# When Sample-Splitting Choices Matter in TMLE and AIPW: Cross-Fitting, Targeting, and Repetition

arXiv: https://arxiv.org/abs/2610.16192 · HTML: https://mlsanity.com/paper/2610.16192/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 9 of its 24 references that are in the map.
- First posted on arXiv 12 Oct 2026.

## Builds on (most cited first, 9 of 9)

- [Semiparametric doubly robust targeted double machine learning: a review](https://mlsanity.com/paper/2203.06469/) (2022)
- [Using simulation studies to evaluate statistical methods](https://mlsanity.com/paper/1712.03198/) (2017)
- [Efficient Estimation of Pathwise Differentiable Target Parameters with the Undersmoothed Highly Adaptive Lasso](https://mlsanity.com/paper/1908.05607/) (2019)
- [Machine learning for causal inference: on the use of cross-fit estimators](https://mlsanity.com/paper/2004.10337/) (2020)
- [Practical considerations for specifying a super learner](https://mlsanity.com/paper/2204.06139/) (2022)
- [Challenges in Obtaining Valid Causal Effect Estimates with Machine Learning Algorithms](https://mlsanity.com/paper/1711.07137/) (2017)
- [Causal machine learning methods and use of cross-fitting in settings with high-dimensional confounding](https://mlsanity.com/paper/2405.15242/) (2024)
- [Performance of Cross-Validated Targeted Maximum Likelihood Estimation](https://mlsanity.com/paper/2409.11265/) (2024)
- [Three Routes to One Answer: Reconciling AIPW, TMLE, and Double Machine Learning for Applied Researchers](https://mlsanity.com/paper/2609.26142/) (2026)

## Abstract (arXiv)

Variants of targeted maximum likelihood estimation (TMLE) and augmented inverse-probability weighting (AIPW) with machine-learning nuisance models differ in which observations train them, how bias is corrected, and whether splits are repeated. In known-truth simulations of a binary-outcome average treatment effect, eight headline estimators shared each setting's library, inner folds and truncation bound; coverage differences were judged against a prespecified two-percentage-point margin. With a low-complexity library and good overlap (n = 3000), all but a half-sample baseline covered equivalently (0.928-0.939). With one random forest on the primary data-generating mechanism, estimators fitting the propensity in-sample undercovered severely (0.12-0.48), whereas cross-fitting both nuisances raised coverage to 0.91-1.00, two overcovering. Under near-positivity, every AIPW estimator covered better than every TMLE estimator with the unweighted fluctuation (0.907-0.917 versus 0.656-0.806 at the primary setting, all undercovering); looser truncation widened the gap. At the primary settings, pooled-versus-fold-specific and single-versus-double cross-fit constructions (involving four additional estimators) were usually equivalent; under extreme weights, pooled-targeting estimates were 1.7-1.8 times as dispersed as fold-specific ones. Repetition reduced partition dependence but made several intervals more conservative. Two real-data analyses illustrate these sensitivities. Results support cross-fitting both nuisances when empirical-process conditions cannot be justified, and assessing truncation and AIPW-versus-TMLE sensitivity under poor overlap.