The effects of modelling on the contingent praise of mental retardation counsellors.

Five counselors watched a five-minute live demo. The model praised adults with intellectual disability while they brushed their teeth.

Researchers then counted how often each counselor gave praise during the same tooth-brushing routine. They used a multiple-baseline design to be sure any jump was caused by the demo.

Four of the five counselors started praising right away. Their praise stayed high when they later helped clients with a different hygiene task.

The boost lasted at least two weeks with no extra training.

Moss et al. (2009) looked at 55 staff-training studies. They found the biggest gains come from mixing lecture and on-the-job coaching. Dodd et al. (1977) is one of the early proofs that a tiny slice of coaching—just watching a model—can work.

Blackman et al. (2022) seems to disagree. They showed that most trainees need direct feedback, not just observation. The gap is about risk: modeling alone can spark change, but you can’t bank on it for every staff member. Plan to add feedback if praise rates stall.

Yaw et al. (2014) kept the modeling-plus-feedback recipe but shifted the target from praise to data accuracy. Staff again reached mastery fast, showing the combo works across skills.

You can lift staff praise in five minutes. Show a clip or demo in the morning huddle, then measure praise during the next session. If someone doesn’t budge, layer on brief verbal feedback. This low-cost move raises client reinforcement immediately and keeps working for weeks.