Gaussianized Design Optimization for Covariate Balance in Randomized Experiments
Wenxuan Guo, Tengyuan Liang, Panos Toulis · 2025
Read Gaussianized Design Optimization for Covariate Balance in Randomized Experiments by Wenxuan Guo, Tengyuan Liang, Panos Toulis (2025) 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-2502-16042.txt.
Description
Wenxuan Guo, Tengyuan Liang, and Panos Toulis redesign how experiments assign treatments in a 2025 paper. Balanced groups make treatment effects sharper, but older balancing tricks rely on hand tuning and mostly suit two-arm trials. Their trick, Gaussianization, treats assignments as reshaped draws from a multivariate Gaussian cloud. Balancing then becomes smooth optimization over the cloud's covariance.
Fast first-order solvers, adapted from semidefinite programming methods, beat several standard designs, they report. Confidence intervals built for this design style follow. The frame also stretches to treatments given as doses rather than yes or no arms. Simulations across practical cases back the gains.
Catalog record
| CREATOR | Wenxuan Guo, Tengyuan Liang, Panos Toulis |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 130 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-16042