A new Slop University paper separates a council's displayed food-hygiene rating series from the inspections behind it, and finds the routine component of a six-year district rise at 0.04 stars.
A decomposition of 8,617 inspections at 1,904 premises puts 0.51 of the 0.79-star rise on re-inspections a premises requests and may withdraw
A district’s displayed food-hygiene ratings rose by four-fifths of a star across six years. The inspections nobody had to ask for account for almost none of it.
Nobody Asks for a Worse One, deposited today through the Office of Research Outputs, takes one metropolitan council’s food-premises register — 1,904 premises and 8,617 inspection events — and separates the rating in the shop window from the rating each premises last received on a routine visit. The district mean moves from 3.62 stars to 4.41. The routine series moves 0.04.
The difference is administrative. A premises may request a re-inspection, and may withdraw that request before the result is published. Of 1,206 requests, 214 were withdrawn, carrying provisional ratings a mean 1.8 stars below the certificate already in the window.
Every entry in that register is correct, and the shape of it was not visible to us from inside the file. It took separating the inspections a premises chose from the ones it did not.
— Dr Fenna Okoro, Senior Lecturer and Convenor, Horizon Register
A shopfront audit of 612 addresses found the median certificate on display 2.9 years old, and a third of them already superseded by a later inspection.
Dr Marek Solheim, Lecturer in the School of Emergent Priorities, located the interest of the case in what the scheme leaves out rather than what it specifies. “There is no procedure by which a rating comes down on its own,” he said. “Nothing here was designed to ratchet. It has nowhere else to go.”
The work was supported by the Anticipatory Capability Fund. The University regards the separation of a published series into the processes that actually move it as work its schools are well placed to do, and welcomes the questions the paper puts to anyone maintaining a public register.
The full paper is available from the University’s research repository under an open licence, doi:10.5555/slop.akmw3y.