Treatment Integrity in ABA: What It Is and Why It Matters

Treatment integrity in ABA is how closely a treatment plan is carried out the way it was written. When integrity is high, the client gets the exact steps the BCBA designed. When it slips, the client gets something else, and the data stop telling the truth.

That simple idea touches almost every part of daily practice. It shapes how RBTs (Registered Behavior Technicians, who run day-to-day sessions) carry out programs. It also shapes how BCBAs (Board Certified Behavior Analysts, who design and oversee those programs) read graphs. It also shapes how supervisors coach staff and how parents learn to carry the plan home. Below, eight clinicians share how they think about it in real sessions.

What treatment integrity means#

In an Article Club session on discrete trial instruction, Sam Mueller opened with a clean definition. Discrete trial instruction, often called DTI or DTT, is a structured way of teaching one skill at a time. Each trial has a few set parts. There is an instruction, a prompt, the child's response, a consequence, and a short pause before the next trial.

treatment integrity is the degree to which an independent variable is implemented with precision.From the talk — Sam Mueller

The "independent variable" is just the thing we change on purpose. In ABA, that is the teaching procedure itself. So treatment integrity asks a plain question. Did the procedure actually happen the way the plan says? Mueller noted that DTI is used to teach a wide range of skills. These include receptive labeling, expressive language, social skills, and academics. As Mueller's co-host put it, this is "stuff we do all day." That is why small slips add up so fast. A step missed once is minor. A step missed across hundreds of trials changes the whole program.

Why low integrity weakens teaching#

Mueller went on to explain why precision matters so much. Prior research shows that errors in delivery can make an intervention less effective. Two parts of DTI stand out in that research. One is giving the controlling prompt, which is the prompt that reliably gets the right answer. The other is withholding information as the procedure calls for.

reduced levels of treatment integrity can decrease the effectiveness of interventionsFrom the talk — Sam Mueller

Mueller also flagged a gap in older studies. Many of them lacked ecological validity. That term means how well a study matches real life. The errors that researchers built into their studies were not always the errors that happen in actual classrooms. Some studies did not even use children as participants. So the article Mueller reviewed tried to study mistakes that look like the ones real staff make. For a practicing BCBA, that framing is useful. It reminds us to watch for the errors our own teams make most, not just the textbook ones.

Integrity protects the data#

Anlara Hoskins came at treatment integrity from the RBT side, through the ethics code. She pointed to RBT code 1.03. It asks RBTs to run only what is in the treatment plan.

which really talks about only implementing things that are written in the treatment plan or as prescribed by the BCBA and not modifying those treatment plansFrom the talk — Anlara Hoskins

Hoskins explained why this matters for the client. If an RBT runs a different procedure than the written one, the data get skewed. A plan that works might look like it fails. A plan that fails might look like it works. Either way, the graph no longer reflects what the plan really does. Then the BCBA may change the plan when it should stay the same. Or the BCBA may keep it when it should change. Hoskins was direct that this outcome is harmful to the client. Good integrity is what lets the data guide real decisions.

Why RBTs follow the plan as written#

Hoskins also gave a second reason for sticking to the plan. This one is about protecting the RBT.

only implementing what is written is really ensuring that you are being protected because you do not have the same liability protection that a BCBA doesFrom the talk — Anlara Hoskins

She used an architect and a construction crew as a picture. The crew builds the building. The architect designs it. If the building fails for structural reasons, the architect's job is on the line. The BCBA and RBT roles work in a similar way. The BCBA designs the plan and carries the clinical risk for it. The RBT carries it out.

Hoskins framed this as a form of self-advocacy, not just a rule. She noted that RBTs sometimes feel trapped in the BCBA and RBT relationship. Yet they hold their own certification and their own ethics code. Following the plan as written is one way RBTs protect that certification while still meeting the client's needs.

The "what" belongs to the BCBA#

Noah McAmis, speaking about teaching beyond the table, drew a very clear line. He split every program into a "what" and a "how and where." The "what" belongs to the BCBA.

The what is the target. The response criteria. The prompting hierarchyFrom the talk — Noah McAmis

The response criteria spell out exactly what counts as a correct answer. The prompting hierarchy is the ordered list of hints, from most help to least. It guides the learner toward that answer. McAmis added reinforcement parameters to that list too. Reinforcement is a reward given right after a correct response. These parameters are the rules for when and how that reward is given. He said the BCBA made those choices for a reason. RBTs do not change them, even when they think they know better.

He tested this with his audience using short examples. In one, an RBT notices a client is getting every trial right. So the RBT simply stops using the prompt. McAmis called that overstepping. It does not matter if the RBT turns out to be right. McAmis said it is still a clinical decision, and the correct move is to tell the BCBA.

Fading a prompt is a real treatment change. So it needs to go through the person who owns the plan.

The "how and where" belongs to the RBT#

McAmis was just as clear that RBTs have real room to be creative. The setting, the activity, and the materials are often open. He called this creative implementation.

If it's not on the program sheet, the toy. The room, the game. You can make a choice on that.From the talk — Noah McAmis

His examples made the line easy to see. One RBT ran a manding program during a game of catch instead of at the table. Manding just means asking for things. The same verbal model prompt was used, so the "what" stayed the same. Another RBT taught receptive color names during an art activity by asking for paint colors. Both counted as creative implementation, not overstepping.

McAmis offered a quick self-check. If the thing you want to change is written on the program sheet, leave it alone. If it is not, you can choose. And if you are unsure, ask before you make the change. He also stressed why this balance matters. Engagement is worth more than one correct response. A rigid session can lose the learner even when every step is perfect.

Report what works back to the BCBA#

McAmis added a step that is easy to skip. When a creative choice works well, the RBT should write it down. Then they should tell the BCBA, saying something like "the client responded really well."

He gave examples like a yoga ball that became a manding tool, or a pretend drum that taught waiting. He framed the report as professional growth, not as asking permission because you are in trouble. RBTs spend more time with the client than anyone else on the team. The BCBA needs to know what they see, because it helps write better programs.

This links back to integrity in a quiet way. When RBTs report their changes, the BCBA knows what was actually done. The BCBA can then decide whether to keep a stricter format or allow the new context. The record stays honest, and the plan can improve on purpose instead of by drift.

Build a checklist for your own setting#

Crystal Harms looked at treatment integrity as a training tool. She spoke about how organizations prepare new RBTs before their formal competency assessment. Her starting tool is the treatment integrity checklist, something most clinicians already recognize.

Harms described a treatment integrity checklist as a task analysis of the skill you want to see. A task analysis breaks a skill into small, ordered steps. She compared it to teaching a child to brush their teeth. You list each step for the child. In the same way, you list each step for your trainee.

Her focus was on embedding these checks during instruction, modeling, and rehearsal. That way, a trainee's skills are assessed bit by bit over time. It is not all saved for one big test at the end. She also named a real barrier. Mid-sized companies often grow by word of mouth and good service. But insurance payment structures can leave them with little capital to build training systems. Harms framed homegrown, task-analyzed checklists as a smart way to work within those limits. Her full session, Press Pause: What the RBT 40-Hour Training Was Really Meant to Do, goes deeper on this approach.

Use the checklist to coach parents#

Penny Holloway brought treatment integrity into parent training. Her team builds a checklist with two layers. The top layer measures the universal protocol, which is the same for every client. The bottom layer holds idiosyncratic items. These are the small moves a clinician uses with one specific client. Holloway's examples were a "funky dance" or putting on a certain song at a key moment.

we put all of this in a treatment integrity for the parents. And during parent training, this is what I do.From the talk — Penny Holloway

Holloway said many clinicians do not know what to do during parent training. Her answer is simple. Everything the clinician does, the parent does too. She then uses the checklist to coach the parent, one item at a time.

Hey, did you offer the empathetic statement? Did you make this a gray EO?From the talk — Penny Holloway

An EO, or establishing operation, is something that changes how much a thing is worth to a person in that moment. Holloway's team calls a "gray EO" a way of "making a sucky moment less sucky." It means adding something pleasant to a hard task. The checklist turns that clinical skill into a step a parent can see and practice.

Supervisors observe and explain integrity#

Tisha Denton teaches new BCBAs how to supervise. For her, observing treatment integrity is a core part of every overlap session. An overlap is when a supervisor watches a trainee work with a client.

Are they implementing things the right way?From the talk — Tisha Denton

Denton described checking whether staff follow the treatment protocol. Supervisors also watch how the client is responding and note any new concerns. But she stressed the next step most. New supervisors must narrate their thinking out loud. She tied this to a grade-school math lesson: show your work. Getting the right answer is not enough if you cannot explain how you got there.

Don't forget to analyze your... visual analysis and and your treatment integrity. And talk about it.From the talk — Tisha Denton

Visual analysis means reading a graph by eye to judge trends. Denton pairs that graph with integrity data. Together they show whether a program needs to change or should stay as it is. She noted that an observation does not automatically mean something must change. When narrating out loud is not appropriate, her team uses a private Microsoft Teams channel instead. That helps protect the client's dignity.

Write programs anyone could follow#

Pauli Catanese and Jasmine Abdel discussed a child's rights approach to ABA in their Article Club session. The article they reviewed tied social validity to treatment integrity. Social validity means the treatment is something the client and family would value and prefer.

And they are designed with the mindset of treatment integrity.From the talk — Pauli Catanese & Jasmine Abdel

The point was about how plans get written. Assessments should be rigorous and use clear definitions. The hosts offered a simple test for any program.

Like if you give this to a random person who's never done ABA before, would they be able to understand the treatment?From the talk — Pauli Catanese & Jasmine Abdel

This flips the usual focus. Integrity is not only about staff following the plan. It also depends on the plan being clear enough to follow. A vague program invites drift, even from careful staff. The hosts also talked about sharing goals in plain, objective terms with parents and teachers. A clear plan helps everyone around the child carry it out the same way.

Measure supervision by integrity, not just hours#

Falon Roberson spoke about using Acceptance and Commitment Training in supervision. Acceptance and Commitment Training, or ACT, builds flexible responses to hard thoughts and feelings. When she turned to measuring whether supervision is working, integrity came first. She said her priority is to:

look at treatment and integrity and procedural Fidelity.From the talk — Falon Roberson

Procedural fidelity is a near-identical term for following the written steps exactly. Roberson also looks for behavioral signs that avoidance is going down. Does the supervisee report their own errors accurately? Do they ask questions when they do not know something? Do they finish their weekly commitments? She suggested that rating scales for burnout or psychological flexibility can help too. But she said to use them only when they serve a clear purpose. Her view keeps supervision grounded in what happens with clients. Logging the right number of hours is not the same as good practice. Integrity data show whether the supervisee's work actually changed.

Where the experts draw the line differently#

The speakers agree on the core. Targets, prompts, and criteria should be run as written. But they differ on what belongs on the integrity checklist itself.

Holloway writes the clinician's personal moves into the checklist. The funky dance and the special song sit on the bottom half of her form. In her approach, those small, client-specific actions are part of the treatment. They get measured, and parents are coached on them item by item.

McAmis draws the line in a different spot. For him, the toy, the room, and the game are not on the program sheet. They are the RBT's creative space. The integrity question covers the "what," such as the target, prompt hierarchy, and reinforcement rules. The "how and where" stays flexible, with good ideas reported back to the BCBA.

Hoskins leans toward the stricter end of this range. She stresses running only what is written and not modifying the plan. She did not carve out a creative zone in the same way McAmis did.

These views are not fully at odds. They reflect different settings and goals. Holloway is coaching parents to copy a working routine. McAmis is training RBTs to keep learners engaged across many contexts. Still, the practical choice is real. A BCBA has to decide how much detail goes on the checklist. More detail can protect what works for one client. Less detail can leave room for staff to adapt in the moment.

What the research says#

The research backs up why these speakers take integrity so seriously. It also shows where the field still falls short.

A parametric analysis tested reinforcer errors on set shares of trials, from 0% up to 50% (Bergmann et al., 2021). A reinforcer error is a mistake in delivering the reward. Errors on 15% or fewer trials, which is 85% integrity, were unlikely to stop learners from reaching mastery. Larger error rates were the ones that risked slowing or blocking learning.

A review of 205 studies found that only 46.83% reported treatment integrity data (Falakfarsa et al., 2021). By contrast, 94.15% reported interobserver agreement, which checks whether two observers record the same thing. Nearly half the studies were judged high risk for implementation errors.

Behavioral skills training raised both the accuracy and the speed of trials delivered by new teaching assistants (Hranchuk & Williams, 2021).

A study of parent-delivered plans tracked parent integrity on each part of a multi-part intervention at home (Nuta et al., 2021). Training plus ongoing coaching supported parents, which fits well with Holloway's item-by-item approach.

FAQ#

What is the difference between treatment integrity and procedural fidelity?

The two terms are often used side by side. Falon Roberson named both when she described what to measure in supervision. In everyday use, both ask whether a procedure was carried out as planned. Many teams use the terms almost the same way.

How do you calculate treatment integrity?

A common method uses a step-by-step checklist. An observer scores each step as done right, done wrong, or not applicable. Then you divide the steps done right by the total steps and multiply by 100. A telehealth DTT study by Lindgren and colleagues (2024) used exactly this method.

What level of treatment integrity is good enough?

Higher is always the goal. In one lab study, reinforcer errors on 15% or fewer trials rarely blocked mastery. That works out to about 85% integrity. Still, that study used undergraduate students and a computer program. Clinical teams should aim for the plan as written and watch the client's data closely.