Twelve Means Nine Things

An Item-Response Analysis of Fit Outcomes for 246 Garments Labelled 8 to 18 Across Nine Retailers, Against a Size Availability Index That Counts Labels Stocked

A paper from the School of Emergent Priorities treats the printed clothing size as a measurement instrument and estimates its item parameters directly. A panel of 118 wearers produced 4,918 analysable binary fit judgements over 246 garments labelled 8 to 18 from nine retailers, with every label concealed at the point of judgement and allocation balanced under an incomplete block design. A two-parameter item-response model places each garment on a common latent scale: the nominal label accounts for 34.1% of the variance in item difficulty and the retailer for 28.6%, and while the mean step between adjacent labels within one retailer is 0.47 logits, thresholds among garments labelled 12 vary across the nine retailers with a standard deviation of 1.11. Retailer intercepts span 2.48 logits, so garments labelled 10 at one shop sit above garments labelled 14 at another. The decomposition is unmoved by model specification. The size availability index the retailers report against counts labels present at audit, and fit returns are coded to a change-of-mind category excluded from quality reporting.

@misc{slop_iv44yy,
  author       = {Anneke Tolan and Anouk Mensah and Verity Marris},
  title        = {Twelve Means Nine Things: An Item-Response Analysis of Fit Outcomes for 246 Garments Labelled 8 to 18 Across Nine Retailers, Against a Size Availability Index That Counts Labels Stocked},
  year         = {2026},
  publisher    = {Slop University},
  doi          = {10.5555/slop.iv44yy},
  url          = {https://slop.university/outputs/slop-paper-clothing-size-labels-iv44yy/},
  version      = {1.0},
  note         = {Research paper},
}