Single-case design for BCBAs: your clinical instinct is already there

Single-case design (SCD) sounds like a research term, but a Board Certified Behavior Analyst (BCBA) already runs the bones of one every week, because the clinical instinct is already there: you set a baseline, you change one thing, you watch the graph, and as Dr. Stephanie Peterson puts it, "our graphs are incredibly convincing" to the people who pay for and approve your work, including parents, clinic directors, school district teams, and managed care reviewers.

What single-case design actually is (in BCBA terms, not textbook terms)#

Strip away the journal language and single-case design is a structured way to ask one question. Did this intervention move this kid's behavior, or did it not? You take a measurement before you start. You start the intervention. You take the same measurement again. You look for a pattern that lines up with when you changed something and when you did not.

The piece that makes it "single-case" is that the kid is their own comparison. You are not running a control group. You are not waiting for a thirty-person sample. The behavior at baseline is the control. The behavior under intervention is the test. The graph is the result.

That is closer to what a BCBA already does on a Tuesday than most textbooks let on. A program book is a baseline. A goal is a target. A weekly progress note is the dependent variable. The missing pieces are the ones a real SCD adds on top: a stable baseline before you start changing things, a clean rule for when you switch phases, and a graph that anyone can read.

If you can write a goal and run a probe, you can run a single-case design. The skill is in tightening the structure, not in learning a new science.

Why SCD is faster than A-B trial and error#

The most common reason BCBAs avoid SCD is time. The classroom is loud, the parent is texting, the schedule shifts every Thursday, and a research-grade structure feels like one more thing on the pile. The reframe Peterson offers is that the time math runs the other way.

if you're just using an A-B design or, you know, sort of playing around without a whole lot of systemization, then typically what happens is you trial and error. And over the course of time, it takes you longer to get to the end result if you ever do it all. Dr. Stephanie Peterson

Here is what that looks like in a real caseload. You try a token board. It kind of works for a week. You add a break card. The behavior changes again, but you are not sure which piece moved it. You add a new prompt level. The parent reports something at home. Now you are six weeks in with no clean read on what is actually responsible for the gain or the regression. Every time you pull a lever, you lose your view of the last one.

A reversal or a multiple baseline gets you to the answer faster because it forces a clean phase change. You either replicated the effect or you did not. You either staggered the intervention and saw the behavior wait its turn, or you did not. You stop guessing. You start adjusting on data.

The trade is honest. SCD costs you a few extra sessions of clean baseline up front. It buys you weeks of not redoing work later.

The four designs you will reach for most#

There are more than four single-case designs, but a working BCBA leans on four of them most of the time.

The reversal design (A-B-A-B) is the simplest. Baseline, intervention, return to baseline, intervention again. You use it when the behavior is safe to let drift back briefly, and when you need to convince a stakeholder that your intervention is what moved the graph. Pulling the intervention and watching the behavior come back is the most clear-cut demonstration you can give.

The multiple baseline design is the one most BCBAs land on for ethical reasons. You start baseline on three behaviors, three settings, or three students at the same time. You introduce the intervention to the first one. When that behavior changes, you introduce it to the second. Then the third. You never have to pull a working intervention to prove it worked. The staggered start is the proof.

The alternating treatments design is for picking between two interventions. You rotate them in a planned way (say, morning versus afternoon, or odd versus even sessions) and you watch which one moves the graph more. This is the design to use when the parent and the school disagree on the plan, or when you are choosing between two reinforcers.

The changing criterion design is for shaping. You raise the bar in steps. The student has to hit five trials before the bar goes to seven. Then seven before it goes to ten. The graph shows the behavior tracking the bar. This is the design that lives quietly inside most fluency work and most token economies, even when nobody calls it that.

You do not need to memorize all four to start. Pick the one that fits the next case in front of you.

Using SCD to convince payers, parents, and school districts#

This is the use of SCD that does not get talked about enough. A clean graph is one of the most persuasive tools a BCBA has, and it works on people who never took a research methods class.

our graphs are incredibly convincing. So even if somebody doesn't have a background in behavior analysis, even if they don't look at data very often, it typically just takes a quick orientation to the X and the Y axis and a visual to be able to demonstrate that, hey, what we're doing is working. Dr. Stephanie Peterson

A parent who is on the fence about continuing services looks at a multiple baseline and sees three lines stair-step down in the same order you introduced the intervention. That tells a story a written progress note cannot. A managed care reviewer who is weighing an authorization sees a reversal that shows the behavior coming back when the intervention paused. That is the kind of evidence that protects hours.

It has meaning to manage care companies. It has meaning to parents. It has meaning to school districts who are paying for your services, to the people who can really be influential on your ability to be impactful in clients' lives. Dr. Stephanie Peterson

The clinical use and the business use are the same use. The graph is the conversation.

How to fit SCD into a session you are already running#

The fear is that SCD turns into a side project. It does not have to. You can fit it into a session you are already running with a handful of small changes.

Pick one target behavior per kid that you actually want a clean answer on. Not every program needs a designed study. Pick the one where the team disagrees, or the one the parent is anxious about, or the one that is driving an authorization conversation. That is your study.

Lock the baseline. Run three to five sessions of pure measurement before you change anything. No new prompts, no new reinforcers, no new instructions. Just count. The instinct here is to start helping right away. Resist it. A stable baseline is the floor every other piece of the design rests on.

Write the phase change rule before you start. Decide in advance what will trigger the move from baseline to intervention, or from intervention back to baseline. Three consecutive sessions at a level. A specific frequency drop. Whatever it is, write it down before you start so you are not negotiating with yourself on a Friday afternoon.

Graph weekly, not at the end. A graph that nobody looks at for a month is a graph that gets surprises baked into it. A weekly graph review with the team turns SCD into a feedback loop instead of a paperwork chore.

Common mistakes BCBAs make their first time#

A few patterns show up the first time a BCBA runs a real single-case design. They are fixable.

The first is starting the intervention before the baseline is stable. The pull to help is strong. If the baseline is still trending or still bouncing around, your intervention phase has nothing clean to land on. Wait the extra session.

The second is changing two things at once. A new reinforcer plus a new prompt level plus a new schedule is not an intervention, it is a mystery. If the behavior moves, you cannot tell anyone which piece moved it. Change one thing per phase. If you want to layer, layer in a second phase.

The third is treating SCD as a one-time event. The strength of single-case design is in replication. The behavior moves once. You pull the intervention. The behavior comes back. You reintroduce it. The behavior moves again. That replication is what separates a story from a finding. Build it in.

The fourth is hiding the graph from the team. The RBT in the room, the parent at pickup, and the teacher in the gen ed class should all be looking at the same graph you are looking at. The graph is the language everyone speaks.

Frequently asked questions#

What is the difference between A-B and a real single-case design?

An A-B design has a baseline and an intervention, and that is it. You can see that the behavior changed when you started the intervention, but you cannot rule out the dozen other things that also changed that week. A real single-case design adds replication, like a return to baseline or a staggered introduction across behaviors, so that the change in behavior lines up with your intervention in a way coincidence cannot easily explain.

Can I use single-case design without IRB approval?

For clinical use with your own caseload, you are running clinical decision-making, not research, and IRB approval is not required. If you want to publish the data or present it outside your clinic, that is the line where IRB review usually enters the picture. Many BCBAs run SCD on every case for clinical purposes and only loop in an IRB for the ones they decide to write up later.

Which single-case design is easiest for a brand-new BCBA?

A multiple baseline across behaviors on a single client is usually the most forgiving first design. You do not have to pull a working intervention to demonstrate the effect, the structure is forgiving of small schedule changes, and the staggered start makes the graph easy to read for parents and teachers. Start there, then add reversal designs as you get more comfortable.

Watch the full talk#

Dr. Stephanie Peterson and Dr. Nicole Valentino walk through the real barriers practitioners run into when they try to do research inside a clinic, including time, IRB access, and writing. They make the case that single-case design is not extra work but the structure that holds clinical work together. If you have ever wanted to publish a case or just run a cleaner study on your own caseload, this is the talk to watch first.