Cluster-level cross-fitting restores valid coverage for survey-weighted TMLE with flexible learners under stratified multistage designs, while single-fit and internal cross-validation versions under-cover.
arXiv preprint arXiv:2106.10577 , year=
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ME 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Cross-Fitted Survey-Weighted TMLE with Design-Based Variance for Causal Machine Learning
Cluster-level cross-fitting restores valid coverage for survey-weighted TMLE with flexible learners under stratified multistage designs, while single-fit and internal cross-validation versions under-cover.