Article Club: Troubleshooting Behavioral Interventions in Schools
Free · 1 Learning CEU · certificate included
A review of the concepts covered in "Troubleshooting Behavioral Interventions: A Systematic Process for Finding and Eliminating Problems" (Witt, VanDerHeyden, & Gilbertson, 2004), and a simultaneous discussion with current clinical practitioners of ways the concepts apply to our clinical practice. Join us to see where the discussion goes. Leave with research-based takeaways you can implement immediately.
Register free & watch nowClaim your certificate as soon as you pass the quiz.
Presented by Pauli Catanese & Morgan Johnson
About this CEU
Article Club: We read the article so you don’t have to! Join our discussion of how to apply research to clinical practice. This Session's Article: Troubleshooting Behavioral Interventions: A Systematic Process for Finding and Eliminating Problems (Witt, VanDerHeyden, & Gilbertson, 2004) Instructor: Pauli Catanese Co-Presenter: Morgan Johnson Article Club Format: A review of the concepts covered in a research article in the field of ABA, and a simultaneous discussion with current clinical practitioners of ways the concepts apply to our clinical practice. Join us as we chat, share experiences, and learn. Leave with research-based takeaways you can implement immediately. We are a group of currently practicing clinicians devoted to professional growth through research-based learning and experience-based discussion. 2611
Abstract
A review of the concepts covered in "Troubleshooting Behavioral Interventions: A Systematic Process for Finding and Eliminating Problems" (Witt, VanDerHeyden, & Gilbertson, 2004), and a simultaneous discussion with current clinical practitioners of ways the concepts apply to our clinical practice. Join us to see where the discussion goes. Leave with research-based takeaways you can implement immediately.
Credits
1 CEU
From the talk
What was covered
A working guide to the four-level Behavioral Intervention Troubleshooter: what to check, in what order, before you change a behavior plan.
- The BIT, or Behavioral Intervention Troubleshooter, is a checklist for fixing stalled plans. Work through its four levels in order before you scrap a plan that looks like it failed.
- Check the behavior definition and data system first, since a flawed definition can mimic a failed plan.
- Check treatment integrity (how closely the plan is followed) by watching it run. Self-report (what staff say they did) often does not match what you observe.
- Replacement behavior (the skill meant to replace the problem behavior) and the problem behavior can both stay high at once. When that happens, suspect the problem behavior is still getting reinforcement (whatever keeps a behavior going).
- Suspect a change in function (what the behavior gets someone) when a plan works for weeks and then stops working.
- Build classroom plans on top of what the teacher already does, so the steps are realistic to run.
The Flat-Data Scenario That Opened the Session
The talk opened with a case that feels familiar. You are a BCBA (a certified behavior analyst) who supervises an RBT (a staff member who runs the plan). A client's aggression has not dropped in three weeks on the plan you wrote. The RBT says they follow the plan, but the data show no progress. What do you do first?
Most of the group gave the same answer: there is not enough to go on yet. We do not know if the plan is being run as written. We do not know if the counts are accurate. We do not know if the client gets the planned reinforcement. Swapping the plan right away feels right when the data look flat. One presenter said that pull felt strongest early in their own fieldwork.
The article argues for a pause. Check what you control first: your definitions, your data, and the room. Then look at the plan itself. The paper is not a study with groups and graphs. It reviews past work and hands you a checklist you can use on Monday.
It's just a conceptual review of previous literature and a proposed checklist for troubleshooting.
From the talk — Pauli Catanese & Morgan Johnson
The Four Levels of the Behavioral Intervention Troubleshooter (BIT)
The BIT has four levels, and each one has a job. Level one is problem definition and data systems. Level two is classroom fundamentals. Level three is intervention integrity, meaning how closely the plan is followed. Level four is intervention design.
Each level holds a set of yes or no questions. You start at level one. If every answer is yes, you move down a level. A no tells you to stop and fix that thing first. The chart looks busy at first glance. The logic under it is plain.
Order matters for a reason. Level four is the designer's level, which is your own work as the analyst. Levels one through three often sit with other people. In schools, that means teachers, aides, and daily routines. The presenters noted how hard those third party variables are to control from the outside. The BIT gives you a way in. It turns "this is not working" into a short list of things to check.
You literally just go through the questions, and if they're all yes, you go to the next and continue on.
From the talk — Pauli Catanese & Morgan Johnson
Why the Behavior Management Pyramid Comes Before Any Individual Plan
Before the checklist, the article lays out a behavior management pyramid. The point is simple. A plan for one student rests on how the whole room runs. The bottom layer is academic skills taught well. Next come behavior expectations that are taught, not assumed. Next are adult responses to problem behavior that stay steady and accurate. At the top sits a functional assessment (work that finds why a behavior happens), so the plan matches the true function of the behavior.
This changes how you read what you see. Say a worksheet goes out and half the class checks out. That is not a one student problem. A general education classroom also lets you compare your client to same-age peers. One presenter said it is hard not to step in when you see management errors all over a room. The pyramid gives you something better than frustration: a list the teacher can work through.
The group then took up a sharp question from the article.
The answers kept coming back to information. If no one states the rule, the student is not breaking it. One member put it in adult terms: you cannot be late to a 5:30 meeting no one told you about. Another raised how praise gets spread around a room. In some classes, attention for redirection far outruns attention for on-task work. Students who need attention will take the cheaper route. A good fix is teaching the student to ask: raise a hand and ask what to do.
could a student's challenging behavior actually be an appropriate response to an inappropriate instructional environment
From the talk — Pauli Catanese & Morgan Johnson
Level One: Problem Definition and Data Systems
Level one asks a blunt question. Can you even tell what is going on? Is the target behavior clearly and objectively defined? Are the data reliable? Was baseline data (counts taken before the plan starts) collected? Are goals set? Are the data sensitive enough to show real change? Frequency counts can be a trap here. If staff must tally 80 events a day, the count will not hold up.
One clinician shared a case that lives at this level. They took over a client on functional communication training, or FCT (teaching a request to replace behavior). Treatment integrity was good. The replacement request was being reinforced. Still, precursor behavior (small signs before the big behavior) stayed very high. So they read the definition. It had a full problem behavior folded into it. That was two behaviors, not one. They pulled it out. Two weeks later the data fell from high to low or moderate. No new plan. No change to the plan at all.
Level one also cleans up vague reports. A teacher says it happens all the time. You ask how often. Every ten minutes? Every five? This week, or last week? The answer starts to look like real baseline. Good data also buys trust. You can show a teacher a falling trend even while the behavior still shows up.
So without adequate definitions and data, there just may not be enough information to troubleshoot an existing intervention.
From the talk — Pauli Catanese & Morgan Johnson
Level Two: Classroom Fundamentals and the Math Group Case
The group worked a short case. A teacher reports that a client keeps leaving their seat during math group. You go observe. Several students are out of their seats. The teacher gives little instruction on the worksheet. Redirection comes as yelling to get back in the seat. The group called it fast. This is a level two problem.
Their fixes focused on the room, not the client. State the expectation before the task starts: sit down, finish the sheet, then move on. Teach the worksheet first. Model a few problems. Pull students up to answer at the board. Offer help to any raised hand. Then praise students for staying in their seat. One presenter named a pattern worth watching: schools often skip praise for expected behavior, so only problem behavior draws attention. Another idea was to post the expectations on the board. Then a redirection becomes a quiet point at the list.
What makes this level usable is how the questions are written. Are classroom rules posted? When observed, is overall appropriate student behavior above 70 percent? Are rule violations enforced per the class-wide plan 100 percent of the time? Do students move between activities in under four minutes? These are countable. They also read as fair feedback you can hand a teacher: here is what works, here is the next step.
One member described a fix built on what the teacher already ran. The class used points handed out after directions. A token for every replacement behavior was not realistic there. So they asked what was. The answer was once per academic period. The client now earns a favorite sticker for staying seated and doing part of the work. The teacher runs it. The client works the whole period. It also cut the need for one-to-one support all day long.
they're quantifying specific things that you can observe
From the talk — Pauli Catanese & Morgan Johnson
Level Three: Intervention Integrity and the Limits of Self-Report
Level three asks if the plan is being run as written. The article is firm on method. Measure integrity directly. Do not lean on self-report alone. The paper points to a study where reported practice and observed practice did not line up.
The group spent time on why that happens. It is rarely on purpose. Often the person was never told what the plan requires. Training may have been one sitting with no follow-up. Some do not see the payoff yet, which is why sharing data early helps so much. Some are simply busy. A teacher with 30 students will not run a token system with a rule for every token. Plans also get updated. If no one tells the staff, they keep running the old version. Drift (slow change in how a plan is run) does the rest.
The article pairs each problem with a fix. Not enough training: model and rehearse the steps. Behavioral skills training, or BST (tell, show, practice, then feedback), is the tool here. Missing materials: fix access first, because no one can run a token system with no tokens. Plan too complex: cut steps. Uneven delivery: observe, then give performance feedback and graph the data. Integrity sliding over months: that calls for ongoing checks and support, not a brand new plan.
the teacher reported implementation did not reliably reflect observed implementation
From the talk — Pauli Catanese & Morgan Johnson
Level Four: Intervention Design and Reading Data Patterns
Level four is the plan itself. You arrive here only after the first three check out. The questions get at fit. Are the maintaining contingencies (what keeps the behavior going) named correctly? Is the plan built on the function? Is a replacement behavior being taught? Does that replacement get real reinforcement? Does it take less effort than the problem behavior? Have the contingencies in the room changed? The presenters also warned about plan reuse. A token system that worked for one client can fail for the next, because the function is not the same.
The article maps data patterns to next steps, which the presenters called a favorite page. Problem behavior high and replacement low: look at how you teach the replacement, how much effort it takes, and whether the problem behavior still pays off. Both high: the problem behavior is still getting reinforced somewhere. Watch for split practice, where one adult reinforces the replacement and another reinforces the problem.
Both low: look at opportunity. If you want hand raising instead of calling out, you have to set up chances to practice, even on days with no group work. Problem behavior low and then rising, with the replacement bouncing around: suspect a shift in contingencies, or an integrity slide. The patterns overlap, but each one gives you a place to start.
Working the BIT Through Two Cases
The article closes with a case the presenters walked through. A class ran a response cost system (losing tokens for rule breaks). Students began each week with ten tokens. Each rule violation cost a token. Six tokens left on Friday earned game time. It worked in October. In November, one student's behavior climbed. Level one: the behavior was defined and measured well. Level two: most students did the academic work accurately, and this one student accounted for nearly all the violations. Level three held the answer. The teacher removed tokens only 40 percent of the time.
The fix was not a new plan. Students were retrained on the rules and the consequences. The teacher got coaching and support. Steady delivery brought disruptive behavior back down. The presenters drew out the cost of the other path. A new plan takes new design time. It means teaching a whole class a new contingency, retraining the teacher, and collecting fresh data. You still would not know why the first plan stalled.
The last case came from the group. A client destroys property to escape tasks (behavior that gets someone out of something). FCT teaches a break request. Break requests rise. Property destruction stays high. Treatment integrity is high. Both behaviors are high, so the first suspect is ongoing reinforcement for the problem behavior. One member raised a second path: the function may have widened. In a case they shared, screaming started as escape and later served attention and access to items. The plan then had to cover the new functions.
The closing frame was four questions. Are we defining and measuring the behavior well? Does the environment support the right behavior? Is the plan being run correctly? Is the plan designed for this problem? One more point landed with the group. The BIT shows you where to ask for help. New analysts often feel they must solve it alone. A level tells you what to ask about.
Common questions
What are the four levels of the Behavioral Intervention Troubleshooter?▾
Level one is problem definition and data systems. Level two is classroom fundamentals, meaning instruction, expectations, and consistent adult responses. Level three is intervention integrity, or whether the plan is run as written. Level four is intervention design, which asks if the plan fits the function of the behavior.
What should I check first when a behavior plan stops working?▾
Start at level one, even when the data clearly show no progress. Confirm the behavior is defined objectively and that the data system is reliable and feasible. Check whether baseline data exist at all. Without that, you do not have enough information to troubleshoot anything else.
How do I know if the problem is treatment integrity and not the plan?▾
Observe the plan being run instead of asking whether it is being run. The article cites a case where reported practice did not match observed practice. Collect integrity data, graph it, and give performance feedback. In the article's case study, the teacher was removing tokens only 40 percent of the time.
What does it mean if both the problem behavior and the replacement behavior are high?▾
The most likely cause is that the problem behavior is still getting reinforced somewhere. Check whether different adults respond differently, since that can keep the behavior going in one setting or with one person. Also consider whether the function has widened to include something new, such as attention on top of escape.
Does this troubleshooting process only work in classrooms?▾
The article was written for classroom-based plans, and the presenters said so directly. Level two in particular covers class-wide instruction and management. Still, one presenter said much of the process carries over to other settings. That includes clinical work outside the classroom, since definitions, data, integrity, and design apply anywhere.
About the speaker
Pauli Catanese and Morgan Johnson led this Article Club session on troubleshooting behavior plans in schools. They walked the group through the four levels of the Behavioral Intervention Troubleshooter. They also covered the behavior management pyramid and the article's response cost case study. They work in school settings themselves and shared their own cases involving teacher-run systems, treatment integrity, and shifting behavior functions.
This summary was generated from the recording’s transcript. Quotes are taken word for word from the talk.
What you'll learn
- 1Identify the four levels of the Behavioral Intervention Troubleshooter (BIT)
- 2Systematically evaluate why behavioral interventions may not be working
- 3Differentiate where the problem in the intervention lies
- 4Identify data-based modifications that can improve the effectiveness of an intervention before re-starting
- 5Apply the troubleshooting process to real-world clinical scenarios
Related talks
Talks that cover the same concepts
Article Club: Assent in ABA- Ethical Considerations and Practical Recommendations
Achieve Behavior Partners
Article Club: A Child's Rights Approach to ABA
Achieve Behavior Partners
Article Club: Using Deposit Contracts and Reinforcement Thinning to Increase Physical Activity
Achieve Behavior Partners
Article Club: Cultural Awareness for Behavior Analysts
Achieve Behavior Partners
Article Club: Treatment Integrity Errors During Discrete Trial Instruction
Achieve Behavior Partners
Article Club: Selecting Measurement Procedures
Achieve Behavior Partners
More from Achieve Behavior Partners
OpenCEU is supported by advertising from third-party sponsors. Sponsors are clearly labeled, have no influence over course content, instructors, or CEU decisions, and never appear on certificates. The BACB does not sponsor, approve, or endorse OpenCEU sponsors or their products.