Evaluation of the high‐probability instructional sequence to increase compliance with multiple low‐probability instructions among children with autism

Rosales et al. (2021) tested the high-probability instructional sequence with three autistic children. They gave two or three easy, high-probability requests right before a harder, low-probability request. The team ran an ABAB reversal design to see if compliance would go up.

Each child faced three different tough instructions during the study. The researchers wanted to know if the brief high-p stack would be enough on its own.

Two of the three kids started complying more when the high-p sequence was in place. The third child only improved after the team added extra reinforcement such as praise and small treats. Overall, the plain high-p sequence worked about two-thirds of the time.

Borgen et al. (2017) got great results for every preschooler in their sample, but they bundled the high-p sequence with lots of reinforcement and skipped most non-compliance trials. Rosales kept the procedure lean and saw mixed outcomes, showing the extras may matter.

Sheppard et al. (2026) also reported mixed effects when they moved the same high-p tactic into the feeding domain. One child ate more, two did not, matching the hit-or-miss pattern Rosales saw with compliance.

Davis et al. (2021) found large, lasting compliance in a child with CHARGE syndrome once they added individually-picked video reinforcers inside the sequence. Their case lines up with Rosales’ third participant: when reinforcement is tailored, the high-p sequence locks in.

You can still start with the simple high-p stack; it is quick and often helps. Just take one-minute baseline data. If compliance does not jump after two sessions, layer in preferred reinforcers or run a 30-second preference assessment first. This saves you from spinning your wheels while keeping the momentum of easy-to-hard requests that kids expect.