Records, interviews, and ABC data all suggest why a behavior might happen. A functional analysis is where you stop guessing and test it. By arranging conditions on purpose and watching what happens, you can show which consequence actually maintains the behavior.
Why it matters
A functional analysis (FA) is the only part of a functional behavior assessment that experimentally demonstrates a functional relation. Treatments matched to function, such as extinction plus differential reinforcement, depend on getting the function right. A plan built on the wrong function can strengthen the very behavior you meant to reduce.
On the exam, expect to identify conditions, read FA graphs, pick a variation for a setting, and decide what to do when results are unclear or the behavior is dangerous.
Key ideas
- An FA is an experiment. You manipulate antecedents and consequences in test conditions and compare responding with a control condition. Function is shown by a difference between conditions, not by a single high data point.
- Standard conditions (in the tradition of Iwata and colleagues):
- Attention: the therapist seems busy and gives brief attention, such as a statement of concern, only after the target behavior.
- Demand (escape): the therapist presents tasks with prompting and removes the task for a short break after the target behavior.
- Alone or ignore: no materials and no social consequences. An ignore condition keeps an adult present when being alone is unsafe or when the behavior is aggression.
- Tangible: a preferred item is removed and briefly returned after the target behavior. Include it only when interviews or observations suggest that access to items matters, because repeatedly giving items back after the behavior can create a tangible function that was not there before.
- Play (control): preferred items, frequent noncontingent attention, no demands, and no programmed consequences for the behavior.
- Design. Most FAs use a multielement design, rapidly alternating conditions, often in a fixed or counterbalanced order. A pairwise design compares one test condition at a time with play, which can clarify unclear multielement results.
- Design the conditions from your hypotheses. Use what you learned from indirect and descriptive assessment to choose relevant test conditions and the specific reinforcers to test, such as a particular adult's attention or a favorite tablet.
- Safety and consent come first. Get informed consent, review medical risk, set clear session termination criteria (for example, a set number of responses or any sign of injury), arrange protective equipment or medical oversight when needed, and keep sessions as brief as the question allows.
- Variations fit real constraints.
- Brief FA: usually one short session (often 5 to 10 minutes) of each condition, followed by a brief contingency reversal to confirm the result, for settings with very little time.
- Trial-based FA: short control and test segments embedded in natural routines, often in classrooms.
- Latency-based FA: each session ends after the first response, and latency becomes the measure, which limits exposure to dangerous behavior.
- Precursor FA: consequences are delivered for a milder behavior that reliably comes before the severe behavior and appears to serve the same function.
Definitions
- Test condition: a condition that presents a specific establishing operation and delivers the suspected reinforcer contingent on the target behavior.
- Control condition: the comparison condition (usually play) in which the relevant establishing operations are absent and the behavior produces no programmed consequence.
- Undifferentiated responding: an FA result in which the behavior occurs at similar levels across all conditions, including the control, so no single social function stands out.
- Trial-based functional analysis: an FA variation that runs brief test and control segments within ongoing activities instead of in separate, longer sessions.
Examples from a session
Reading FA results
- Home: Mateo, age 5, screams. His BCBA runs short multielement sessions in the family's living room. Across 12 sessions, screaming is high in the attention condition and near zero in play, demand, and alone. His BCBA concludes it is maintained by social positive reinforcement (attention).
- School: Aaliyah, age 13, throws materials. Her teacher cannot run long sessions, so the BCBA uses a trial-based FA during class activities. Throwing happens mostly in the demand test segments and rarely in the matched control segments, which supports an escape function.
- Residential: Owen, 34, rubs and picks at his skin. Responding stays high in the alone condition and throughout a series of extended alone sessions, which points to automatic reinforcement.
- Clinic: Hiroshi, age 9, hits his head hard enough to bruise. The team uses a latency-based FA, with termination criteria and protective measures planned with his physician. Latencies are consistently shortest in the tangible condition, which points to access to items.
- Non-example: Zara, age 12, has outbursts. Her BCBA records ABC data for two weeks and finds that most outbursts follow task demands. This is a descriptive assessment, not an FA: nothing was arranged or changed, so it shows only a correlation.
Common mistakes / exam traps
- Calling a descriptive assessment an FA. ABC data show correlations, not functional relations. Only manipulating conditions can demonstrate function.
- Forgetting the control condition. High responding in the attention condition means little unless it is higher than in play.
- Ignoring multiple functions. Elevated responding in two test conditions can mean the behavior serves more than one function.
- Running an FA without safety planning. Severe behavior calls for termination criteria, risk review, and consent, not for skipping the FA or running unlimited sessions.
Exam trap
High vs. low undifferentiated results. High levels in every condition, including play, often suggest automatic reinforcement; check it with a series of alone or ignore sessions. Low or variable levels everywhere do not; they may call for longer sessions, a pairwise design, or more individualized conditions.
Exam trap
Elevated responding in the alone condition does not mean the client wants to be left alone. It means the behavior persists without social consequences, which suggests automatic reinforcement.
Quick check
Grace, a BCBA, runs a multielement FA for Leo, age 8, who bites his hand. Biting is consistently high in the attention and demand conditions and stays low in the alone, tangible, and play conditions. What is the most reasonable interpretation?