Descriptive Assessment and Functional Analysis

The Three Tiers of Functional Behavior Assessment

A full functional behavior assessment (FBA) has three tiers that move from weakest to strongest evidence.

1. Indirect assessment — interviews and rating scales completed from memory. Examples: the FAST (Functional Analysis Screening Tool) and the QABF (Questions About Behavioral Function). Fast, broad, low-risk, but vulnerable to report bias and memory limits. Generates hypotheses, never confirms function.
2. Descriptive assessment — direct observation of behavior in natural routines while recording ABC data continuously or by narrative, plus scatterplots that map when behavior occurs across the day. Preserves real context.
3. Functional analysis (FA) — experimental manipulation of antecedents and consequences to test which variable controls the behavior.

The decisive conceptual line the exam draws: descriptive data are correlational (behavior often follows hard tasks, or is often followed by attention), while an FA is causal because it controls the variables. Correlation can mislead — attention may follow problem behavior in observation yet not maintain it.

Why can descriptive data deceive? In natural settings, several events co-occur. After a child screams, a teacher often reprimands (attention) and the worksheet stops (escape) and peers look over (attention) — all at once. ABC observation records that attention followed the behavior, but it cannot isolate which consequence is responsible. Only by manipulating one variable at a time, as an FA does, can you show that escape (not attention) controls the behavior. This is a core assessment principle: descriptive = correlation, functional analysis = causation.

Descriptive Methods in Detail

Descriptive assessment is direct but uncontrolled. Three formats appear on the exam:

• ABC continuous recording — log every instance with its antecedent and consequence across an observation period; rich but labor-intensive.
• ABC narrative recording — open-ended notes of antecedents, behaviors, and consequences only when the behavior occurs; faster, less systematic.
• Scatterplot — a grid of time intervals marked when behavior occurs, revealing temporal patterns (e.g., clustering at transitions or a specific class) without specifying consequences.

Descriptive methods are correlational because natural consequences are not controlled. Their job is to sharpen the hypothesis the FA will test, or — when an FA is unsafe — to provide the best available, clearly caveated evidence.

Functional Analysis: Iwata Conditions and Variations

The standard FA (Iwata, Dorsey, Slifer, Bauman & Richman, 1982/1994) alternates conditions and compares each test condition against a control:

• Attention — behavior is ignored until it occurs, then brief attention is delivered (tests social-positive/attention function).
• Escape / demand — academic or task demands are presented; behavior produces a break (tests social-negative/escape function).
• Tangible — a preferred item is withheld and delivered contingent on behavior (tests access-to-tangibles function).
• Alone / ignore — the client is alone with no materials or attention; persistent behavior here implicates automatic (self-stimulatory) reinforcement.
• Play / control — enriched environment, attention freely available, no demands; expected to show the lowest responding and serves as the comparison baseline.

Identifying the function: the condition with clearly elevated responding relative to play names the maintaining variable. Elevated only in alone suggests automatic; multiple elevated conditions suggest a multiply-controlled behavior.

The four functions you must be able to name from FA results:

• Attention (social-positive) — behavior gets others' attention; elevated in the attention condition.
• Escape (social-negative) — behavior removes or postpones demands; elevated in the escape/demand condition.
• Tangible — behavior produces access to items/activities; elevated in the tangible condition.
• Automatic (sensory/self-stimulatory) — behavior produces its own reinforcement independent of others; persists in the alone condition, where no social consequence is available.

A key tell for automatic function: high, steady responding in the alone condition. Because no one is present to deliver attention, escape, or tangibles, behavior that continues there must be maintained by its own sensory product. By contrast, behavior that drops to near zero when alone is almost certainly socially mediated.

Variations (safety/feasibility trade-offs)

• Brief FA — a condensed session series for time-limited settings; less data, faster decisions.
• Trial-based FA — embeds short test/control trials in the natural classroom routine; high feasibility, good for teachers, slightly less control.
• Latency-based FA — measures time to the first response rather than rate; useful for dangerous behavior because a single response can end the condition, limiting exposure.

When NOT to run a standard FA

An FA evokes the very behavior of concern, so it requires consent, trained implementers, safety procedures, and data-based stopping rules. The exam often rewards choosing a medical referral, descriptive data, or a modified (latency/trial-based) FA when severe self-injury, setting limits, or client variables make a standard analysis unsafe or invalid. Do not run a standard FA automatically.