Observed changes in the alertness and communicative involvement of students with multiple and severe disability following in-class mentor modelling for staff in segregated and general education classrooms.

The team placed an expert mentor inside classrooms. The mentor showed teachers and aides how to keep students with multiple and severe disabilities awake and talking.

They used a multiple-baseline design across staff. Classrooms were both special-ed and general-ed. The mentor gave tips, modeled moves, and watched staff try the same moves with students.

After coaching, students looked alert more often. They also started more back-and-forth communication. Gains were real but jumped around from day to day.

Both segregated and inclusive rooms saw the same upward trend. Staff kept using the new moves even after the mentor stepped back.

Frantz et al. (2019) ran a near-copy test in preschool. They coached paraeducators and saw the same bumpy but positive child communication gains. The match says the model travels across age groups.

Balikci (2026) swapped the live mentor for an AI self-coach app. Teachers still hit high fidelity and kids still learned. Tech can scale the same idea when a human mentor is not on site.

Castañe et al. (1993) did an early version thirty years earlier. Brief daily coaching lifted toddler social play. The 2014 study widened the lens to older students and added communication targets.

Gevarter et al. (2025) pushed the coaching chain further. Student clinicians first learned NDBI, then coached caregivers. Big communication gains followed. Cascaded coaching keeps working even when the expert is two steps away.

You do not need to pull staff out for long workshops. Bring the coach into the room for short, real-time hints. Pick one student who zones out. Track alert faces or communication turns for a week. Then invite a peer BCBA to shadow you, model prompts, and give quick feedback. You should see the same lift P et al. saw, without extra funding or sub days.