Records Review and Intake Hypotheses

Records Review in Domain F

Behavior Assessment is Domain F of the BACB Test Content Outline (TCO) and carries roughly 13% of the 185-question exam (about 24 scored items). The assessment process is a funnel: it starts wide with indirect information and narrows toward direct, controlled data. Records review is the first, widest step.

Records may include referral forms, prior assessments, behavior-support plans, progress graphs, medical and psychiatric notes, IEPs or service plans, incident reports, interviews, and treatment histories. They are inexpensive, fast, and already on hand — which is exactly why they are reviewed first and why over-trusting them is the first error to avoid.

The central exam principle: records are not the assessment result. They tell you what to ask, where to observe, which safety issues to plan for, and which data systems already exist. A prior plan that says behavior is 'attention-maintained' is a hypothesis to test, not a finding to adopt.

Think of records review as setting the agenda for direct assessment. A good review answers four questions: What is the referral concern in measurable terms? What is already known, and how reliable is it? What safety risks must I plan around? And which data are missing that only direct observation can supply? Every later method — interview, skill probe, descriptive observation, functional analysis — is chosen to close one of those gaps.

What to Extract From Each Source

Review is purposeful, not a transcription exercise. Each source answers a different assessment question, and a skilled reviewer reads for patterns, risk, and gaps rather than copying narrative.

Notice that medical notes can reveal setting events (also called establishing operations for problem behavior) such as poor sleep, illness, constipation, or a medication change. These do not occur at the moment of behavior but raise its overall likelihood, and missing them is a classic assessment error the exam tests.

Reading graphs and incident reports critically

Prior graphs are gold when they show a real baseline and a clear response (or non-response) to past intervention. Read them for level (where the data sit), trend (slope/direction), and variability (bounce). A flat high trend across a previous plan tells you that plan failed; a sharp drop that later rebounds suggests the contingency was not maintained. Do not assume a label on the graph (the prior 'function') is correct — read the data, not the title.

Incident reports are often the richest source of topography and risk because they are written close to the event. Extract: the behavior's form, the time and place, who was present, what was happening just before (antecedent), what staff did after (consequence), and any injury or property damage. Patterns across many incidents — same class period, same staff member, same transition — are scatter you can later confirm with a formal scatterplot. The risk is recall and reporter bias: incident reports are written by people with a stake in the narrative, so treat their stated causes as hypotheses too.

Building Intake Hypotheses

Records feed an intake hypothesis — a tentative, testable statement linking behavior to environmental events. The hypothesis drives method selection: which assessment is least intrusive yet sufficient to answer the open question.

A disciplined intake sequence:

1. Define the referral concern in observable, measurable terms (topography, not interpretation).
2. List reported antecedents and consequences from records and interviews.
3. Note setting events, medical variables, cultural/contextual variables, and safety needs.
4. Identify the missing data that only direct observation or assessment can supply.
5. Choose the least intrusive method that can still answer the question.

Strong vs. weak hypotheses

• Strong (tentative, functional): 'Aggression may be maintained by escape from demanding writing tasks.'
• Weak (final, label-based): 'The student is defiant and unmotivated.'

The weak version is a diagnosis-by-record; it assigns a trait, stops inquiry, and cannot be tested. The strong version names a behavior and a candidate function and invites an experimental or descriptive test.

Assessment-to-decision link

Every question should narrow the next decision. If prior data show no independent toileting steps, a skills assessment should precede any reduction goal. If incidents cluster at transitions, descriptive assessment is warranted before selecting replacement-behavior goals. Records that are stale, contradictory, or label-heavy are exactly the cue to gather current direct data before acting.

A practical decision rule: the age and consistency of a record set how much weight it carries. A two-year-old plan written for a different setting is context, not truth; a chart documenting a seizure last week is an urgent safety flag. When records conflict — a teacher reports escape, a parent reports attention — the conflict itself is data: it tells you the function may differ by setting, or that nobody has confirmed it yet, and it points toward a method (often descriptive observation across both settings, then a functional analysis) that can resolve the disagreement empirically.

Finally, records review protects against duplicating services and against doing harm. If an active speech-language plan already targets requesting, the BCBA coordinates rather than competes. If notes flag a medical contraindication — say, a feeding restriction — that constraint must shape any goal involving food reinforcers. The output of a strong review is not a summary; it is a prioritized list of testable questions and safety constraints that the rest of Domain F will resolve.