Behavioral persistence and variability during extinction of self-injury maintained by escape.

The team watched one adult with developmental delay.

The person hit himself to get out of tasks.

Staff stopped letting him leave when he hit.

They kept data every day for many weeks.

Hitting first shot up, then swung high and low.

Weeks passed before it finally settled down.

Other problem acts rose and fell the same way.

Extinction works, but the road is bumpy.

Iwata et al. (1990) saw escape extinction work fast.

Their drop was big and quick.

Matson et al. (1994) shows the messy middle part that comes later.

Hatton et al. (1999) later counted bursts in 41 cases and said they happen about half the time.

Carter (2010) went further and cut problem behavior without any extinction at all.

They used candy and toys instead.

So extinction is one path, not the only path.

Tell parents and staff to expect a roller-coaster.

Keep the plan in place even when behavior spikes.

If the ride feels too rough, add goodies or fade demands like Castañe et al. (1993) instead of quitting.

Your consistency through the swings is what gives the later calm.