A new Slop University research poster tests menu-board kilojoule labelling at 34 food-court counters and reports a null, with the estimate sitting inside a margin of indifference fixed before the data were opened.
A pre-registered equivalence test across 1.84 million transactions rules out a change larger than 100 kilojoules either way
One point eight four million itemised till records, gathered fortnight by fortnight from thirty-four food-court counters, sit behind a result the School of Continuous Improvement has published as a finding of no difference.
Reached, and Unmoved, deposited today through the Office of Research Outputs, follows 18 outlets across the fortnight their menu boards began carrying kilojoule figures, against 16 matched counters in the same food courts that did not. Mean energy per transaction moves by seven kilojoules, downward, with a 90 per cent interval running from minus 41 to plus 27. The margin of indifference, 100 kilojoules either way, was lodged before the team was given the data, and the estimate sits well inside it.
A study that finds nothing is usually a study that could not have found anything. This one was built the other way round. We said in advance how small a difference we would be willing to call nothing, and then we went and looked.
— Dr Solveig Adeyemi, Lecturer, School of Continuous Improvement
Alongside the tills, the team stopped 412 diners at the same counters. Three in five recalled seeing the figures, and one in thirteen said the figures had changed what they ordered.
“People are not misremembering,” said Dr Joost Nwosu, Postdoctoral Fellow in the School. “They read the number, they can tell you they read it, and the register records the same lunch either way.”
The health district’s healthy-choices indicator counts consumers reached — transactions at premises displaying the figures — and climbs with every outlet that comes into scope. The 1.84 million transactions behind the study are the same 1.84 million the indicator counts.
Slop University takes some satisfaction in depositing a null on the same footing as everything else it publishes, and the School regards the margin, rather than the estimate, as the contribution.
The full poster is available from the University’s research repository under an open licence, doi:10.5555/slop.puo2j3.