A functional analysis (FA) is a structured assessment that identifies why a problem behavior occurs by testing specific environmental conditions one at a time. Each condition is designed to isolate a possible reason for the behavior, such as getting attention, escaping a task, or accessing a preferred item. By comparing how often the behavior occurs across these conditions, you can determine what’s reinforcing it and design an effective treatment. The standard FA, first developed by Brian Iwata and colleagues in 1982, remains the most reliable method for identifying behavioral function and has been refined significantly in the decades since.
The Four Standard Conditions
A standard FA uses a multielement design with four conditions: attention, demand (escape), alone, and play (control). Each condition tests a different hypothesis about what maintains the behavior. You rotate through these conditions across multiple sessions, typically in a randomized or semi-randomized order, and graph the results to see which condition produces the highest rates of the target behavior.
Play (Control) Condition
This serves as your baseline. The person has access to preferred items and activities, receives attention on a regular basis without having to engage in problem behavior, and faces no demands. Because all potential reinforcers are freely available, you should see low or zero rates of the target behavior. If the behavior is consistently higher in one of the test conditions compared to this control, that tells you something meaningful about function.
Attention Condition
The person has access to items but you withhold social attention. You occupy yourself with something else, like reading or doing paperwork. When the target behavior occurs, you deliver brief attention, often in the form of a mild reprimand or concerned statement like “Don’t do that, you might hurt yourself.” If the behavior spikes in this condition compared to the control, it suggests the behavior is maintained by social attention.
Demand (Escape) Condition
You present tasks or instructions using a structured prompting sequence. When the target behavior occurs, you remove the task materials and turn away for a brief period, giving the person a break from the demand. If behavior is consistently elevated in this condition, it’s likely maintained by escape from or avoidance of tasks.
Alone Condition
The person is in a room without social interaction, attention, or task demands. No materials are provided (or only minimal items). If the behavior occurs at high rates in this condition, it suggests the behavior is automatically reinforced, meaning the behavior itself produces some form of sensory stimulation that maintains it. When the target behavior is aggression directed at others, an “ignore” condition is substituted: a therapist is present in the room but does not interact with the person throughout the session.
Whether to Include a Tangible Condition
Many practitioners add a fifth condition to test whether the behavior is maintained by access to preferred tangible items. In this condition, you give the person a highly preferred item, then remove it. When the target behavior occurs, you return the item briefly.
However, this condition requires careful thought. Research published in the Journal of Applied Behavior Analysis has shown that tangible conditions can produce false positives, meaning the behavior appears to be maintained by tangible access when it actually isn’t. The recommendation is straightforward: don’t include a tangible condition unless there’s strong reason to suspect that problem behavior is actually reinforced by tangible items in the person’s daily life. When you do include it, the items used should reflect what is actually delivered as a consequence for problem behavior in the person’s natural environment, not just things the person likes. If you can’t identify specific tangible items that are regularly given contingent on problem behavior, the tangible condition is likely unnecessary.
Session Duration and Repetition
The traditional FA uses 15-minute sessions per condition. Research comparing different session lengths found that 10-minute sessions produced interpretations identical to those from 15-minute sessions across all cases studied. Five-minute sessions yielded a small number of discrepancies, mostly because some behaviors increased toward the latter parts of longer sessions. For most cases, sessions of 10 minutes work well and improve efficiency without sacrificing clarity.
You’ll need multiple exposures to each condition to draw reliable conclusions. There’s no fixed number of sessions that guarantees clear results, but you continue rotating through conditions until you see stable, differentiated patterns in the data. Visual analysis of graphed data is the primary method for determining when you have enough information.
What Success Rates Look Like
Not every FA produces a clear answer on the first attempt. A large study of 176 inpatient cases found that the standard FA identified a function in 46.6% of cases on the first try. That number might seem low, but it’s the starting point, not the end. After modifying the FA once for the remaining unclear cases, an additional 67.1% of those became differentiated. After a second modification round, another 66.7% of still-unclear cases resolved. Across the full sample, a function was ultimately identified in 86.9% of all 176 cases, and in 93.3% of cases where up to two modifications were attempted.
The takeaway: if your initial FA doesn’t produce clear differentiation, modifying your conditions (adjusting how you deliver consequences, changing the types of attention or demands, or adding conditions) will likely get you there.
Brief Functional Analysis
When time with a client is limited, such as in outpatient clinics or consultation visits, a full multielement FA with repeated sessions may not be feasible. The brief functional analysis (BFA) was developed for exactly these situations. It consists of a single exposure to each 5-minute test and control condition, with replication of the key test condition if time allows, followed by a treatment probe. The entire process can be completed within a 90-minute window.
The BFA trades some of the confidence you get from repeated measures for practical efficiency. It’s best suited for situations where you need a working hypothesis quickly and can confirm it through treatment outcomes.
Trial-Based FA for Natural Settings
Standard FAs are typically conducted in controlled clinic rooms, but many practitioners need to assess behavior in classrooms, homes, or community settings. The trial-based functional analysis (TBFA) adapts the process for these environments. Instead of timed sessions, the TBFA uses pairs of brief trials: a control trial and a test trial for each condition, each lasting about 2 minutes.
You run 6 to 9 pairs of trials per condition, with the exact number determined by data stability. During each trial, you simply record whether the target behavior occurred or did not occur, rather than counting exact frequency. This binary scoring system makes data collection much simpler for teachers, parents, or other caregivers. Research has shown that parents can be coached to implement TBFAs in their own homes with support, making this a practical option when clinic-based assessment isn’t possible.
Using Precursor Behaviors for Safety
Some target behaviors are dangerous enough that repeatedly evoking them raises ethical and safety concerns. One well-supported approach is to conduct the FA on precursor behaviors instead. Precursors are lower-intensity behaviors that reliably occur before the severe behavior, such as tensing up before self-injury, or verbal threats before aggression.
Research has confirmed that precursor behaviors often share the same function as the severe behavior they precede. In published studies, all participants showed clear differentiation when functional analyses were conducted on precursors, and the identified functions were successfully used to design effective treatments. Sessions in precursor FAs are typically shorter (around 5 minutes each), and because you’re measuring a less dangerous behavior, you can stop a session before it escalates to the severe target behavior.
Graphing and Interpreting Results
FA data are graphed on a line chart with sessions on the horizontal axis and the rate (or percentage) of the target behavior on the vertical axis. Each condition gets its own data path, typically distinguished by different markers or line styles. You’re looking for one condition to produce consistently higher rates of behavior compared to the others, particularly compared to the play/control condition.
Clear differentiation looks like one data path that separates visually from the others with minimal overlap. If the attention line is consistently elevated while the demand, alone, and play lines remain low, the behavior is likely maintained by attention. If two conditions are elevated, the behavior may have multiple functions, and treatment will need to address both. Undifferentiated results, where all conditions look similar, mean the FA hasn’t identified a clear function yet and modifications are needed.
When interpreting your graph, look for level differences between conditions (is one consistently higher?), trends within conditions (is behavior increasing over sessions in one condition?), and variability (are the data points scattered or stable?). Stable, separated data paths give you the most confidence. High variability within a condition or substantial overlap between conditions makes interpretation harder and often signals the need for more sessions or a modified protocol.