Translating a Research Article Into a Real BIP

The skill nobody really teaches you in grad school is how to pull the active ingredient out of a paper's procedure section and rebuild it for your kid, your clinic, and your staff. The clearest example: Matt Harrington took a multiple-schedule paper on destructive behavior, kept only the signaled availability principle, layered in percentile shaping so screaming never spiked, and used it to get a three-year-old to sit in a church pew long enough that the family could attend a wedding.

Why the paper's procedure section is not your BIP#

A research procedure is written so a reviewer can replicate it. A BIP is written so an RBT can run it on a Tuesday with two other kids in the room. Those are not the same document. The paper assumes a controlled setting, a trained implementer, and a learner who fits the inclusion criteria. Your client probably matches one of those, maybe.

Matt put it this way when he was talking about a fieldwork student who could not find a research article for a tough case:

It wasn't that she didn't know that we needed to use research. It's that she couldn't extrapolate the principles of behavior change, the valuable nuggets from the research, and then apply them to the totally separate situation where the principles are similar enough.

That gap is the whole job of translation. You are not copying the procedure. You are pulling out the principle and rebuilding around your context.

Pulling the active ingredient out of the intervention#

Every intervention has a load-bearing piece and a bunch of scaffolding. The load-bearing piece is the behavioral principle doing the work. The scaffolding is the room, the materials, the implementer, the data sheets, the schedule.

When you read a JABA paper, ask one question first: if I stripped everything away, what is the one mechanism that produced the change? Was it extinction? A signaled schedule? A prompt delay? A motivating operation shift? Once you can name it in a sentence, you have the active ingredient. Everything else is replaceable.

For Matt's church case, the paper he pulled from used multiple schedules to reduce destructive behavior. The active ingredient was not the destructive behavior protocol. It was signaled availability. One signal means the reinforcer is on, another signal means it is off. That principle does not care whether the setting is a clinic or a back pew.

What changes when you swap the setting (clinic to home to school)#

Settings change three things that matter: the resources, the predictability, and the social rules. A clinic has the resources. A home has the predictability. A school and a church have neither, and they both have social rules that limit what your intervention can look like.

Matt described a transition-tolerance plan that worked beautifully in his outpatient clinic and then fell apart the second the district moved his client to a school placement. Same kid, same skills, same rapport. Different setting. The original paper was a clinic study. The principles still applied. The procedure had to be rebuilt around what the school would actually allow.

The right move was not to abandon the research. It was to ask which variable changed and whether the principle still made sense. Shaping and skills-based treatment still made sense. The materials and pacing did not.

What changes when you swap the implementer#

This is the one that gets missed. Almost every JABA paper has a line near the end telling you who ran the sessions. It is almost always the first author or a doctoral student running every condition.

First author did every implementation session. And I'm like, well, that's a little different from my 19 year old technician who has a background in early daycare settings.

If the intervention only works at 100% fidelity, and 100% fidelity in the paper required a doctoral student, you have to either simplify the intervention or invest heavily in BST before rollout. Both are fine. Pretending the fidelity gap does not exist is not.

A useful test: write down each step of the procedure as you would put it in a BIP. If a step has more than one decision branch, your RBT will not run it the same way twice. Collapse it. Add a flowchart. Pre-train the discrimination. Whatever the paper authors did to keep fidelity high, you have to invent a smaller version of that for your team.

What changes when you swap the learner profile#

Papers report a VB-MAPS level, an IQ, a diagnosis, an age, a language profile. You should read those numbers the way a chef reads an ingredient list. You are not looking for an exact match. You are looking at whether your client is close enough that the principle still has a path to work.

A few things to actually compare:

• Prerequisite skills. PECS will not move a learner who does not have one-to-one matching yet. That is the paper's setting talking, not the intervention.
• Verbal repertoire. An FCT paper trained on full sentences does not translate cleanly to a one-word manner. The principle is the same. The response form has to change.
• History. A learner with two years of inconsistent extinction at home will not behave like the paper's learner who was naive to the intervention. Plan for the burst, plan for the recovery.
• Tolerance. Some learners need micro-shaping where the paper used larger steps. You are not deviating from the research. You are honoring what the response data is telling you.

Writing the BIP so an RBT can run it with fidelity#

A BIP that quotes the paper is not a BIP. A BIP an RBT can run has four parts: a setup section, a step-by-step procedure with no branches, a data sheet that takes less than a minute per trial, and a short list of what to do when something unexpected happens.

Setup should list the materials, the location, the signals, and any pre-session prep. Procedure should read like a recipe. Data should be checkboxes wherever possible. The unexpected list should cover the two or three things you actually saw in baseline and a one-sentence response for each.

If your BIP needs a glossary, your RBT will not run it. Cut the jargon. Replace it with the action the staff should take. "Signaled availability" becomes "green card on table means coloring is available, red card means coloring is put away." Same principle. Different audience.

Worked example: turning a multiple-schedule paper into a church-tolerance plan#

Matt's son was three. He loved running around the back of the church. He hated sitting in the pew. The family was about to lose access to a major source of community support because Sunday was the hardest day of their week.

The paper Matt pulled from was a multiple-schedule study on destructive behavior. On its face, the topography is wrong. The setting is wrong. The age is close but the profile is different. So why use it?

Because the active ingredient was signaled availability. One signal: the reinforcer (running around) is on. Another signal: it is off, and the only behavior available is sitting in the pew.

What we're applying here is extinction at one point and FR1 at the other. And we can't have screaming. And of course, if we put extinction onto running around in the back of the church, we're going to get screaming. So then we'll do some shaping.

The shaping piece came from a percentile-schedule paper. Percentile schedules let you set the reinforcement criterion based on the learner's recent behavior, which keeps the probability of reinforcement near 100% during a fade. That is the math behind micro-shaping. It is what kept the extinction component from triggering a burst.

So the BIP was not "do the JABA multiple-schedule procedure." The BIP was:

1. Signal availability with a clear visual.
2. Start with very short unavailable windows. Use percentile rules to grow the window only as fast as the data allows.
3. Reinforce sitting in the pew on FR1 during the available signal.
4. When the unavailable signal is up, the reinforcer is gone, but the window is short enough that screaming does not get reinforced by escape.

This was able to help them get that real social validity of attending one of their best friends' weddings and having their kids participate.

Two papers. One principle from each. A plan built around the family's actual Sunday. That is the translation skill.

FAQ#

Can I just copy the procedure from the research article into my BIP?

No. A paper's procedure is written for replication in a controlled study. A BIP is written for a real RBT in a real setting. Copy the principle. Rebuild the procedure around your context, your implementer, and your learner.

How do I know which parts of the intervention I can change?

Name the active ingredient first. The active ingredient is the behavioral principle producing the effect. Everything else, the setting, the materials, the pacing, the implementer credentials, is scaffolding you can swap as long as the principle still has a path to work.

What if my staff are not as trained as the implementers in the paper?

Either simplify the procedure so a less trained implementer can run it with fidelity, or invest in BST before rollout. Both are valid. Skipping the question is how good interventions fail in the field.

How much modification is too much modification?

You have gone too far when the active ingredient is no longer present. If you keep the principle intact and only change the scaffolding, you are still doing research-informed work. If you have to change the principle to make it fit, you need a different paper.

What if the setting in the article looks nothing like mine?

Settings are almost never identical. Ask whether the setting was controlling the effect or just hosting it. If the setting was hosting it, swap it. If the setting was controlling it, the intervention may not generalize, and you need to find research that ran in a setting closer to yours.

Keep going#

The full CEU walks through three case studies, the search framework Matt uses to find papers fast, and the parsimony test he runs before any intervention goes live. If you want the worked examples in his voice, the recording is the fastest way to get them.