Descriptive assessment

Observe ordinary conditions without arranging a test

During a descriptive assessment, an observer records a predeclared response and selected features of the surrounding context. Those features may include activities, requests, choices, communication access, people present, noise, pain or illness reports, time, setting, events immediately before and after the response, and how communication partners act.

The ABAI Basic Principles in Behavior Analysis page is a higher-education training resource. It supplies foundational context, rather than a clinical assessment standard. The current BACB BCBA Test Content Outline, 6th edition specifically includes designing and evaluating descriptive assessments in item F.5, alongside separate content for measurement, functional analysis, interpretation, client-informed goals, and contextual fit. The outline describes examination content and does not authorize practice.

Common recording formats include:

• antecedent-behavior-consequence, or ABC, records
• structured narrative records with operational definitions
• scatterplots showing occurrence across time blocks or activities
• frequency, duration, latency, or interval measures
• conditional-probability comparisons that include relevant base rates
• measures of communication-partner, staff, access, or environmental implementation

The format follows the question. A scatterplot may locate time patterns but miss immediate sequences. ABC records preserve sequence but can overrepresent memorable events. Conditional calculations can be misleading when definitions, sampling, or comparison events are weak.

Descriptive assessment, indirect assessment, and FA differ

A 2012 paper by Hanley uses these three distinctions in a discussion of functional assessment for problem behavior. It is a clinical commentary grounded in that literature, not a rule that every assessment needs all three methods.

Descriptive observations can generate or challenge hypotheses and reveal practical conditions that an interview missed. They remain correlational. A consequence may often follow behavior because caregivers appropriately respond to risk, pain, or communication. That sequence does not establish that the consequence caused or maintains the behavior.

The limitation appears in empirical work. A 2009 study by Pence, Roscoe, Bourret, and Ahearn compared three descriptive methods with functional analyses for six participants. Descriptive outcomes were similar across methods for five participants, yet descriptive and functional-analysis outcomes differed substantially for five. This small study does not supply a universal error rate; it shows why agreement among descriptive formats cannot establish a functional relation.

Design observations around a decision

Before collecting data:

1. define the person-selected or meaningfully agreed question
2. identify who may consent and obtain assent when applicable
3. define behavior and context in observable, neutral language
4. choose settings, times, people, and comparison events that represent the question
5. define eligible opportunities, exclusions, start and stop rules, and missing data
6. select measures that capture the important dimension
7. train observers on examples and nonexamples
8. plan observer-agreement and procedural-integrity samples when decisions depend on the data
9. set safety, privacy, health, and referral routes
10. decide how the person and stakeholders will review the result

Sample both occurrence and relevant nonoccurrence. If every ABC form begins only after the target response, the record cannot show how often the same antecedent occurs without it. Report raw counts and exposures before percentages. Preserve dates, settings, observation duration, observer, ordinary supports, exclusions, data loss, and every procedure change.

For BCBA and BCaBA certificants and people who completed an application for either credential, the BACB Ethics Code addresses client involvement, informed consent and assent when applicable, medical needs, scientifically supported and context-responsive assessment, risk minimization, written results, and appropriate data collection. The BACB has no separate jurisdiction over organizations or corporations. Other roles remain governed by their own competence, supervision, licensure, setting, payer, and legal requirements.

Protect access and avoid manufactured exposure

Observation should preserve ordinary communication and essential supports. ASHA's AAC practice portal describes aided and unaided communication and says AAC users should always have their tools or devices available. Record whether the system is present, positioned, charged, understood by partners, and usable before interpreting an absent response.

Do not withhold food, water, bathroom access, rest, mobility or sensory supports, prescribed care, pain care, or a safe exit to generate an event. A descriptive assessment observes naturally occurring conditions; deliberately arranging conditions moves toward experimental assessment and needs a different rationale, authority, consent, design, and safety review. Immediate risk calls for safety action, not continued data collection.

A fictional descriptive assessment

Niko is a fictional eleven-year-old who wants after-school cleanup to allow an understandable pause message. With Niko, the clinician defines an eligible transition as a planned cleanup request with speech and AAC available. A recognizable pause message can use speech, AAC, or a preidentified gesture. A partner response means acknowledgment within 30 seconds and either the agreed pause or a clear, accessible explanation.

Across 16 naturally occurring transitions, Niko communicates pause in six. Partners give the defined response after 2 of 6 messages. A choice of cleanup role is absent in five of the six message transitions, or 5 of 6. It is also absent in four of the ten transitions without a message, or 4 of 10.

The higher observed proportion supports examining role choice, partner response, task features, and other differences. It does not show that missing choice caused the message. The sample is small, several variables change together, and Niko may communicate for more than one reason. The report includes all 16 opportunities, raw counts, prompt and access information, Niko's account, partner behavior, observation dates, agreement samples, missing data, and competing explanations.

Possible next steps include improving partner response and choice immediately when those changes fit Niko's preferences, gathering more representative observation, seeking medical or interdisciplinary input, or considering a stronger comparison under qualified review. Useful supports do not need to wait for a causal test when their value and safety are already clear.