Distilling heterogeneous treatment effects: Stable subgroup estimation in causal inference †

Melody Huang, Tiffany M. Tang, Ana M. Kenney · 2025

Read Distilling heterogeneous treatment effects: Stable subgroup estimation in causal inference † by Melody Huang, Tiffany M. Tang, Ana M. Kenney (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-2502-07275.txt.

Description

Huang, Tang and Kenney introduce in 2025 causal distillation trees, a way to turn black-box estimates of who benefits from a treatment into plain subgroups. Flexible models estimate varied effects well but cannot say who is at risk in words a clinician can use. Their method fits any such model first, then grows a simple tree that mimics its estimates. Theory backs the subgroups' consistency.

Stability diagnostics help judge whether the subgroups can be trusted. On an HIV treatment trial, the method produced steadier, more clinically sensible groups than rival approaches. Interpretability, they argue, need not cost accuracy.

Catalog record

CREATORMelody Huang, Tiffany M. Tang, Ana M. Kenney
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE224 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2502-07275

Browse the full library · About the Tower of Babel