Key takeaways
- A hypothesis names the antecedent, the behavior, and the function.
- State the supporting data pattern before you state the function.
- There is one worked example here for each of the four functions.
- When two functions fit, split by context or name a primary and a secondary.
- Strong wording is observable and specific; weak wording is a label.
A function hypothesis statement is the single sentence that carries an FBA into a behavior plan. It says, in plain terms, what triggers the behavior and what the behavior accomplishes for the student. Get it right and the plan almost writes itself. Get it vague and the plan addresses the wrong thing. This article gives you a reliable format, a worked example for each of the four functions, a hard case where two functions compete, and a side-by-side look at wording that holds up versus wording that does not.
The format
A function hypothesis statement follows a simple shape:
When [antecedent], the student [behavior], in order to [function].
The antecedent is what happens right before. The behavior is observable, described the way two people would agree on. The function is the purpose: escape or avoidance, attention, access to a tangible or activity, or automatic and sensory. Behind every good hypothesis is a data pattern that supports it. State the pattern first, then the hypothesis, so the reader can see why you believe it.
It helps to remember what the statement is for. The behavior plan that follows will try to change the antecedent, teach a replacement behavior that serves the same function, and stop reinforcing the problem behavior. Every part of that plan reads directly off the hypothesis. If the function is wrong, the plan addresses a purpose the behavior never had, and it fails no matter how well it is written. That is why the hypothesis is worth getting precise rather than plausible.
A note on the four functions before the examples. Escape and avoidance is getting out of something non-preferred. Attention is gaining a social response from an adult or peer. Access is getting a preferred item or activity. Automatic or sensory is producing an internal result that does not depend on anyone else. The same topography of behavior, say leaving a seat, can serve any of these, which is exactly why you cannot name the function from the behavior alone. You name it from the antecedent and consequence pattern around it.
Worked example: escape or avoidance
Data pattern: A composite seventh grader we will call Liam. Across four observation sessions, work refusal (pushing the paper away and putting his head down) occurred in eleven of fourteen instances right after a multi-step math worksheet was handed out. It rarely occurred during partner work or warm-ups. In nine of those eleven instances, the worksheet was set aside or shortened after the refusal.
Hypothesis: When Liam is given a multi-step independent math worksheet, he refuses the work and puts his head down, in order to escape or delay the non-preferred task.
Worked example: attention
Data pattern: A composite first grader we will call Sofia. Across five sessions, calling out and leaving her seat clustered in the minutes after the teacher turned to work with a small group. In ten of twelve instances, the consequence was that the teacher spoke to her or walked over. The behavior was rare when she was already receiving one-on-one help.
Hypothesis: When adult attention shifts away from Sofia to other students, she calls out and leaves her seat, in order to gain adult attention.
Worked example: access to a tangible or activity
Data pattern: A composite kindergartner we will call Theo. In three sessions, crying and dropping to the floor followed being told that tablet time or the block center was finished. In eight of nine instances the behavior stopped within a minute of the item or activity being returned or a timer being added. It did not occur during transitions that did not end a preferred activity.
Hypothesis: When a preferred activity such as the tablet or blocks is ended, Theo cries and drops to the floor, in order to regain access to the activity.
Worked example: automatic or sensory
Data pattern: A composite fifth grader we will call Noah. Repetitive hand flapping and humming occurred across every setting observed: math, lunch, the hallway, and independent reading. It happened whether or not an adult or peer responded, and it was most frequent during unstructured downtime. No social consequence reliably changed it.
Hypothesis: During unstructured or low-demand moments, Noah engages in hand flapping and humming, in order to obtain sensory input, with no evident social function.
When two functions compete
Sometimes the data does not cleanly point to one function. The honest move is not to pick the tidier answer. It is to look at whether each function lives in a different context.
Data pattern: A composite second grader we will call Mara. Out-of-seat behavior showed two clusters. During independent writing, leaving her seat was usually followed by the task being delayed (an escape signature). During reading, the same behavior was usually followed by a peer or adult talking to her (an attention signature). Antecedents and consequences differed by setting.
What to do: Treat both as live hypotheses split by context. Write two statements rather than forcing one. During independent writing, Mara leaves her seat in order to escape the task. During reading, she leaves her seat in order to gain attention. If the two truly overlapped in the same context, you would name the better-supported function as primary and note the second, then keep collecting to settle it.
Splitting by context is not a failure of the FBA. It is often the most accurate finding, and it lets the plan respond differently in writing than it does in reading.
There is a related trap worth naming. When the data is genuinely thin, two functions can look equally possible simply because you do not have enough rows to separate them. That is not a competing-hypothesis situation, it is an under-collected one, and the right response is more observation rather than a coin flip. The way to tell the difference is to look at whether the antecedents and consequences cleanly sort by setting. If they do, you have two real functions in two contexts. If they are just jumbled together with no pattern, you do not yet have a hypothesis at all, and forcing one now will send the plan in a direction the next month of data may contradict.
Strong versus weak wording
The difference between a hypothesis a team can act on and one they cannot usually comes down to wording. Strong wording is observable and specific. Weak wording is a label dressed up as a finding.
| Weak wording | Strong wording |
|---|---|
| The student is defiant during math. | When given an independent math worksheet, the student pushes the paper away and puts his head down, in order to escape the task. |
| She acts out to get attention. | When adult attention shifts to other students, she calls out and leaves her seat, in order to gain adult attention. |
| He has tantrums when he does not get his way. | When a preferred activity is ended, he cries and drops to the floor, in order to regain access to the activity. |
| The behavior is just sensory. | During low-demand moments, he flaps and hums regardless of who responds, in order to obtain sensory input. |
The weak versions name a trait or a guess. The strong versions name the antecedent, an observable behavior, and the function the data supports, which is the only version a behavior plan can be built on. If you can read your hypothesis and not know what the plan should change, rewrite it until the antecedent and the function are both visible.
A checklist before you commit the statement
Run a finished hypothesis past four quick questions. Could two people agree on whether the behavior happened, or is the behavior word actually a judgment? Is the antecedent a real event in the room, not a restatement of the behavior or the student's mood? Does the named function match what your data shows the behavior produced most of the time, rather than the function that feels most familiar? And could a colleague who never met the student read the statement and know what the plan should change? If any answer is no, the statement is not ready.
One last honesty point. A hypothesis is a best current explanation, not a verdict, and the plan built on it is the real test. If a plan that matches the hypothesis does not move the behavior, the first thing to revisit is the function, not the student. Strong wording makes that easier, because a precise statement is one you can clearly confirm or rule out, while a vague one can be quietly true and useless at the same time.
The statement is only as good as the rows under it. The ABC observation log shows how to tally the antecedent and consequence columns that feed the data pattern, and the practical guide to running an FBA places the statement in the full process. Once the statement is settled, the function-to-strategy map turns it into prevention, teaching, and response strategies.
Common questions
What is the standard format for a function hypothesis statement?
A common and reliable format is: when [antecedent], the student [behavior], in order to [function]. It forces you to name the trigger, the observable behavior, and the purpose the behavior serves, which is exactly what a behavior plan needs to address.
What if the data supports two functions?
Look at whether each function travels with a different setting. Often one function drives the behavior in one context and another drives it elsewhere. If they genuinely overlap, name the better-supported one as primary and note the second so the plan can cover both.