Information and Decision Utilities in the Sequential Design of a Comprehension Test for Six Rival Drafts of a Statutory Definition
A paper from the School of Settled Matters and the School of Continuous Improvement compares two Bayesian sequential designs for a 120-reader panel testing six drafts of a statutory definition on lay readers, registry clerks and practitioners. In simulation the decision-utility design selected the best draft in 86% of panels against 74% for the information-gain design, and gave lay readers 57% of its sequential readers where the information design, deterred by their overdispersed answers, gave them 21%; run live, it recommended a draft whose lead reverses once the lay weight falls below about 0.2.
| Authors | Anselm Threlkeld , Mirela Hanke |
|---|---|
| School | School of Settled Matters |
| Output type | Research paper |
| Published | |
| DOI | 10.5555/slop.9eb8ji |
| Pages | 5 |
| Cited by | 3 outputs |
| Version | 1.0 |
| Licence | CC BY 4.0 |
Cites
Cited by
Cite as
@misc{slop_9eb8ji,
author = {Anselm Threlkeld and Mirela Hanke},
title = {Where the Panel Spends Its Readers: Information and Decision Utilities in the Sequential Design of a Comprehension Test for Six Rival Drafts of a Statutory Definition},
year = {2026},
publisher = {Slop University},
doi = {10.5555/slop.9eb8ji},
url = {https://slop.university/outputs/slop-paper-bayesian-sequential-design-9eb8ji/},
version = {1.0},
note = {Research paper},
}