Supervision Articles Deep Dive

Supervision Is a Relationship, Not a Task List

Matt Harrington opens this session with a simple claim. Supervision is a relationship first. The paperwork comes second. He then works through a stack of research on supervision. The topics look scattered at first glance. Bilingual clinicians, breaks, hour tracking, turnover, burnout, and feedback all show up.

One thread ties them together. In every study, someone asked practitioners what they needed. Then someone wrote it down and acted on it. Matt argues that this is the core job of a supervisor. You ask, you listen, and you fix what you can.

That framing changes what a supervision hour looks like. It is not a review of a checklist. It is a working talk about real barriers. Most of the fixes in this session cost little money. They cost attention, planning, and a willingness to hear hard things.

This is the same type of candid relationship that we need to be having with our RBTs, our supervisees, our behavior techs.

From the talk — Matt Harrington

What Bilingual Behavior Analysts Said They Need

The first article is by Martin Loya and colleagues in Behavior Analysis in Practice. It studied the needs of bilingual behavior analysts. The team interviewed 14 bilingual BCBAs or BCaBAs (assistant level certified behavior analysts). They worked in California, Texas, Florida, Arizona, and Connecticut. Each one did a demographic survey and a semi structured Zoom interview.

Matt sets up the problem with his own practice. He speaks only English and works in Florida. He once taught functional communication training (teaching a better way to ask for something) through an interpreter. It took about three times as long. He did not think it worked well.

Three themes came out of the interviews. The first was translation and interpretation load. There was no formal training for translating assessments. On the fly interpreting created constant mental strain. Mapping behavior analytic terms across languages raised ethical gray zones. The second theme was discrimination and microaggressions, including bias about accent and appearance. The third was overwork and blurred roles. It included unpaid translation hours and messy dual relationships (juggling two conflicting roles with the same family).

The participants also named what would help. They wanted translated materials, standardized Spanish assessments, and shared bilingual glossaries. They wanted trained interpreters for intakes, IEP (Individualized Education Program) meetings, and caregiver training. They wanted more bilingual staff so one person is not the only link to a family. They also asked for culturally responsive supervision and training for providers who speak one language.

And when language is a barrier, it is one of the most significant barriers to effective service that can be entered into the equation.

From the talk — Matt Harrington

RBT Breaks Are a Clinical Variable

The second article is by Natasi and colleagues from 2024. It asks a plain question. Do registered behavior technicians actually get breaks? The team interviewed 15 RBTs across clinics, schools, and homes. They ran a quick burnout check first. About 40 percent screened as burned out.

The themes were clear. Breaks helped staff manage stress and refocus. Technicians said breaks protected the quality of care they gave. Several reported that client behavior improved after they had taken a break. Still, guilt and a hero mindset led many to skip breaks or never ask.

Break conditions varied a lot. Official breaks were often rushed or taken in a cold room. Unofficial breaks got improvised during admin time or drive time. Nobody was free to cover high risk moments, so a break could vanish in seconds. Home based schedules were the tightest of all.

Support cut both ways. Coworkers often stepped in and helped. Some supervisors protected breaks well, and some were absent. Matt's point for leaders is blunt. No amount of OBM, or organizational behavior management (behavior science applied to workplaces), fixes a tired technician. If the work setting is unsustainable, the therapy will be too.

Final takeaways, breaks are not a luxury, they're a clinical necessity.

From the talk — Matt Harrington

Teach Fieldwork Hour Tracking Before It Gets Hard

The third article is by Atkinson and colleagues. It trained graduate students to enter fieldwork data using asynchronous online instruction. Matt tracked over 3,500 supervision hours across his own path. He says it never got easy. He thinks supervisors forget that fast.

The study ran two modules. One taught daily hour tracking in an Excel file. The other taught monthly forms, including BACB verification forms and university site forms. Twenty two graduate students took part, with no prior fieldwork. Each module had about 18 to 20 people.

The design was six video lessons and then mock scenarios with color coded guides. Optional Zoom supervision sat on top of that. Mastery was set at 85 percent accuracy. Extra probes checked maintenance and generalization (using a skill in a new situation).

Most students scored 50 percent or less at baseline (the starting point measured before training). About 12 of 18 hit mastery early. Sixteen of 18 got there by the midpoint. All 18 kept mastery, and 16 of 18 generalized to a new scenario. Half of the participants rated the training, and every one of them called it useful. Usefulness scored 6.3 out of 7, and stress reduction scored 6.2 out of 7. Matt's advice is to build this module once and assign it before supervision even starts.

I would argue that being a good behavior analyst has nothing to do with tracking your hours in an Excel spreadsheet.

From the talk — Matt Harrington

Teaching Empathetic Listening With Peer Led BST

The fourth article is by Law and colleagues. It used peer led BST, or behavioral skills training (teach, show, practice, then feedback). The target was active and empathetic listening. Six candidates worked in three dyads (pairs of two) over Zoom. They watched a 35 minute expert video that modeled 14 observable listening skills.

The skills included reflective statements, eye contact, and emotional mirroring. Pairs then practiced together and gave each other feedback. If a pair missed mastery, an expert stepped in for more coaching. The study also checked whether the skills held up in a real interview.

Four of the six reached mastery and generalization without expert help. They got there in under two hours of training. Two of the six still needed the expert. One side effect was both funny and telling. Sessions ran longer than planned because trainees started listening and asking more questions.

Matt's takeaway is about fuzzy skills. Empathy is hard to see and harder to define. This team defined it anyway and taught the parts. He suggests the same move for your most engaging technician. Watch what they actually do, name it, and teach those pieces to everyone else.

We take personality traits, we cut them into observable behaviors, and we teach them and we make sure folks reach mastery.

From the talk — Matt Harrington

Why BCBAs Leave: Turnover Data for Your Supervisees

The fifth article is by Blackman and colleagues from 2024. It looked at variables behind BCBA turnover. The team surveyed 500 BCBAs, and 476 finished. Every participant had left at least one BCBA role. The survey covered job history, reasons for leaving, and workplace factors.

The numbers are worth reading twice. Turnover among direct service providers sits near 44 to 45 percent. In this sample, 82 percent named other workplace factors. Thirty seven percent named too few or too many hours. Twenty percent left for promotions, money, or family reasons.

Burnout showed up for 60 percent of respondents. Of those, 70 percent worked more than 40 hours a week. Heavy caseloads, travel, poor leadership, and thin support came with it. Pay and benefits drove 47 percent. And 35 percent reported ethical concerns, with half of those tied to fraudulent billing.

Matt pulls this article out when supervisees start job hunting. His advice is practical. Set the pay floor your family needs. Then pick the leader you want to learn from. For owners, he says to price the real cost of churn before cutting support roles.

They're fixable, but the organization has to actually take the time to do that.

From the talk — Matt Harrington

Compassion Burns You Out Without Supervisor Support

The sixth article is by Austin and Fisk from 2023. It studied compassion, perceived supervisor support, and burnout. About 1,400 professionals took part. The sample included RBTs, BCBAs, and related service providers. The team used published burnout and compassion scales, not a quick rating.

One finding surprised Matt. BCBAs showed higher burnout than technicians and classroom assistants. He had guessed the opposite. His own theory is the constant switching across a caseload. Each family expects your full attention, and you have already given it nine times that week.

The top burnout predictors were supervisor support, training satisfaction, and organizational support. All three can be fixed without much cash. Supervisor support also changed what compassion did. High compassion with weak support pushed people toward burnout. High compassion with real support looked more like retention.

The sharpest line is about how support lands. Support given and support received are not the same thing. A slow, buggy new tool is not support. A clinic director who only audits is not support either. Ask your staff whether the help actually feels like help.

Are you giving them support to allow them to be compassionate?

From the talk — Matt Harrington

Match Feedback Frequency to Skill Level

The last article is by Theobodu and colleagues from 2025, in the Journal of Applied Behavior Analysis. It looked at how often learners want feedback. Thirty five undergraduates learned two computer tasks with 16 steps. After each step they could request feedback or decline it. A feedback phase followed a baseline with no feedback. Mastery was 100 percent accuracy across three sessions.

Requests tracked accuracy closely. Below 50 percent correct, over 90 percent of people wanted feedback. Between 50 and 75 percent, requests fell to about 78 percent. Above 75 percent correct, only 18 percent asked for it. That drop showed up for 86 percent of participants.

That 75 percent mark matters on the floor. It sits near where many teams call procedural fidelity (how closely a plan gets followed) acceptable. So your technician may stop wanting feedback right before the last stretch. Matt says change the feedback rather than just pushing more of it. Explain why the final 25 percent matters, including intermittent reinforcement (a reward that only comes sometimes).

He also says to walk in with a priority list. When one skill is mastered, move your feedback to the next target. Matt pairs this with Carol and colleagues 2022 on video modeling with voiceover (watching a clip while someone narrates it). In that study, five supervisors or trainees watched a 15 minute video of an eight step feedback process across 10 examples. One expert films it once, and the training can reach many people.

Unsurprisingly, as performance improved, feedback requests dropped.

From the talk — Matt Harrington