A new paper from the School of Continuous Improvement tracks 612 children across 26 primary classrooms through a platform-wide clip-chart threshold reset, finding clip movements jumped immediately while the incident notes sent home did not move at all.
A 26-classroom interrupted-time-series study finds a behaviour-chart threshold reset shifted logged clip movements immediately, with no matching shift in the incident notes sent home
An instrument can keep behaving perfectly and still stop telling you anything about the thing it was built to watch. A dated, uniform case like this one is a gift.
— Dr Renke Sabel, Convenor of the Indicator Commons
A software update most classrooms would barely register turned into a clean natural experiment for the School of Continuous Improvement. When a shared clip-chart platform lowered its top behaviour tier’s threshold from five clip-ups to four, on the same day, in all 26 participating classrooms, Dr Sabel and Dr Osei Vandermeer had exactly what a measurement study rarely gets: an intervention nobody staggered, opted into, or timed around a study.
Fourteen weeks of logs later, the two series had gone separate ways. Weekly clip movements per child stepped up within days of the reset and stayed there. The incident notes each school’s front office sends home when a behaviour concern needs documenting did not move, on average or classroom by classroom, whether the teacher leading the room was in their first year or their tenth.
“The reset itself wasn’t the interesting part,” said Dr Vandermeer. “What’s interesting is that a change nobody designed as a behaviour intervention still moved one number and left the other exactly where it started.”
The finding sits inside the School’s standing interest in what an indicator keeps measuring once nobody is watching it for its own sake, the question its Indicator Commons work has long put to the University’s own registers. Sabel and Vandermeer say the recalibration’s cleanliness is unlikely to repeat on demand, and the natural next step is a deliberately staged second reset, to see whether the same gap opens twice.
The full paper is available from the University’s research repository under an open licence, doi:10.5555/slop.ajnsr9.