How to Operationally Define Assent Withdrawal for a BIP

When you write an assent withdrawal definition for a behavior intervention plan (BIP), you have to capture both the active version (the client pushes the task away, walks out, says "no") and the passive version (the client sits still, looks at the floor, stops engaging). A Board Certified Behavior Analyst (BCBA) who only writes the active half will miss most of what real withdrawal looks like in session. And if your definition does not match the response class the client is actually using, your data will lie to you, and so will your plan.

Active withdrawal vs passive withdrawal: write both#

Assent provision is forward motion. The client picks up the worksheet, walks to the table, makes the vocalization, taps the card. Something is happening that moves the behavior chain along.

Assent withdrawal is the opposite. The client pauses the chain, or moves in the other direction. Both count. The trap most BIPs fall into is only writing the loud version.

The active withdrawal list is easy. Pushing materials away. Standing up and walking out. Saying "no" or "stop." Crossing arms and turning the body away from the technician. These show up on a video review without anyone needing to squint.

The passive list is what gets missed. Sitting still while the technician presents a task. Staying outside the session room after a transition. Going quiet when the prompt comes. Resting the head on the table. Refusing to look at the materials. None of these involve a big motor response, but every one of them is a clear signal that the client is not consenting to the next step.

Any behavior that is a clear indication of choice away from an option may communicate the lack of assent. If the client is sitting still and not engaging with the technician, they're not engaging in their assent provision behavior. They're sitting outside the session room. There may be some resistance during transitions to tasks or away from other items. From the talk — Matt Harrington

If you only put the active list in your BIP, your registered behavior technicians (RBTs) will under-report withdrawal. They will score the kid as "compliant" when the kid is actually shut down. That data ends up driving the wrong clinical decision.

So write both lists. Two short bullets in the definition. One says what active withdrawal looks like for this client. One says what passive withdrawal looks like. Then your team has a chance of measuring the same thing.

The clenched-fists example: how a real definition gets built from observation notes#

The cleanest way to write a definition is to stop guessing and start watching. Sit in two or three sessions. Take notes that describe what the body is doing, not what you think it means. You are not labeling yet. You are just describing.

Your notes might look like this. Tech presented the task. Client clenched fists. Client frowned. Client stopped moving. Client sat on the floor. Five lines of plain description.

Now read them back. Which one shows up first? Which one is the easiest to see across the room? Which one happens before the bigger, more disruptive behavior? That is the candidate behavior for your operational definition.

Individualize to your client. The tech presented task. Client clenched fists. Client frowned. Client stopped moving. Client sat on the floor. Things like that where you start looking and you're like, okay, well, maybe clenching fists is an easy observable behavior that is within that assent withdrawal behavior chain. From the talk — Matt Harrington

Clenched fists wins a lot of the time. It is early in the chain. It is visible. It does not require the technician to read the client's mind or guess at internal state. An RBT can mark a tally without having to interpret anything.

Once you have the candidate, write the definition the way you would write a definition for any target behavior. Include the topography. Include the boundary (when does the behavior start, when does it stop). Include one positive example and one non-example. If you cannot write a non-example, your definition is too vague.

Response class hierarchy: where does the withdrawal sit?#

Assent withdrawal is almost never one behavior. It is a class of behaviors that all do the same job for the client: escape, delay, or signal "I am not okay with this."

That class usually has a hierarchy. Quiet withdrawal at the bottom. Verbal protest in the middle. Property destruction or aggression at the top. The client moves up the hierarchy when the lower-cost behaviors do not get a response.

When you write your definition, you want it to land on a behavior near the bottom of the hierarchy. That gives the team a chance to honor the withdrawal before the client has to escalate.

We had a client who I would hold out the instructional material and then I would only proceed with prompting following a tap. And it resulted in escape behavior when it did occur, was not providing a tap rather than a more dangerous form in that response hierarchy. From the talk — Matt Harrington

That tap-or-no-tap setup is the whole game in one example. The withdrawal behavior was defined as "client does not tap the material." It is quiet. It is low cost. It sits at the bottom of the response class. And because the team honored it (no tap, no prompt, no task), the client never had to use the louder, more dangerous version.

If your definition only catches the top of the hierarchy (the meltdown, the aggression), you are too late. You are measuring the alarm, not the warning.

The 'movement toward, movement away, pause' test#

Here is a quick test you can run on any candidate definition before you put it in the BIP. Ask three questions about the behavior.

One. Is it movement toward the task? If yes, that is provision, not withdrawal. Take it off the list.

Two. Is it movement away from the task? If yes, that is active withdrawal. Keep it.

Three. Is it a pause in the chain? Sitting, freezing, going silent, dropping eye contact. If yes, that is passive withdrawal. Keep it.

If you cannot answer one of those three for a behavior, the behavior is too fuzzy. Either tighten the description or pick a different behavior.

This test takes about ten seconds per item. Run it on every behavior in your draft definition before the plan goes out.

Common operational definitions that fail treatment integrity#

Treatment integrity means everyone on the team scores the same behavior the same way. If two RBTs watch the same five minutes of session and one marks four withdrawals while the other marks zero, your definition is broken.

A few patterns show up over and over.

The mind reader. "Client appears upset." "Client seems frustrated." If your RBT has to guess at an emotion, you do not have an operational definition. You have a guess.

The catch-all. "Any sign the client does not want to do the task." This is a paragraph, not a definition. It pushes the work of deciding onto the RBT in real time, and they will all decide differently.

The missing passive list. "Client says no, walks away, or pushes materials." This only catches the active version. The quiet kid who shuts down for forty minutes will get scored as "no withdrawals" all session.

The unit-less. "Crying." For how long? Loud or quiet? With or without tears? A unit of measurement plus a duration cutoff turns this into something an RBT can actually tally.

When you spot one of these, send the plan back to the drawing board. It is faster than running six months of bad data.

A copy-paste template for your BIP#

Use this as a starting point. Edit it for your client. Do not ship it unedited.

Assent withdrawal is defined as any of the following behaviors occurring within ten seconds of task presentation or during ongoing task demands:

Active withdrawal. Pushing materials more than six inches away from the body. Standing up and moving more than three feet from the work area. Verbal protest including "no," "stop," "I'm done," or similar phrases.

Passive withdrawal. Sitting still without responding to the task prompt for more than five seconds. Remaining outside the session area after the transition cue. Dropping head onto the table or arms. Going silent after previously vocalizing during the same activity.

Onset. The first of any listed behavior after the task is presented.

Offset. Ten consecutive seconds without any listed behavior occurring, or the task is removed by the technician.

Non-example. Client looks away briefly to track a noise, then returns to the task within three seconds. Client asks a question about the task. Client requests a short break using the agreed-upon signal.

That template is not the final answer for your client. It is scaffolding. The point is that every line tells the RBT exactly what to score, and tells your supervising BCBA exactly what to audit during a treatment integrity check.

Frequently asked questions#

Should crying count as assent withdrawal?

Sometimes, but not by default. Crying can be assent withdrawal, a pain response, a tired response, or a delayed response to something that happened earlier in the day. If you put "crying" in the definition with no qualifier, you will get noisy data. A cleaner approach is to define the type of crying you mean (for example, "crying that begins within five seconds of task presentation and stops when the task is removed") and pair it with at least one other behavior in the class. Crying alone, with no context, almost never makes a good standalone marker.

What if the client uses different withdrawal behaviors across staff?

That is real, and it usually means the response class is shaped differently with different people. The client has learned that one technician responds to a quiet behavior and another technician only responds to a louder one. The fix is not to write three different definitions. The fix is to write one operational definition that covers the full range of behaviors the client uses, then run treatment integrity checks to make sure every staff member honors the same low-cost behaviors. If you only honor withdrawal for the loud version, you are training the client to escalate with that staff member.

How is this different from defining a precursor behavior?

A precursor is a behavior that reliably comes before a target problem behavior. Assent withdrawal is broader. Every precursor to escape-maintained problem behavior is a form of assent withdrawal, but not every assent withdrawal is a precursor to something dangerous. A client who quietly looks at the floor when a non-preferred task comes out is withdrawing assent. They may never escalate. The job of the BIP is to honor the withdrawal regardless of whether a bigger behavior would have followed.

Where to go from here#

The definition is half the work. The other half is what the team does with the data once they start tracking it. If your RBTs can score withdrawal but the plan does not tell them what to do when withdrawal happens, the data goes in a drawer.

Watch the full talk for the assent provision side of the equation, the data sheet setup, and the response protocol when withdrawal occurs in session.