GAAVI: Global Asymptotic Anytime Valid Inference for the Conditional Mean Function
Brian M Cho, Raaz Dwivedi, Nathan Kallus · 2026
Read GAAVI: Global Asymptotic Anytime Valid Inference for the Conditional Mean Function by Brian M Cho, Raaz Dwivedi, Nathan Kallus (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-2602-08096.txt.
Description
Experiments that run continuously need decisions at any time, not just at a planned endpoint, and standard inference breaks under such peeking. Brian M Cho and two co-authors provide anytime-valid tests for the conditional mean function, the function mapping covariates to expected outcomes, covering global nulls and contrasts between functions.
Under mild conditions, their tests control type-I error at all times beyond a minimum sample size, achieve power one, and attain optimal sample complexity relative to Gaussian location testing. Inverting the tests yields function-valued confidence sequences. Synthetic and real-data experiments confirm power while preserving error rates under continuous monitoring. Experimenters in adaptive trials, treatment assignment, and fairness auditing gain inference they can trust mid-stream.
Catalog record
| CREATOR | Brian M Cho, Raaz Dwivedi, Nathan Kallus |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 126 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-08096