Reducing job coach assistance for supported workers with severe multiple disabilities: an alternative off-site/on-site model.

Three adults with severe multiple disabilities had jobs in the community. A job coach stood next to them for most of each shift.

The team built a small training room away from the work site. They practiced each work step there first. They moved tables, taped outlines on the floor, and set timers to match the real line speed.

After the worker hit the practice goal three times, the coach moved to the real job. Coach time was clocked every day.

Coach minutes dropped the very first day on the real job. All three workers kept the same output rate.

The drop stayed for the whole study. No extra errors or late pieces occurred.

Ewing et al. (2002) tried a different way to cut coach time. They split one job across three workers so each person did only the steps that needed least help. Both studies reached the same goal—less coaching—using opposite tactics. Pick the tactic that matches your workplace layout.

Lerner et al. (2012) added audio cues after video modeling failed. Their teens with autism met competitive speed in retail jobs. The 2001 off-site practice and the 2012 audio prompts can be stacked: train off-site first, then fade in audio if the worker still needs a prompt.

McMillan et al. (1999) trained residential staff in active support two years earlier. They also saw less hands-on help, but in homes, not factories. The same coach-fading logic works across settings.

You can shrink paid coach time without hurting production. Build a cheap mock-up in a back room or garage. Run the exact job at the exact pace. When the worker beats the speed goal three times, move to the floor and step back. You will see the difference in your timesheet the same day.