Prior-Free Sample Size Design for Test-and-Roll Experiments
Kentaro Kawato, Shosei Sakaguchi · 2026
Read Prior-Free Sample Size Design for Test-and-Roll Experiments by Kentaro Kawato, Shosei Sakaguchi (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-02414.txt.
Description
Kentaro Kawato and Shosei Sakaguchi study how large a trial should be when the winner will be rolled out to everyone else. In a test-and-roll design, a decision maker tests on some units, then gives the rest the treatment that did better. A bigger trial improves the final choice, but it also leaves more trial units stuck with the weaker option for a time.
The pair argue that a standard worst-case regret rule can shrink trials too far. They propose a rule that weighs the worst-case gain from adding one more matched pair against its cost. For both simple yes-no outcomes and measured outcomes, the analysis points to a clear benchmark: the trial should cover about a third of the group. No prior belief about effects is needed.
Catalog record
| CREATOR | Kentaro Kawato, Shosei Sakaguchi |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 81 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.02414