To separate observation, report, and inference in an ABA plan, label the evidence class and source before describing its clinical meaning. Direct observation records what the observer defined and measured. Client, caregiver, staff, and record reports retain their authorship and context. Interpretation, hypothesis, and recommendation identify the qualified author, evidence considered, uncertainty, and decision scope. One class does not automatically outrank another.

Define the evidence classes

Use direct observation, client report, proxy report, record review, instrument result, implementation evidence, interpretation, hypothesis, recommendation, and decision as distinct labels.

Create the labels in the record template instead of relying on prose style. “Ravi left the room during 3 of 5 eligible transitions” is direct observation when the definitions and conditions are documented. “Ravi said the room was too loud” is client report. “The clinician hypothesizes noise affected departure” is an interpretation or hypothesis. None should silently replace another.

Keep implementation evidence separate as well. A procedure delivered in 2 of 5 opportunities changes how the observed outcome can be interpreted. The eventual recommendation and team decision each receive their own qualified author, date, and rationale.

Preserve the source person's words

Record Ravi's communication and the caregiver's report separately, with question, response mode, date, period, setting, support, uncertainty, and any later correction.

Use the reporting person's words when practical or a faithful attributed summary. Note whether Ravi used speech, AAC, writing, gesture, or another form, who was present, what question was asked, and whether he had privacy and adequate time. Preserve “unsure,” mixed responses, and declined questions.

The caregiver may add observations across a longer period, but their report remains a separate source. Do not average perspectives or rewrite the caregiver's interpretation as Ravi's view. When either person corrects an entry, add the new version and keep the earlier one traceable.

Operationalize direct observation

Name the response, eligible opportunities, observation window, observer, setting, ordinary supports, prompts, access, exclusions, and numerator and denominator.

Write the definition so another trained observer could score the same event. Include the start and end of an opportunity, how partial responses and interruptions are handled, which supports are ordinary, and why any event is excluded. Preserve unavailable conditions as system evidence rather than erasing them.

Report raw counts with percentages. A 60% result could mean 3 of 5 or 60 of 100 and carries different uncertainty. Identify whether the observation period was representative and whether observer agreement or implementation evidence was sampled.

Attribute interpretations

State who interpreted the evidence, qualifications and role, records considered, competing explanations, confidence, missing evidence, and what the interpretation may guide.

An interpretation should name the specific facts it joins and the alternative explanations considered. For example, departure may relate to sound, denied access, illness, partner behavior, or an unobserved event. State the author's confidence and the observation or referral needed to narrow the question.

Stay within professional authority. Medical symptoms, hearing, pain, trauma, legal issues, communication assessment, or environmental access may require other qualified reviewers. The ABA interpretation can guide a cautious next step without declaring their conclusions.

Avoid a false hierarchy

A short observation may miss daily context; a report may have recall limits; a record may be stale. Compare each source's relevance and quality for the decision. Prestige alone adds no evidentiary weight.

Evaluate proximity to the question, collection quality, recency, setting match, access conditions, and known bias. Ravi's direct report is central to his experience, while a trained observation may answer a narrow frequency question. A physician record may establish a medical instruction but may not describe current classroom implementation.

Conflicting sources need reconciliation, not a blanket rank order. State what each can support, where they diverge, and which qualified person owns the decision. Additional evidence should be collected only when it will change that decision and can be obtained without unreasonable burden.

Carry classes into review

Graph or summarize like evidence together, preserve discrepancies, and show when new observation, direct client input, medical review, or another assessment is needed.

Do not graph proxy estimates beside direct counts as though they share a unit. Use separate panels or narrative labels and connect them through an interpretation field. Carry source IDs and versions into reviews so a later reader can find the underlying record.

When a decision changes, show which evidence class moved it and which gap remains. If access conditions change, treat the new observation as a new context rather than silently merging it with older data. Review discrepancies with Ravi in an accessible form.

Build Ravi's evidence-class matrix

Create one versioned evidence-class matrix for the community assessment. Include statement or evidence ID, person and source, author, method, setting, collection window, as-of date, version, access conditions, missingness, uncertainty, interpretation, qualified decision, plan component, client response, privacy route, correction history, owner, due date, and review trigger. A second qualified reviewer should be able to reconstruct how the evidence entered the plan and what it changed.

Work through Ravi's example

Ravi's review includes four source classes. Direct observation shows 3 help messages in 10 opportunities; Ravi reports that the setting is too noisy; a caregiver reports two difficult visits; and the clinician hypothesizes that communication access and noise merit testing. The plan keeps all four entries separate and labels the hypothesis provisional. Preserve every statement, source class, person, numerator, denominator, exclusion, missing item, correction, and decision link. This fictional example illustrates one provenance control. It supplies no universal clinical threshold, medical conclusion, privacy permission, payer result, or outcome guarantee.

Address Ravi's main provenance risk

Writing a report as an observed fact changes its authority. Ravi's matrix preserves who knew what, how they knew it, and where it applied. A source can be accurate for one period or decision and unfit for another. Keep original evidence, interpretation, recommendation, consent, operational release, payer state, claim, and outcome as separate records.

Choose Ravi's next evidence step

The team measures noise and communication access in representative visits while continuing to ask Ravi directly about comfort and priorities. Record the qualified owner, authority, person and scope affected, interim protection, evidence needed, access and privacy route, due date, correction method, recipients, closure state, and next review. Software may coordinate workflow while qualified people interpret evidence and decide within scope.

Apply professional evidence standards to Ravi's record

For Ravi's evidence record, the BACB ethics hub identifies the current Ethics Code, which addresses competence, client and stakeholder involvement, assessment, medical needs, documentation, confidentiality, risk, and data-based evaluation for covered people. BACB has no separate organizational jurisdiction. The BCBA outline provides examination content and carries no practice authority. The CASP public summary supplies high-level planning context for ABA treatment of autistic people. An evidence-based ABA framework supports integrating research, clinical expertise, client values, and context.

Use integrity and access sources carefully for Ravi

In Ravi's case, the treatment-integrity practitioner guide, Essig review, impact study, and reporting review support explicit procedures, implementation evidence, observer quality, and cautious interpretation. They create no universal provenance threshold. ASHA supports continuous AAC access. For a HIPAA covered entity, HHS minimum-necessary guidance generally applies to uses, disclosures, and requests for protected health information, with rule-specific exceptions. It supports role-based access and focused plan fields.

Close Ravi's evidence review

Review the evidence-class matrix with Ravi, the responsible clinician, affected participants, and the specialists named in the manifest. Preserve direct client communication, source attribution, disagreements, versions, limitations, decisions, corrections, and open gaps. Keep this page draft and noindex until the required clinical, client or family, measurement, AAC, access, medical, records, privacy, ethics, research, and legal reviews are complete.

Related resources

Sources