A new Slop University paper tests whether a warehouse's oral food-safety checkbox predicts what workers still know three months later, finding the tick accurately records that a supervisor asked something, but nothing about the worker's answer or what became of it.
A three-month retest across 900 warehouse workers finds the oral-check tick predicting nothing about what they still know
A shift supervisor asks a warehouse worker two or three food-safety questions, listens to the answers, and ticks a box confirming the exchange took place. The box does not say what was asked, what was answered, or how long the conversation ran; a new Slop University paper tests what everyone downstream assumes the tick is standing in for anyway.
The box was never built to hold an answer. It was built to hold a supervisor’s word that a question got asked, and everyone reading it since has added a promise the form itself never made.
— Dr Ingeborg Nwachukwu, Lecturer and Convenor, Capability Sprints, School of Continuous Improvement
Working with a national grocery-distribution operator across 24 warehouses, the School of Continuous Improvement team compared three-month written-retest scores for 900 workers against whether their original oral check had been ticked, finding no meaningful difference between groups. An audio-consented sample of 258 sessions showed the tick doing exactly one job well: predicting whether a conversation happened at all, with near-perfect accuracy. It said nothing about how long that conversation ran, or what became of it three months on.
“What we found almost admirable about the checkbox,” said Dr Dagny Okafor, Lecturer and Convenor, Evaluation of Evaluation, “is that it never claimed to measure retention. That claim got layered on afterwards, by everyone who read the form.”
The paper joins a run of recent University findings on records that hold up as evidence an event occurred while saying nothing about its substance, from a badge tap standing in for who is at a desk to a compliance photograph standing in for a condition it measured once — a distinction the School has made increasingly central to how it audits its own compliance instruments.
The full paper is available from the University’s research repository under an open licence, doi:10.5555/slop.amx2a8.