting level, prompt topography, or fading rule creates a new condition rather than rewriting earlier data.

Related resources

Sources

The sources vary in participant, skill, setting, prompt topography, sequence, design, and outcome. They support explicit levels, case-specific assessment, natural-cue tests, direct measurement, acceptability, fidelity, and bounded interpretation. They do not define a universal controlling prompt, hierarchy, contact permission, fading step, advancement or rollback threshold, trial count, mastery criterion, expected effect, or clinical decision.