← Every exhibit

Table Figure 1 From the research bench

The synthetic statistical-inference review

Significance is one field in an inference record. Effect, uncertainty, testing family, model, and permitted wording must remain visible together.

IDReported resultUncertainty and informationTesting or model contextConclusion permittedStill not established
T-01Estimate +2.0 points, p = 0.0195% interval +0.5 to +3.5; adequate planned powerOne declared outcome and modelEvidence of a positive association in the studied samplePractical importance or causality
T-02Estimate +0.2 points, p < 0.001Narrow interval +0.15 to +0.25Large sample; outcome scale is smallPrecisely estimated small differenceMaterial business value
T-03Estimate +8 points, p = 0.08Wide interval -1 to +17One outcome; limited informationData are insufficient to rule in or out the declared effectNo-effect conclusion
T-04Estimate +5 points, p = 0.03Interval +0.4 to +9.6Twelve outcomes tested; no family rule statedAt most a flagged result pending multiplicity reviewConfirmed discovery
T-05Estimate +10 points, p = 0.02Interval +2 to +18Model changed after inspecting outcomeConditional post hoc resultEx ante test interpretation
T-06Estimate +6 points, p = 0.01Interval +2 to +10Studied in one segment and 14-day windowPositive result in that segment and windowTransfer to all customers or 90-day renewal

Swipe or scroll horizontally if the table is wider than your screen.

Cite Embed

Reference & Evidence

Source: Author's synthetic inference review grounded in Cohen (1994) and Ioannidis (2005). Estimates, intervals, testing histories, and permitted conclusions are illustrative.

Each line is a claim from the register this journal publishes against, resolved from the register at build time.

  • A "the ritual of null hypothesis significance testing mechanical dichotomous decisions around a sacred 05 criterion still persists", and the article reviews the problems with it Cohen (1994), article · COH94-C1
  • A What he recommends instead: "an emphasis on estimating effect sizes using confidence intervals and the informed use of available statistical methods", with graphic methods and better measurement Cohen (1994), article · COH94-C2
  • A "the probability that a research claim is true may depend on study power and bias the number of other studies on the same question" and the ratio of true to no relationships in the field Ioannidis (2005), PLoS Medicine article · IOANNIDIS05P-C1
  • A Bias is the lever: "with increasing bias the chances that a research finding is true diminish considerably this is shown for different levels of power and for different pre study odds" Ioannidis (2005), PLoS Medicine article · IOANNIDIS05P-C2
  • B Statistical conclusion validity asks whether the inference from data and model to the stated statistical conclusion is warranted under its assumptions and error controls Author framework grounded in COH94-C1 to COH94-C3 · R08-OWN-C1
  • B A review should report effect estimate, uncertainty interval, sample or information basis, testing family, model specification, and conclusion wording together Author framework · R08-OWN-C2
  • B The six-row inference table contains illustrative estimates and labels only; it is not an observed study result or benchmark Author synthetic object · R08-OWN-C7
  • B A significant result should be bounded by the population, outcome, model, testing family, and observation window that produced it Author operating rule · R08-OWN-C9

Grades: A, verified against the printed page of the primary source · B, primary source, text layer only · C, authoritative secondary · D, reported.

Related exhibits