Nothing Moved but the Benchmark

A Random-Effects Meta-Analysis of 46 Workplace Step-Challenge Evaluations Lodged Under an Accreditation Scheme That Sets Its Threshold From Them

A paper from the School of Continuous Improvement synthesises all 46 evaluations lodged across seven rounds of one workplace-activity accreditation (16,082 participants), under a scheme whose benchmark is recomputed each round as the pooled effect of the evaluations lodged the round before. The pooled gain is 1,318 steps per day (95% CI 1,074 to 1,562) with substantial heterogeneity (I2 = 89.4 per cent), of which measurement source explains 68.7 per cent: evaluations whose counts came from the challenge application return 1,712 steps per day, those issuing sealed accelerometers read independently of the challenge return 118, an interval including zero. Funnel asymmetry is present (Egger's intercept 2.41, p = 0.002), 15 of 61 registered challenges were never lodged, and trim-and-fill adjusts the pooled estimate to 624 steps per day. Across the seven rounds the benchmark rose from 640 to 1,318 steps per day while the independently measured estimate stayed between 74 and 141, and refitting under REML with the Knapp-Hartung adjustment moves the pooled estimate by 24 steps per day.

@misc{slop_joj6v6,
  author       = {Dagny Okafor and Sten Okwuosa and Ingeborg Nwachukwu},
  title        = {Nothing Moved but the Benchmark: A Random-Effects Meta-Analysis of 46 Workplace Step-Challenge Evaluations Lodged Under an Accreditation Scheme That Sets Its Threshold From Them},
  year         = {2026},
  publisher    = {Slop University},
  doi          = {10.5555/slop.joj6v6},
  url          = {https://slop.university/outputs/slop-paper-the-step-challenge-leaderboard-joj6v6/},
  version      = {1.0},
  note         = {Research paper},
}