Evaluation of treatment integrity errors on mand acquisition.

The team taught children with developmental delays to ask for items they wanted. They used an alternating-treatments design to compare three conditions: perfect delivery every time, 40% correct delivery, and 0% correct delivery.

Each child got the same teaching steps. The only thing that changed was whether the adult gave the right item after the child asked.

Kids only learned to ask when adults delivered the correct item every single time. At 100% integrity, mands appeared quickly.

When integrity dropped to 40% or 0%, no child learned to ask. Partial or random delivery stopped learning cold.

Gauthier et al. (2020) looks like a contradiction at first. They saw that response-interruption worked even at 33% integrity. The difference: they were trying to reduce stereotypy, not build a new communication skill. Stereotypy suppression can survive sloppy delivery; mand acquisition cannot.

Russell et al. (2019) used the same alternating-treatments setup but swapped integrity for prompt type. They also found individual differences—one child learned faster with a verbal prompt, one did not. Together the two studies tell us to test both prompt type and integrity for each learner.

Drasgow et al. (2016) came next and asked, "Which replacement mand should we teach?" They showed kids benefit from having two ways to ask. Alaimo et al. (2015) says teach those options with 100% integrity or the new forms will never appear.

If you run mand training, guard integrity like a hawk. Use a data sheet with a simple yes/no column for correct delivery. Review the last five trials before each session; if integrity slips below 100%, stop and retrain staff on the spot. A moment of accidental non-delivery can erase days of progress for a child who is just learning to speak up.