How to Tell If a Research Article Actually Matches Your Client

Picking a research article for your client is about finding the boundaries of the treatment, ruling articles in instead of ruling them out, and being willing to keep a paper on a kid with autism for your own client with Down syndrome when the question you have does not turn on that one variable. A BCBA is a board certified behavior analyst. Most BCBAs throw out good articles over one mismatch. The whole point of this page is to stop doing that.

Demographics Are About Boundaries of Treatment, Not Age and Diagnosis#

The first move most people make is wrong. They open an article, see a participant who is six years old with autism, look at their own client who is nine with Down syndrome, and toss the paper. That is throwing money in the trash. The right question is not "do the labels match." The right question is "which variables, if different, would break this intervention."

That is what boundaries of treatment means. It is the set of conditions under which the intervention does what the authors said it did. Outside those conditions, the bet is shakier. Inside them, the bet is good.

We're going to be looking at boundaries of treatment essentially, right? We're going to be looking at what, when present makes it so this intervention is going to work as we assumed it was going to work. Think about prerequisite skills such as verbal language. From the talk — Matthew Harrington

Notice the example. Verbal language. Not age. Not diagnosis. A prerequisite skill. That is the level you have to read at.

The Six Variables That Actually Change an Intervention#

Here is the short list of what to look at when you are trying to decide if an article fits. Age is on the list. Diagnosis is on the list. But neither is at the top.

1. Prerequisite skills. Can the participant do the thing the intervention assumes they can already do? For a complex FCT study (functional communication training, where you teach the kid a new way to ask), the article might assume the participant has a VB MAPP score in a certain range. VB MAPP is the Verbal Behavior Milestones Assessment and Placement Program. It tells you what language level the kid is at. If the article needs Level 2 and your kid is mid Level 1, the intervention may not run as written.
2. Mode of communication. Vocal, sign, gestures, AAC (a speech device). This changes the steps of almost any verbal-behavior intervention.
3. Setting. In-home, clinic, school, telehealth. The same procedure in a clinic with two adults and one kid is a different procedure in a kitchen at dinner time.
4. Function of the behavior. If the article is about behavior maintained by escape and your kid's behavior is maintained by access to tangibles, that is a real mismatch. The intervention is built around the function.
5. Reinforcement schedule and reinforcers used. If the article ran on edible reinforcers and your client will not eat in session, that is a barrier you have to plan around.
6. Caregiver involvement assumed. Some studies assume a parent runs trials at home between sessions. If your case has zero parent buy-in this week, that step is missing.

Age and diagnosis sit lower than people think. They matter when they change one of the six above. They do not matter on their own.

The Same Variable Matters Sometimes and Not Other Times: How to Tell#

This is the part most people miss. The same variable can be the thing that breaks the article for one question and a non-issue for another. The trick is to ask "does this variable touch the mechanism the intervention runs on."

Take a VB MAPP score. If your question is "how do I run a complex FCT program where I shape one mand into a full sentence," the VB MAPP score is everything. The intervention assumes the kid has the building blocks. If you do not check, you will start a program your client cannot do.

If your question is "which functional analysis should I use, trial-based or latency-based," the VB MAPP score barely matters. Those FAs (functional analyses) run on whether the behavior shows up, not on what the kid can say. Tossing a paper because the participant had a different VB MAPP score on a question like that is a waste.

The rule is short. Ask what the intervention runs on. If your variable touches that mechanism, match hard. If it does not, let it slide.

Rule In, Not Out: Why Perfectionist Matching Kills the Search#

There is a way to do this work that ends with you reading nothing for three weeks. You set the match bar so high that no article clears it. Then your case is still stuck and your supervisor is asking what you found.

The fix is a mindset shift. Rule articles in, not out. Hold them until you have a real reason to drop them. One mismatch on a variable that does not touch the mechanism is not a real reason.

Try to match relevant demographics when possible. But also remember, it's better to rule in articles and rule in solutions to your problem rather than rule out. I don't want you to never include any article that doesn't precisely match the demographics of your client because then you'll just be searching forever and ever. From the talk — Matthew Harrington

If you read fifteen abstracts and rule out fourteen, you might be doing it right. If you rule out all fifteen, you are doing it wrong. The job is to keep the good-enough articles on the table long enough to learn from them.

Translating Score-Talk Into Walking-Around-Talk#

There is a gap between how researchers describe a participant and how a clinician describes a client. Researchers write in scores and acronyms. You think in stories. You have to translate one into the other or the match check is meaningless.

A research paper will say the participant scored a 42 on the VB MAPP and used a Speech Generating Device. Your head needs to turn that into "this kid is roughly where my Tuesday client is, uses an iPad with a Proloquo grid, does not speak in full sentences, points and taps for most asks." That is the level your match check has to live at.

In research, all the researchers I've met look at clients the same way. However, they describe them differently in research because of the demands of academia and journal settings. So rather than talking about kind of what skills they have, like a narrative ways, they're going to refer to a VB MAP or a speech assessment. From the talk — Matthew Harrington

If you do not translate, you will either over-match (rule out a perfectly fine paper because the score wording is foreign) or under-match (keep a paper whose participant is nothing like your kid because the labels happened to overlap). Either failure costs you a case.

Worked Example: Down Syndrome Case vs Autism Participant Article#

Here is the move in real time. The case is a kid with Down syndrome. The article is about a kid with autism. Most BCBAs would close the tab.

Larry, eight years old, looks like he is the only participant. Diagnosed with autism. I would say that we're close enough that this is still worthwhile in terms of reading. Other thing to note, Larry was nonverbal and used gesture. So that's going to be important in the intervention piece. From the talk — Matthew Harrington

Walk through what just happened. The diagnosis did not match. The clinician did not stop there. He looked at the variables that touch the mechanism. Age was close. Communication mode (nonverbal, used gesture) was the variable that actually mattered for the intervention being studied. Two of the three things that move the needle for this question lined up. The mismatch on diagnosis did not touch the mechanism. The article stayed on the table.

That is the rule-in move. You are not pretending the kids are the same. You are checking, variable by variable, whether the differences would break the intervention. If they would not, you keep reading.

The 60-Second Match Check Before You Open the Methods Section#

Before you spend twenty minutes on an article, run this check on the abstract and the participants section. It is fast on purpose.

1. What is the intervention. Write it in one line.
2. What is the mechanism. What does the intervention run on. Communication mode? Reinforcer access? Prerequisite skill? Pick one.
3. What is my client's status on that mechanism. Match, close, or off.
4. What are the secondary variables. Age, diagnosis, setting, function. List them.
5. Any of those secondary variables touch the mechanism for my specific question. Yes or no.
6. If the mechanism matches and no secondary variable kills it, keep the article. If the mechanism is off, drop the article even if everything else looks great.

The whole check should take less than a minute. You are not deciding whether to cite the paper. You are deciding whether to read it. A "keep" here just means it earns a deeper read in the methods.

If you do this on five abstracts and end up with two keeps, you are in a good spot. If you end up with zero, look at how strict you are being on the secondary variables. You are probably ruling out for matches that do not touch the mechanism.

Frequently asked questions#

How do I know if an ABA research article applies to my specific client?

Match on the mechanism the intervention runs on, not on age or diagnosis. Find the variable the intervention depends on (communication mode, prerequisite skill, function of behavior, reinforcement type). If your client lines up there, the article is worth reading even if other things do not match. If the mechanism is off, the article is not a fit even if the labels look identical.

Does the diagnosis in a research article have to match my client's diagnosis?

Most of the time, no. Diagnosis only matters when it changes one of the variables the intervention runs on. A kid with Down syndrome and a kid with autism can both be valid matches for an FCT paper if their communication mode and skill level line up. A diagnosis that does not touch the mechanism is a secondary variable. Do not let it kill the read.

What demographic variables matter most when applying research in ABA?

In order: prerequisite skills, communication mode, function of behavior, setting, reinforcement schedule, caregiver involvement assumed. Age and diagnosis come in only when they change one of those six. The single most useful question is "what does this intervention assume the participant can already do, and is my client there yet."

Watch the full talk#

If you want the rest of the method, the Research Finding Framework, the Key Places Framework for reading a paper in under fifteen minutes, and the worked cases that go with each, the full free CEU is one hour and counts for one general learning credit.

Watch the full CEU: Solving Clinical Challenges with Research