FA vs FBA: What's the Actual Difference?

The functional behavior assessment (FBA, the full process behavior analysts use to figure out why a behavior happens) is the umbrella term for three distinct phases: the indirect assessment, the descriptive assessment, and the functional analysis (FA, the experiment that confirms the function). The FBA is the whole process. The FA is one step inside it. The reason this matters: research shows caregiver interviews only match the FA's answer about 50 to 60 percent of the time for closed questions, and 50 to 75 percent of the time for open ones. So if you stop at the FBA and never run the FA, you are guessing the function about half the time.

Short answer: FBA is the umbrella, FA is the experiment#

The FBA is not one tool. It is a process with three phases. Each phase has a different job. Mixing them up is the most common confusion new behavior analysts (BCBAs, board certified behavior analysts who design treatment) bring to their first case.

The indirect assessment is the interview phase. The descriptive assessment is the observation phase. The functional analysis is the test phase. You do not pick one. You do them in order, and each one feeds the next.

I like to think of the functional behavior assessment, the functional assessment in general, as an umbrella term for three distinct phases. You have the indirect assessment, the descriptive assessment, and then the functional analysis. From the talk — Matt Harrington

This is the cleanest way to keep the terms straight. FBA = the umbrella. FA = the experiment. Everything else lives inside the FBA umbrella with it.

The three phases inside an FBA#

Think of an FBA like building a case. Each phase gathers a different kind of evidence.

Phase one is the indirect assessment. You talk to the people who know the client. Parents, teachers, RBTs (registered behavior technicians, the staff who run sessions). You ask what the behavior looks like, when it happens, and what tends to come right before and right after.

Phase two is the descriptive assessment. You watch the behavior in real time. You take ABC data, which is antecedent, behavior, and consequence data. You see for yourself what is happening in the natural setting.

Phase three is the functional analysis. You stop watching and start testing. You set up short conditions where you control what comes before and after the behavior. Then you see which condition makes the behavior happen most.

Each phase narrows the question. By the time you run the FA, you should already have a strong guess. The FA is what turns that guess into a confirmed answer.

What the indirect assessment is actually for#

Here is where new BCBAs get it wrong. They treat the interview like it is supposed to give them the function. It is not.

The interview is for setting up the FA. That is its job. Caregivers know things you cannot see in one session. They know what the kid loves. They know what sends the kid into a tailspin. They know which routines tend to spark the behavior. You use that information to build your FA conditions, not to skip the FA.

What about informing the conditions of a functional analysis? Bingo. That's what we're looking for. We've seen that caregivers can really accurately report on these conditions. And so we want to use what they know and then use what we observe to then inform and better refine our functional analysis conditions. From the talk — Matt Harrington

So when you sit down with a parent, you are not asking "what is the function?" You are asking "what should I put in my test conditions?" That is a different question, and it changes how you run the interview.

If you ask a parent what they think the function is, you are asking them to do your job. They cannot. They have not been trained to. That is not on them.

What the descriptive assessment is actually for#

The descriptive piece (the part where you watch the behavior in the natural setting) has the same job as the interview. It is not there to confirm the function. It is there to sharpen your test.

You watch what really happens before the behavior. You watch what really happens after. You see if the parent's report matches what you observe. Sometimes it does. Sometimes the parent says "he hits when I tell him no" and you watch for an hour and see he actually hits when his sister walks in the room. Now you know something you did not know before.

That new information goes into your FA design. Maybe you add a tangible condition. Maybe you build a sibling presence into one of your test conditions. The point is the same as the interview: refine the test.

If you skip the FA and just write a treatment plan from the descriptive data, you are guessing. You may guess right. You may guess wrong. You have no way to know which.

Why a hypothesized function is not a confirmed function#

This is the part most pages skip. There is a real, technical difference between a hypothesized function and a confirmed function.

A hypothesized function is your best guess based on what people told you and what you saw. A confirmed function is what shows up when you actually control the variables and run the experiment. The first is a maybe. The second is an answer.

If you're using an indirect assessment and a descriptive observation, well, then you can then say that it's a hypothesized function, but it's not until you really start playing with variables and manipulating the environment that you can say confirm that there is a functional relationship. From the talk — Matt Harrington

Why does this matter for treatment? Because a function-based treatment plan only works if it is matched to the actual function. If your guess is wrong, your plan is wrong, and the behavior does not get better. Sometimes it gets worse, because you are accidentally reinforcing the thing you are trying to reduce.

The research backs this up. When you compare what indirect assessments say the function is against what the FA later confirms, the match rate is not great.

Really, why is it that this research is showing such really bad results? Well, there's a couple of reasons, right? The probably most primary is that there's really minimal training in clinical interviewing... For closed-ended interviews, you're looking at about 50% to 60%. For open-ended interviews ... you're looking at anywhere from 50% to 75%. From the talk — Matt Harrington

So if you write a plan based only on the interview, you are right about half the time. Coin-flip odds. That is not a standard you would accept for a kid in your care.

When you can stop at FBA and when you cannot#

Real talk. Not every case gets an FA. Time, settings, and safety all push back.

When you can probably stop at the FBA: - The behavior is mild and not getting in the way of learning. - The environment is so chaotic an FA would not be valid anyway (think early-stage school placements with too many moving parts). - A trial-based or brief FA already gave you a clean differentiated answer in session. - The team is doing a quick check-in on an old case where the function was already confirmed years ago.

When you really should not stop at the FBA: - The behavior is severe or dangerous. - The treatment plan you are about to write involves extinction, blocking, or anything that could spike the behavior before it drops. - Two reasonable interviewers disagree on the likely function. - The first treatment plan based on the hypothesized function already failed.

In short, the harder the case, the more you need the experiment. The lighter the case, the more you can lean on the indirect and descriptive phases. But the default should be: if a treatment plan depends on knowing the function, the function should be confirmed, not guessed.

Frequently asked questions#

Is an FBA enough for a function-based treatment plan?

Sometimes, but not for the hard cases. If the behavior is mild and the team agrees on what is keeping it going, you can often move forward with a treatment plan from the indirect and descriptive phases alone. For severe behavior, or any plan that involves extinction or restrictive procedures, you need the FA to confirm the function. A plan built on a guess can fail or make things worse. The riskier the plan, the more you need the experiment behind it.

Do BCBAs have to do the FA themselves or can a BCaBA?

A BCaBA (board certified assistant behavior analyst, who works under a BCBA's supervision) can run FA sessions, but the BCBA is responsible for the design, the safety plan, and the read of the data. The BCBA decides which conditions to test, what to do if behavior spikes, and when the data shows enough to stop. The BCaBA can carry out the sessions hands-on. The clinical call stays with the BCBA. Same answer for RBTs running sessions: they can run conditions with training, but they do not own the design.

Does insurance pay for the functional analysis separately?

It depends on the payer and the state. Many insurance plans bundle the FA into the broader assessment authorization (often coded as adaptive behavior assessment hours, like 97151). Some plans authorize FA hours separately if you justify it in the treatment plan. The key is documenting why the FA is needed. If a hypothesized function is not enough to safely guide treatment, write that down. Reviewers look for that reasoning before they approve extra hours.

Watch the full talk#

If you want the long version, that is in the recording. Matt walks through his actual case stories. The 90-minute PICA FA that Santa's magnets explained. The telehealth trial-based FA he had to stop and restart. The siblings whose individual conditions fell apart the second they were in the same room. The article gives you the framework. The talk gives you the field notes.

Watch the full talk →