A School of Continuous Improvement replication extends the barefoot-bowls-night membership study to men's sheds, community choirs, allotment societies, and netball raffle nights, finding no effect the original single-code study reported.
A pre-registered four-code replication of the barefoot-bowls-night finding turns up no membership effect, and the facility- and plot-allocation formulas built on it keep reading the same headcount regardless.
A finding travels further than the room it was measured in, and the School of Continuous Improvement has spent three years finding out how far. Dr Marit Osayande, Lecturer and Convenor of the Living Dashboard, led a pre-registered attempt to reproduce an earlier finding — that a club’s open-participation night lifts the headcount a funding formula reads while quietly emptying the room the headcount is meant to describe — across four codes the original study never tested.
Working with Associate Professor Casimir Beng of the Adaptive Metrics Lab and Dr Mirela Hanke, Osayande’s team matched 61 clubs that introduced a comparable open-participation event against 68 that had not, following the original study’s own protocol. None of the four codes reproduced the original membership effect, and the pooled result sits close to zero. The counted-participation figure each event produces rose regardless, and continues to weight the allocation formula it was adopted to influence — a use the University’s pre-registration register does not require any code to revisit until three independent replications concur, and this is the first.
A finding that travels to four more codes and comes back empty is still telling you something, even if what it’s telling you is quieter than what we started with.
— Dr Marit Osayande, Lecturer and Convenor of the Living Dashboard
Associate Professor Beng called the result “exactly the kind of finding a pre-registration register exists to protect — one nobody would have gone looking for a second time otherwise.” The School frames the three-replication threshold as the discipline that keeps a single result, in either direction, from being over-read before its own funding formulas act on it.
The full paper is available from the University’s research repository under an open licence, doi:10.5555/slop.xwy7so.