Qualification: Plea for routinely presenting prediction intervals in meta-analysis

Record: kaal:position:2026-08-08-332 · 2026-08-08

IntHout, Ioannidis, Rovers, and Goeman qualify the inference from a pooled confidence interval. Under between-study heterogeneity, a pooled 95 percent confidence interval may remain wholly on one side of zero while a prediction interval reaches both sides. Their review found that 72.4 percent of statistically significant Cochrane meta-analyses with observed heterogeneity had prediction intervals that included the null. This result does not dispute the registered calculation or the negative direction in each campaign unit. It shows that pooled significance and directional consistency do not alone establish how the effect will generalize to another exchangeable unit. That stronger claim requires a heterogeneity-aware prediction analysis. The paper does not test Kaal's cohort, hypotheses, estimator, or campaign units.

Affirmed commentary position. This record extends a source-bound scholarly claim but is not a verbatim paper claim.

Holds when

The response is limited to the exact BMJ Open passages and the one mapped Kaal claim.
External evidence level: complete public BMJ Open article in Europe PMC JATS XML with a concordant Crossref record and exact proposition-bearing passages.
Mapping review tier: independent substantive scholarly-growth qualification.
The paper studies clinical meta-analyses and does not evaluate Kaal's multi-model cohort, registered hypotheses, estimator, or campaign units.
The source does not recalculate Kaal's pooled intervals or estimate between-campaign heterogeneity.
Every observed campaign contrast being negative is stronger descriptive evidence than a pooled confidence interval alone, but it is not a prediction interval for another exchangeable campaign unit.
The qualification addresses generalization only. It does not dispute the registered calculation or its reported directional consistency.
Crossref Spanish and Semantic Scholar search returned HTTP 429. Several inherited identity refreshes were rate-limited or unresolved. No blocked response was promoted.

Current debate

Plea for routinely presenting prediction intervals in meta-analysis

Scholarly basis

kaal:claim:7261018-036
Wulf A. Kaal, Empirical Evaluation of the Agentic Reputation Substrate: Deliberation, the Composition of Error, and the Registered Measurement of Agency Costs in a Controlled Multi-Model Cohort (2026). SSRN: https://ssrn.com/abstract=7261018
Source PDF sha256: 1d6cbe544bd0055133f7cf8ff308be4fde955867bd8dc764992b8d516af15fa8

Evidence and mapping

Evidence: complete public BMJ Open article in Europe PMC JATS XML with a concordant Crossref record and exact proposition-bearing passages
Review tier: independent substantive scholarly-growth qualification
Mapping confidence: 0.98
Mapping ambiguous: false

Topics

research-methods consensus-and-security ai-and-agents scholarly-growth-coverage scholarly-literature statistical-inference confidence-intervals meta-analysis heterogeneity prediction-intervals evidence-provenance

Provenance

Affirmed in kaal-review:2026-08-12:scholarly-growth-7261018-036-reviewed-v1 on 2026-08-08. Review record.

Verify

Canonical markdown sha256: 27e0dd506806c5d2f46cdba6d0336126ea8fb7c6557f63d116188511a2cfd66f
curl -s https://wulfkaal.github.io/positions/2026-08-08-332.md | sha256sum