RMP.7:11 - SoTA-Echoing

What counts as independent support? The selected line recovers the underlying studies and source relationships while retaining additional information from their reports. The cheap default is to count publications or discard all apparent duplicates. The former inflates support; the latter can lose follow-up detail. Adapt: 4.2 combines information at the appropriate contribution and preserves consequential overlap, as illustrated by A/A2 in 5.1. Cochrane chapter 5 supplies the best-known-line candidate for these report/study distinctions because it resolves both failures, not because it is an institutional handbook. This recovery adds work when relationships are unclear; it is needed only where they can change the finding. New overlap, provenance or supplementary results reopen the affected contribution.

Which numerical answer can this corpus support? The selected line chooses a synthesis for a stated quantity and dependence model, with explicit alternatives when that quantity is unsupported. The serious alternatives are pooling every numerical result or refusing all heterogeneous material. The first invents comparability; the second can discard a warranted combined estimate. Adapt: 4.4 and 5.1–5.2 preserve justified combination and an informative narrower answer. Cochrane chapter 10 supplies the quantitative best-known-line candidate; chapter 12 supplies the competing non-meta-analysis route when its conditions fit. Their substantive question and interpretation rules justify the choice; intervention-review conventions do not govern every research family. The trade-off is model and grouping judgment in exchange for an interpretable answer. A changed target, dependence relation or defensible alternative model reopens that choice.

How should missing contributions affect a finding? The selected line locates what is missing and the particular inference it can change. Treating an available corpus as complete is cheaper but can hide selective results; rejecting every incomplete corpus would also erase bounded useful answers. Adapt: 4.1 and 4.3 retain question-relative coverage and distinguish unavailable, unreported, unsought and lost material. Cochrane chapter 13 is a best-known-line candidate for these missing-evidence questions; its meta-analysis tools retain that scope. New eligible material, an identified selective omission or a stronger model of its effect reopens the affected finding.

How can situated and mixed findings be related? The selected line preserves context and finding-specific confidence, integrating transformed contributions only when they answer a shared question. The serious rival is separate synthesis of all quantitative and qualitative contributions; a different default would merge everything into one theme. Cochrane chapter 21 supplies a best-known qualitative line and the separate-contribution choice; JBI §8.3 supplies the serious question-relative integration alternative. Adapt: 4.4 and 5.3 select between them by the question. This permits a useful shared answer while keeping different aspects separate and transformation losses recoverable. GRADE-CERQual supplies an applicable finding-level confidence candidate, not a numerical truth scale. The accepted cost is explicit interpretation and method choice. A changed tradition, question or stronger account of what transformation loses reopens the comparison.

What can historical agreement establish? The selected line compares source dependence, dating, purpose, contrary material and rival explanations. Narrative count is an attractive cheap default, but copied accounts cannot supply independent causal support. Adapt: 4.2–4.4 and 5.4 recover the shared memoir and the different contribution of the decision record. AHA professional standards supplies the best-known-line candidate for source integrity and evidence/interpretation discipline here; it supplies no verdict about the particular factory. The extra source criticism preserves the useful chronology while leaving the unsupported cause open. New provenance, dating or a substantive rival explanation reopens that inference.

These comparisons support different research traditions. Their combination here is question-relative guidance, not evidence that one synthesis procedure works universally.