Inference after selection on randomized p-values
Abstract
We consider inference on parameters and hypotheses selected using p-values. We provide a simple procedure to decompose a p-value into two components, one for inference and one for selection, that are independent under the null hypothesis. To control the Type 1 error rate conditional on selection of the null hypothesis from randomized p-values, we develop tests motivated by the selective inference approaches of data thinning, fission, and carving. Unlike the typical application of these approaches, however, our tests only make use of the p-values rather than the full data. Furthermore, we construct confidence intervals that attain the nominal conditional coverage and corresponding estimators. In the context of the file drawer problem, the winner's curse, and sequential testing, we show that our proposals offer both theoretical and empirical advantages over existing randomized and non-randomized approaches. Lastly, we apply our methods for selective inference on genetic associations using publicly-available p-values from the UK Biobank.
Related
- SLICE: Sequential Likelihood-Free Inference via Conformal E-Values
- A Tutorial on Safe Anytime-Valid Inference: Practical Maximally Flexible Sampling Designs for Experiments Based on e-Values
- p ‐Values Are Not Evidence: Navigating the Gap Across Modes of Inference
- Automated, Efficient, and Model-Free Inference for Randomized Clinical Trials via Data-Driven Covariate Adjustment
- Collaborative representations for targeted causal inference under outcome selection
BibTeX
@misc{perry2026inference,
title = {{Inference after selection on randomized p-values}},
author = {Ronan Perry and Daniela Witten},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.07459},
eprint = {2611.07459},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.07459},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Inference after selection on randomized p-values
AU - Ronan Perry
AU - Daniela Witten
PY - 2026
DA - 2026/11/05
PB - arXiv
DO - 10.48550/arxiv.2611.07459
UR - https://arxiv.org/abs/2611.07459
AB - We consider inference on parameters and hypotheses selected using p-values. We provide a simple procedure to decompose a p-value into two components, one for inference and one for selection, that are independent under the null hypothesis. To control the Type 1 error rate conditional on selection of the null hypothesis from randomized p-values, we develop tests motivated by the selective inference approaches of data thinning, fission, and carving. Unlike the typical application of these approaches, however, our tests only make use of the p-values rather than the full data. Furthermore, we construct confidence intervals that attain the nominal conditional coverage and corresponding estimators. In the context of the file drawer problem, the winner's curse, and sequential testing, we show that our proposals offer both theoretical and empirical advantages over existing randomized and non-randomized approaches. Lastly, we apply our methods for selective inference on genetic associations using publicly-available p-values from the UK Biobank.
ER -