A new poster from the School of Emergent Priorities tests the Timetabling Office's occupancy forecast, the figure that decides whether a subject's enrolment is capped or grown, against a term of clicker-recorded lecture-theatre attendance.
A term-long comparison across nine lecture theatres finds the Timetabling Office's pre-term occupancy forecast consistently overshoots clicker-recorded attendance, with no correction as the term wears on
Nine lecture theatres, 214 timetabled sessions, and one number the Timetabling Office sets before a single student enrols: a forecast of how full the room will be. A new poster from the School of Emergent Priorities asks how well that number holds up once the term is actually under way.
Dr Fenna Okoro and Dr Ingrid Vasseur set a term of ceiling-mounted clicker counts against the Office’s own pre-term capacity model, tracking nine theatres from the first week of semester to the last. The forecast, once issued, was left to run untouched — the same policy the Office applies to every subject it caps or clears to grow.
Attendance never caught up to what the forecast predicted. Across the nine theatres, recorded occupancy tracked the model only weakly, and the shortfall widened rather than closed as the weeks went on.
A forecast is a bet placed once and rarely revisited. We would like ours revisited a little more often than a term.
— Dr Fenna Okoro, Senior Lecturer and Convenor of the Horizon Register
The University holds that a forecast earns its keep only for as long as it keeps describing the room it was built to predict.
Dr Vasseur, whose work follows how forecasting talk becomes committee business, said the pattern was a familiar one: “the model outlives the moment it was calibrated for, and nobody’s job is to notice.”
The full poster is available from the University’s research repository under an open licence, doi:10.5555/slop.qeq1w5.