Comparative Probability Metrics: Using Posterior Probabilities to Account for Practical Equivalence in A/B tests
Bibliographic record
Abstract
Nathaniel T. Stevens* & Luke Hagar Department of Statistics & Actuarial Science University of Waterloo, Waterloo, ON, CanadaSupplementary materials for this article are available online. Please go to www.tandfonline.com/r/TAS.CONTACT Nathaniel T. Stevens nstevens@uwaterloo.ca Department of Statistics & Actuarial Science, University of Waterloo, Waterloo, ON, Canada N2L 3G1.ABSTRACTRecently, online-controlled experiments (i.e., A/B tests) have become an extremely valuable tool used by internet and technology companies for purposes of advertising, product development, product improvement, customer acquisition, and customer retention to name a few. The data-driven decisions that result from these experiments have traditionally been informed by null hypothesis significance tests and analyses based on p-values. However, recently attention has been drawn to the shortcomings of hypothesis testing, and an emphasis has been placed on the development of new methodologies that overcome these shortcomings. We propose the use of posterior probabilities to facilitate comparisons that account for practical equivalence and that quantify the likelihood that a result is practically meaningful, as opposed to statistically significant. We call these posterior probabilities comparative probability metrics (CPMs). This Bayesian methodology provides a flexible and intuitive means of making meaningful comparisons by directly calculating, for example, the probability that two groups are practically equivalent, or the probability that one group is practically superior to another. In this article, we describe a unified framework for constructing and estimating such probabilities, and we develop a sample size determination methodology that may be used to determine how much data are required to calculate trustworthy CPMs.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.123 | 0.522 |
| Meta-epidemiology (narrow) | 0.003 | 0.001 |
| Meta-epidemiology (broad) | 0.003 | 0.002 |
| Bibliometrics | 0.011 | 0.008 |
| Science and technology studies | 0.002 | 0.012 |
| Scholarly communication | 0.008 | 0.016 |
| Open science | 0.005 | 0.007 |
| Research integrity | 0.005 | 0.009 |
| Insufficient payload (model declined to judge) | 0.006 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".