A Preliminary Analysis of Mastery Criterion Level: Effects on Response Maintenance

Fuller et al. (2018) asked a simple question. Does it matter if we say a child has mastered a skill at 50%, 80%, or 90% accuracy?

Three autistic students in a public school learned new tasks through discrete trial training. Each child worked on three different skills. One skill had to reach 50% correct before moving on, the next 80%, and the last 90%.

Later the team checked each skill once a week to see which ones the kids still used.

Skills that met the 90% bar stayed strong during weekly probes. Skills that met only 50% or 80% faded much faster.

In plain words, the higher the mastery line, the longer the skill lasted.

Bigby et al. (2009) looked at how we record data and saw no difference in maintenance whether you score every trial or just the first. Fuller now shows the mastery level itself does matter, so the two studies team up: pick any data method you like, but keep the 90% cut-off.

Dunham (1972) ran a college class with an A-or-fail rule and got near-perfect pass rates. Fuller brings that old idea into special-ed DTT: high mastery still pays off almost fifty years later.

Matson et al. (1989) warned that self-management studies often fail to prove all-day maintenance. Fuller gives a clear fix: raise the mastery criterion instead of hoping self-monitoring will carry the load.

If you run DTT sessions, stop accepting 80% as good enough. Push each program to 90% correct for at least two consecutive days before you fade prompts or move to the next target. You will spend a few extra trials up front, but you will save time later because you will not have to re-teach the skill. Your learners will also look more fluent to teachers and parents during weekly reviews.