Delayed prompting: a review of procedural variations and results.

Green et al. (1987) read every delayed-prompt paper they could find. They looked at kids with delays, kids with ID, and typical kids. They wanted to see which delay tricks worked fastest.

They counted how many trials it took to reach mastery. They also noted who kept the skill later.

Most kids learned the new skill faster when the teacher waited a few seconds before helping. A short wait beat no wait.

But some kids never got it. Only a few studies checked if the skill stuck around weeks later.

Bottjer et al. (1979) had already shown that a 15-second wait prompts spontaneous meal requests. Green et al. (1987) later wrapped that tactic into their bigger review.

Frank-Crawford et al. (2024) looked at 82 DTT studies and found the same hole: almost no one tracks maintenance. The 1987 warning still holds 37 years on.

Hattier et al. (2011) seems to clash at first. Their embedded prompts hurt toy play in typical preschoolers. But they used tiny hidden cues during free play, not the clear delay procedure L et al. reviewed. Different tactic, different result—no real fight.

Petit-Frere et al. (2021) later paired least-to-most prompts with BST and got safety skills to stick for autistic kids. They added probe checks at follow-up—exactly the step L et al. said was missing.

Start your next program with a 3-5 second delay before you prompt. It usually cuts trials. Then calendar a maintenance probe two weeks out. If the skill drops, you will catch it early and re-teach. That one extra probe is the difference between ‘learned’ and ‘lost.’