A/B testing peeking: always-valid p-values vs fixed-horizon false positives
Takeaway Detail Fixed-horizon tests analyzed once at the end hold false positives to 5%. The 5% false-positive rate applies only when the test is analyzed a sin Independent technical note mirroring the canonical version: https://bteanalytics.co/blog/ab-testing-peeking-always-valid-p-values-vs-fixed-horizon-false-positives.php
Authors
- Sofia Lindström
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-21
- DOI
- https://doi.org/10.5281/zenodo.24416094
- Primary Topic
- Statistical Methods in Clinical Trials
- Type
- preprint