Response-rate ratio¶
The ratio of the response proportion in a treatment or exposed group to the response proportion in a control or reference group, used as a relative efficacy measure for a prespecified binary outcome over a declared follow-up period.
Core Idea¶
The response-rate ratio is the response proportion in a treatment or exposed group divided by the response proportion in a control or reference group for the same prespecified binary outcome and follow-up period. It is a response-event risk ratio, not an odds ratio. It is a risk ratio for the event called response, not an odds ratio. It is a risk ratio for the event called response, not an odds ratio.
Scope of Application¶
The ratio is used in randomized trials, comparative effectiveness, meta-analysis, epidemiology, product experiments, marketing campaigns, quality improvement, and binary-outcome reporting. Use it with response definition/threshold/timepoint, treatment/reference interventions, response counts and full denominators, estimand and ITT/per-protocol population, missing and intercurrent events, clustering/strata/adjustment, formula direction, interval and zero-cell method, absolute difference and baseline rate, harms, multiplicity/subgroups, and practical threshold. Do not substitute odds, hazards, mean changes, raw-count ratios, or relative efficacy without uncertainty and bias analysis.
- Clinical trials. Compares responder proportions.
- Meta-analysis. Pools compatible log ratios.
- Marketing. Compares defined conversion/response events.
- Subgroups. Requires interaction-aware inference.
- Decision support. Pairs relative and absolute effects.
Clarity¶
Report response definition/threshold and validation, timepoint/window, treatment and reference interventions, randomized/exposed denominators, response counts and proportions, estimand and ITT/per-protocol population, missing/intercurrent-event handling, cluster/strata/adjustment model, ratio formula and direction, confidence/credible interval and p-value if used, zero-cell method, absolute difference and baseline risk, harms, multiplicity, subgroup interaction, and clinical/practical threshold. The closest near miss sets the boundary: Risk ratio is the nearest statistical synonym when 'response' is the event; response-rate difference is the nearest alternative effect scale.
Manages Complexity¶
The measure compresses four counts into a scale-free comparison, enabling cross-risk interpretation while hiding absolute benefit, missingness, and the response threshold. The central relative comparability–absolute impact tradeoff is this: Ratios scale across baselines while decisions depend on absolute events. A second simple responder rule–threshold sensitivity tension matters because Binary response communicates clearly while discarding magnitude near the cutoff.
Abstract Reasoning¶
Use three linked moves: freeze a common response event, timepoint, and estimand; construct comparable treatment and reference denominators; compute both proportions, the ratio, and absolute difference. As a collapse test, the result becomes undefined or unstable with a zero reference rate unless an explicitly justified model/correction is used, and uninterpretable when outcomes or denominators differ. A fourth check is to estimate uncertainty with dependence, zero cells, and missingness handled.
Knowledge Transfer¶
The ratio transfers from medicine to marketing or quality settings only if response, denominator, exposure assignment, time window, and causal assumptions are redefined. The reviewed DAG parent relation is ; it carries the broader structural comparison without erasing the specialist conditions. Formal operation, not sufficient identity on its own. Statistical synonym when response is the event.
Relationships to Other Abstractions¶
Current abstraction Response-rate ratio Domain-specific
Parents (1) — more general patterns this builds on
-
Response-rate ratio is a kind of Ratio Prime
Response-rate ratio is a strict kind of Ratio: Response-rate ratio is a measure of efficacy of therapy in clinical trials.
Hierarchy path (1) — routes to 1 parentless root
- Response-rate ratio → Ratio → Comparison → Self Checking
Neighborhood in Abstraction Space¶
Response-rate ratio sits in a moderately populated region (57th percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.
Family — Empirical Measurement & Statistical Inference Methods (50 abstractions)
Nearest neighbors
- Clinical-Trial Stratification — 0.87
- Assay sensitivity — 0.87
- Obesity paradox — 0.85
- False coverage rate — 0.84
- Length time bias — 0.84
Computed from structural-signature embeddings · 2026-10-08