A new research poster runs the Review Stewardship Maturity Model against 154 booking-app hosts and finds the tier effect that held for tradesperson reviews does not survive the trip to a new platform.
Scored against 154 booking-app hosts, the five-level tier that explained a tradesperson's star rating explains almost nothing, and the University has revoked its cross-platform certification.
A School of Emergent Priorities team took an instrument built for one setting and ran it, unmodified, in another. What predicted a tradesperson’s online reputation did not survive the trip to a booking platform’s own review score.
A tier that only holds up where it was built isn’t wrong, but it isn’t portable, and we would rather learn that from a controlled trial than from a certification nobody thought to re-test.
— Professor Verity Marris, Director, Trajectory Analytics Group, School of Emergent Priorities
The team scored 154 hosts across eleven metropolitan booking-app markets against STAMP, the five-level stewardship tier first built to grade how deliberately a tradesperson curates their review presence, over a fourteen-week window, then benchmarked the result against the platform’s own record of verified-guest complaints rather than self-reported service quality. Displayed host score rose 0.14 stars from the lowest tier to the highest, against the 0.95-star rise the original tradesperson coefficient predicted. “I built the original rubric to catch a tradesperson quietly managing their own score,” said Dr Bram Ntuli, Lecturer and Convenor of Improvement Grand Rounds, School of Continuous Improvement. “Running it against a platform’s own smoothing was always the harder test, and it lost.”
The School counts the negative result among the clearest signals its portability audits have produced all year. On its strength, the Office of Research Outputs has revoked STAMP’s cross-platform certification with immediate effect, listed the instrument “forum-only” in the Indicator Commons, and opened a review of every instrument certified on a single setting.
The full research poster is available from the University’s research repository under an open licence, doi:10.5555/slop.8yvsk7.