Comparison of direct observational methods for measuring stereotypic behavior in children with autism spectrum disorders.

Ciotti and team watched preschoolers with autism in a lab playroom. They compared three ways to score stereotypy: continuous real-time recording, partial-interval recording (PIR), and momentary time sampling (MTS).

Each child was filmed for a set period. Two observers scored the same clips using stopwatches and interval sheets. The goal was to see which quick method came closest to the true duration.

PIR blew the numbers up. It made short bursts of stereotypy look like they lasted almost the whole session. MTS stayed closer to the real time, but it sometimes nudged the number up and sometimes down.

Bottom line: PIR is too inflated to trust for duration. MTS gives a fairer picture, though it still wiggles a bit.

Neef et al. (1986) and Decasper et al. (1977) said the same thing years earlier: PIR overestimates, MTS is safer. Gardenier et al. (2004) now shows the pattern holds for kids with autism, not just general lab subjects.

Sisson et al. (1993) warned that MTS can miss very short, scattered bouts. The 2004 data agree: MTS errors went both ways, but they were smaller than the giant PIR inflation.

Iwata et al. (1990) found 15-s MTS worked well in classrooms. Ciotti’s lab study backs that choice for stereotypy, adding confidence that the same rule applies in tighter autism samples.

If you write goals like “stereotypy will drop from 40% to 10%,” the method you pick decides whether that goal is real or fantasy. Drop PIR for duration targets; use 10- or 15-s MTS and you’ll track true change without the bloat. Your data sheets get cleaner and your treatment decisions get honest.