Ineligible by Return Code

A Validation-Subsample Correction of Container-Return Success Rates Across Fourteen Reverse Vending Points

A paper from the School of Continuous Improvement treats a container-deposit scheme's reverse vending machines as a diagnostic test for container eligibility and assesses them against a blind hand-sort. Trained sorters classified all 41,392 containers presented in 216 validated sessions across fourteen return points, without sight of the machine's decision. Specificity held at 0.991 while sensitivity fell from 0.881 to 0.724 over eight quarters, and a standard misclassification correction moves the network's published eligible-presentation success rate from 94.2% to 79.1% (95% bootstrap CI 77.4 to 80.6). The divergence is structural rather than behavioural: containers the machine codes as out of scheme leave the denominator against which the machine is scored, and by the final quarter 46.1% of them were eligible.

@misc{slop_iis0cs,
  author       = {Torun Ezeigwe and Thandiwe Solberg and Marit Osayande},
  title        = {Ineligible by Return Code: A Validation-Subsample Correction of Container-Return Success Rates Across Fourteen Reverse Vending Points},
  year         = {2026},
  publisher    = {Slop University},
  doi          = {10.5555/slop.iis0cs},
  url          = {https://slop.university/outputs/slop-paper-hand-sorting-every-container-iis0cs/},
  version      = {1.0},
  note         = {Research paper},
}