Maintaining instructional behavior after on-the-job training with process-based performance feedback.

The authors trained special-education staff to run the Distar Language program with autistic students. Training happened right in the classroom while kids were present. After each lesson, the trainer gave quick verbal feedback about the teacher’s steps, not just scores.

They used a multiple-baseline design across three instructors. Once each teacher hit the accuracy goal, the trainer stopped giving process feedback. From then on, teachers only checked their own lesson results.

All three teachers kept running Distar correctly for 15 weeks after the feedback stopped. Student talking and correct answers also stayed high. The key: every teacher had first reached the exact accuracy goal before feedback ended.

Matos et al. (2021) conceptually replicated the finding. They trained psychology interns with BST and saw the same quick, durable gains. Both studies show brief BST produces lasting staff skill.

Laske et al. (2024) extended the model into the remote world. They gave feedback through video instead of in-person and still saw maintenance. The 1996 face-to-face method works, but video is now a proven option.

Berube et al. (2021) used a similar plan: BST first, then a lighter follow-up. They added in-situ training for one child who needed it. The pattern is the same—solid BST up front lets you fade extra help later.

You do not need to hover forever. Hit the accuracy criterion during initial BST, then step back. Teachers can keep quality high by simply graphing their own student data. Use this when you train paraprofessionals, peer tutors, or new BCBAs. Front-load the coaching, then trust the data.