Sequential Validation of Treatment Heterogeneity

TL;DR

This paper introduces a sequential validation method for detecting treatment heterogeneity using martingale-based tests, offering an alternative to traditional cross-validation techniques with various validation metrics.

Contribution

It develops a novel sequential testing framework for treatment heterogeneity leveraging martingale constructions, adaptable to multiple validation metrics.

Findings

Provides a new sequential testing approach for heterogeneity detection

Offers an alternative to cross-validation for validation metrics

Applicable with various metrics like BLPs, GATES, QINI curves

Abstract

We use the martingale construction of Luedtke and van der Laan (2016) to develop tests for the presence of treatment heterogeneity. The resulting sequential validation approach can be instantiated using various validation metrics, such as BLPs, GATES, QINI curves, etc., and provides an alternative to cross-validation-like cross-fold application of these metrics.