How Should Prompted and Independent ABA Responses Be Reported?

Define prompted versus independent responses

An independent response occurs within the defined window after the natural cue and with only the ordinary supports named in the plan. A prompted response follows added teaching help. The definitions should name the cue, response topography or outcome, response window, ordinary supports, and each prompt level before staff collect data.

Ordinary support and teaching prompt are not synonyms. A person's usual AAC, glasses, mobility aid, recipe, timer, visual schedule, or adaptive tool may remain available during independent performance. A therapist's gesture, model, verbal hint, positional cue, or physical guidance may be an added prompt. The plan should reflect the real daily condition rather than remove useful supports to create a stricter score.

Requested help deserves its own category. A person may independently identify a problem and request a hint, model, or partner-completed step. That is different from responding independently on the task step and different from receiving an unrequested prompt. Step-level data can show both help-seeking and authorship accurately.

Record partial responses and self-corrections according to prewritten rules instead of deciding after seeing the result. If a response begins before the prompt but finishes after it, the plan should say how it is classified. If staff prompt before the response window ends, mark the early prompt and avoid crediting an independent opportunity that never occurred.

Build mutually exclusive categories

Every eligible opportunity should enter one response category once. A practical set may include:

• Correct independent with named ordinary supports.
• Correct after an independently requested help type.
• Correct after an unrequested prompt, separated by prompt level.
• Partial response or self-correction under the written definition.
• Error.
• No response after the full window.
• Withdrawal or declined opportunity.
• Invalid because the planned cue, material, access, safety, or partner condition failed.

Mutually exclusive categories keep totals reconstructable. A response should not count as both independent and prompted merely because it ends correctly. If a report also uses broader labels such as successful participation, show exactly which categories are combined.

Lock the opportunity cohort

Declare which events are scheduled and which become eligible before review. Locking the cohort prevents difficult responses from disappearing after data collection. The report can show scheduled, presented, eligible, invalid, withdrawn, and unpresented counts before calculating response percentages.

Invalid means the planned opportunity was unavailable for a reason outside the learner's defined responsibility. Examples include a missing material that staff were supposed to prepare, unavailable AAC, wrong cue, device outage, unsafe condition, partner interruption, or prompt delivered so early that the independent window was lost. Preserve the reason and staff responsibility.

Withdrawal remains visible but may or may not enter a particular learning denominator under the prewritten rule. The key is transparency. A family should be able to see how many opportunities were planned, which ones the person actually encountered, and exactly what was excluded.

Keep raw counts beside percentages. Six of eight and sixty of eighty both equal 75 percent while offering different amounts of evidence. Small denominators can shift sharply after one response, so the count matters for interpretation.

Avoid disappearing partner errors

A prompt delivered too early, a missing AAC device, an unavailable material, or a partner who fails to provide the planned outcome may make the event invalid or reveal an implementation error. Report it separately. If preparing the material is a scored staff duty, it stays in the staff-fidelity denominator.

Fidelity data should name the procedure steps that were due. A whole-trial fidelity score can hide whether staff repeatedly missed the response window, chose the wrong prompt, failed to reinforce, or ignored a help request. Step-level fidelity tells the supervisor what needs coaching.

Versions also matter. Changing the prompt hierarchy, wait time, cue, materials, response definition, or ordinary supports creates a different condition. Mark the date and avoid silently pooling the old and new procedures. A sudden improvement might reflect the revision rather than a change under the original plan.

Separate data by staff, setting, item type, or time when those differences matter to the decision. A pattern across people may reveal inconsistent prompting. A pattern with one material may reveal an access issue. These comparisons guide observation but do not establish cause on their own.

Review the measures in a useful order

Families can ask the team to:

1. Confirm definitions, ordinary supports, response window, and procedure version.
2. Account for every scheduled event as presented, unpresented, eligible, invalid, withdrawn, or stopped.
3. Review first independent responses before looking at prompted correctness.
4. Break prompted responses down by level and timing.
5. Review requested help and whether the partner provided the promised support.
6. Check staff fidelity, invalid-event reasons, latency, duration, and client feedback.
7. Examine later independent probes, maintenance, and relevant new examples separately.
8. Apply the written rule for continuing, fading, retraining, redesigning, or pausing.

High prompted accuracy can show that a prompt supports performance during teaching. It does not establish independent responding, maintenance, generalization, or client comfort. Low independent accuracy with high fidelity may call for a prompt or teaching review. Low fidelity means the team has limited evidence about the written procedure because it was not delivered consistently.

Burden belongs in the review. A final correct response after several prompts, long latency, or visible discomfort is different from a quick response after one accepted hint. Duration, repetitions, help requests, withdrawal, and the person's accessible report keep the path to the response visible.

Use source concepts narrowly

The ABAI basic-principles curriculum and BACB outline cover measurement, prompting, fading, stimulus control, integrity, and evaluation concepts. The prompting decision article highlights prompt selection, implementation, efficiency, and transfer concerns.

The Libby comparison examined two prompt arrangements in a narrow play-skill context. The Ethics Code addresses documentation, consent and assent when applicable, client involvement, risk, and continual evaluation. These sources define no universal data sheet, prompt hierarchy, independence rule, or acceptable percentage.

The ASHA AAC guidance supports ongoing communication access. An AAC system should not be removed to test independence unless an appropriately reviewed goal truly concerns performance without that system. Device or access failures need their own record.

A practical example

Hana has 16 scheduled library-checkout opportunities across four visits. One is not presented because the library closes early, and one is invalid because the checkout screen is offline. Of the 14 eligible opportunities, she responds independently in seven, follows a visual prompt in three, requests help in two, makes an error in one, and withdraws in one.

The report shows 7 of 14 independent, 3 of 14 prompted, 2 of 14 independently requested help, 1 of 14 error, and 1 of 14 withdrawal. It separately shows one unpresented event and one invalid technology event out of 16 scheduled. Combining the first three response categories as 12 of 14 “successful” may be useful for a participation question, but it would hide materially different performance unless the component counts remain beside it.

Staff deliver the correct visual prompt on time in two of the three prompted opportunities and prompt early once. Prompt-timing fidelity is therefore 2 of 3 prompts due under that review definition. In four later no-prompt probes with the ordinary library card and AAC available, Hana checks out independently three times and requests help once. Those probes have their own 4-opportunity denominator.

The supervisor reviews the early prompt, the technology failure, Hana's two help requests, and the later probe result before deciding whether to fade. The data do not prove which procedure caused the three independent probes.

A family reporting checklist

• Obtain written definitions for the natural cue, response window, independence, ordinary support, each prompt, requested help, error, withdrawal, and invalid event.
• Account for scheduled, presented, eligible, invalid, withdrawn, and unpresented opportunities.
• Keep independent, requested-help, prompted-by-level, partial, error, and no-response counts separate.
• Show raw numerators and denominators beside percentages.
• Mark early prompts and opportunities where the independent window was lost.
• Track partner response and staff fidelity by step.
• Label procedure, material, and support version changes.
• Review latency, duration, repetitions, burden, and client feedback.
• Keep later probes, maintenance, and generalization samples separate.
• Set the next decision, responsible clinician, and review date.

Questions families can use

What enters the denominator? Which supports count as ordinary? How is prompt timing recorded? Are requested help and withdrawal visible? Which partner or material failures are reported? Can the raw counts be reconstructed? Which result will lead to fading, retraining, redesign, or a pause?

Limits of this guidance

Response categories depend on the target, task, communication method, motor access, ordinary supports, and purpose of measurement. A percentage cannot establish why performance changed, whether the skill generalizes, or whether prompting was acceptable. A qualified clinician should define and interpret the record with the client and family and involve AAC, motor, medical, trauma-informed, school, safety, privacy, or legal specialists when applicable.