Described, Then Decided

A Factorial Vignette Experiment on Home-Insurance Claim Assessment, and the Consistency Audit That Reads the Same Sentence

A paper from the School of Continuous Improvement reports a pre-registered factorial vignette experiment in which 219 licensed claims assessors, across eleven claims teams of one general insurer, each determined twelve home-insurance files drawn from a resolution-IV fractional design (2,628 determinations). The customer's own description of the cause of loss was varied across three wordings while every assessor-facing physical fact --- loss date, photographs, damage extent, sum insured --- was held identical: acceptance in full runs 70.2, 66.6 and 54.6 per cent across those wordings, a spread of 15.6 points (95% CI 11.4 to 19.8) exceeding that attributable to documentary evidence (6.1 points), claim value band (3.8) and prior claims history (1.4, n.s.) combined. A within-study reproduction of the insurer's own assessment-consistency audit on the same determinations returns 95.8 per cent and does not vary across the wording conditions, and the mandatory justification field cites the physical evidence in 91.7 per cent of determinations against the customer's description in 2.1 per cent.

@misc{slop_j1llhz,
  author       = {Torun Ezeigwe and Dagny Okafor and Solveig Adeyemi},
  title        = {Described, Then Decided: A Factorial Vignette Experiment on Home-Insurance Claim Assessment, and the Consistency Audit That Reads the Same Sentence},
  year         = {2026},
  publisher    = {Slop University},
  doi          = {10.5555/slop.j1llhz},
  url          = {https://slop.university/outputs/slop-paper-the-what-happened-box-j1llhz/},
  version      = {1.0},
  note         = {Research paper},
}