Procedural Fidelity: Running the Plan as Written

Procedural fidelity is how closely a plan gets run the way it was written. When a behavior plan says do five things in order, fidelity asks a simple question. Did all five happen, every time, the right way?

This matters because a plan on paper is not the same as a plan in the room. If the plan is sound but nobody runs it correctly, the data will look bad anyway. You could throw out a good intervention for the wrong reason. Fidelity protects both the client and your own conclusions about what worked.

What procedural fidelity actually means#

Different experts use different names for the same idea. You will hear procedural fidelity, procedural integrity, and treatment fidelity used almost the same way. Dr. Jessica Osos keeps the definition plain. It is the extent to which an intervention is implemented as it was designed.

That definition sounds obvious, but it changes how you read your data. Before you decide a plan failed, you check whether the plan was even run right. Two things get measured. First, was the intervention done. Second, how often and how well it was done.

Fidelity separates a dosage problem from a bad plan#

Matt Harrington makes a sharp point about why the count matters. Knowing that something happened is not enough. You need to know how much of it happened and what changed as a result.

Picture a caregiver running homework from an occupational therapist, a specialist who works on daily-living and motor skills. If they run it half the time and behavior does not move, that is a dosage problem. If they run it all the time and behavior still does not move, the plan itself may be wrong. Fidelity data is what lets you tell those two cases apart.

Everyone agrees it matters, but few track it with data#

Here is the gap. Fidelity is treated as basic, yet real fidelity data is rare. Harrington points to a survey of 140 BCBAs, board certified behavior analysts. Almost all of them gave feedback in the moment. Very few of them graphed any of it.

90% provided feedback consistently. Almost all of it was in-the-moment vocal feedback. But only 20% used graphs to track their feedback. From the talk — Matt Harrington

Feedback that lives only in a hallway chat disappears. Without a number on a graph, you cannot show the plan matched the data. Harrington frames it as a trust problem. If there is no integrity data, how do you know the write-up matched what was actually done?

Low fidelity poisons the graph before you even start#

Several talks land on the same warning. Weak fidelity wrecks your data at the very start. Harrington says a graph will never look clean if the fidelity and data integrity are not there. Poor fidelity is one of the most common reasons an intervention looks like it is failing.

Dr. Shane Spiker takes it further into safety. Poor fidelity does not just slow progress. It can create crisis moments. Spiker warns that teams with poor treatment fidelity can re-traumatize the very people they serve. A sound plan, run wrong, can become the source of harm.

So fidelity goes beyond clean charts. Running a plan wrong can harm the person you are trying to help.

It is a system, not a checkbox#

The strongest theme across these talks is that fidelity takes real, ongoing work. Dr. Osos is blunt about the mindset. To her, fidelity is far more than a compliance checkbox.

It's how we protect our clients and our own conclusions and our own services. From the talk. Dr. Jessica Osos

What does the system look like in practice? Osos recommends direct observation with component level checklists. You watch the session and score each step of the plan on its own. Mellanie Page describes monthly integrity checks on learner plans. The supervisor rates how closely an RBT, a registered behavior technician, follows the written plan, then gives feedback. Lauren Donovan builds procedural integrity right into the therapist observation form, so the supervisor checks for it during every observation.

Where BCBAs can bring fidelity into team care#

Fidelity is also useful outside your own sessions. Harrington calls caregiver fidelity tracking the one thing he would recommend most in cross-team work, like partnering with an occupational therapist.

I really like fidelity tracking for OTs. If there was one thing that I would recommend, it is helping the caregiver self-track fidelity to OT interventions. From the talk — Matt Harrington

The idea is simple and it travels. Teach the caregiver to log how often they run the homework. Then you can tie that count to the target behavior. That turns a vague "it's not working" into a clear read on whether the plan was ever tested. The same self-tracking habit helps when a plan has to hold up across home, school, and clinic.

What the research says#

The research backs up the idea that fidelity can be taught, watched, and even automated.

Paden and Carroll (2024) found that video self-monitoring raised staff fidelity and held it high. It worked even during covert observations, when no supervisor was watching (Journal of Applied Behavior Analysis, 58(1), 213-224). That last part matters, because fidelity often drops when staff think no one is looking.

Training also helps people see errors in the first place. Katechis, Ruiz, Schnell-Peskin, and Livanis (2026) used behavioral skills training, a coach-model-practice method, with RBTs. After training, staff could spot both correct steps and fidelity errors. The task was discrete trial teaching, a structured format that teaches one skill at a time (Behavioral Interventions, 41(1)). Staff cannot fix an error they cannot see, so training the eye comes first.

Newer tools also help novices. Yagafarova and colleagues (2025) tested an AI platform that guides people through an intervention step by step. It raised fidelity for both technicians in training and caregivers with little prior training (Behavioral Interventions, 40(1)).

Not every fidelity error hurts the same amount. Diffley, Kubina, Noone, and Mc Tiernan (2025) studied Precision Teaching, a method that charts learning speed and uses the chart to guide decisions. Learners still did well in the short term when one decision step was skipped some of the time (Behavioral Interventions, 40(4)). That does not make fidelity optional. It means some steps carry more weight, so find those steps and guard them hardest.

Frequently asked questions#

What is the difference between procedural fidelity and treatment integrity?

They point at the same idea. Both ask how closely an intervention was run as designed. You will also see procedural integrity used the same way. Pick one term and stay consistent inside your team so nobody gets confused.

How do you measure procedural fidelity?

The recommended method is direct observation with a component level checklist. You break the plan into steps and score each step as done correctly or not during a real session. Then you turn those scores into a percentage and graph it over time. A count in the moment is fine, but the graph is what makes it usable.

Why does fidelity drop when a supervisor is not present?

People tend to run plans more loosely when they think no one is watching. That is exactly why tools like video self-monitoring help. They give staff a way to check their own work. Fidelity then stays high during covert observations, not just when the boss is in the room.

To see fidelity built into a full supervision workflow, watch Treatment Integrity Is Not a Checkbox: Building Meaningful Fidelity Systems. For the caregiver self-tracking angle in team care, see New Year, New Care Collab Goals.