startthat.work home

HE1-11 Meta-analysis

Combines effect estimates from separate studies into a weighted average with a confidence interval and produces a forest plot, a pooled effect and heterogeneity statistics.

Handle evidenceEvidence: DocumentedPrimary technique, runs in step 5.3Strongest statement: estimate the effect of an action

What it does

Combines effect estimates from separate studies into a weighted average with a confidence interval. Artefact: a forest plot, pooled effect and heterogeneity statistics; read heterogeneity before the pooled number.

When to use it

Use when several similar studies report comparable effect estimates and you need one pooled estimate with its uncertainty.

When it does not fit

Studies differ too much, or are few and small. Funnel plot guidance was not read.

Result format

The result has a forest plot with one row per study, a pooled effect with its confidence interval, and heterogeneity statistics. The heterogeneity is stated before the pooled number.

Requirements on the material

PropertyRequirement typeRequirement
Content typeAlways appliesrequires=tabular; requires=text
Unit of analysisAlways appliesrequires=document; requires=dataset
Variable typesRaises a warningprefers=interval; prefers=ratio; prefers=counts or fractions
Level of detail availableAlways appliesrequires=aggregate tables
Number of sourcesAlways appliesrequires=two or more
Appraises source credibilityApplies in some casesno

Not restricted on: Structure, Time structure, How the sample was chosen, Whether sources can be joined, Amount of material, Known data problems, Producer interest, Source dependence, Origin, Document genre, Language, Personal information, Security classification, Licence.

Minimum materialat least two studies with effect estimates and standard errors, or the counts or means needed to compute them
Additional sources needednone
Not for (rules)none
Not for (text)Studies differ too much in design or population, or are few and small.

Choosing and ordering

Kind of techniquePrimary technique, runs in step 5.3
Phase taganalyse
Other phase tagsconclude
Kind of findingpattern; difference
Question answeredhow much
Strongest statement it supportsestimate the effect of an action
Needs as input
Makes as output
Must exist firstNone
Loop targetreturn to gather

Library checks by stakes

Each stakes level includes the checks of the level below it. Checks that need a person apply at high stakes only, so the process leaves them out at lower stakes (decision 38). Repeatability: deterministic.

StakesCheckPerson involvedCounts as a different method
exploratoryinput validationnoneno
decision-informinginput validationnoneno
decision-informingspot checkoptionalno
decision-informingsensitivity analysisnoneno
decision-informingcode execution for arithmeticnoneno
high-stakesinput validationnoneno
high-stakesspot checkoptionalno
high-stakessensitivity analysisnoneno
high-stakescode execution for arithmeticnoneno
high-stakesindependent reviewneededno
high-stakesprovenance lognoneno

Failure modes and limits

Model suitabilitymodel plus code
Typical model failureComputes the pooled estimate by hand with arithmetic errors, or reports the pooled number without reading heterogeneity.
Error sourcesinput data; statistical assumptions; source selection
Certainty limitsCertainty drops with heterogeneity (inconsistency), small studies and publication bias (reporting bias), and biased included studies (risk of bias).

Evidence and tag basis

Documented. The Handbook warns that results can mislead if bias, variation and reporting bias are ignored, that the heterogeneity test has low power with few studies, and that common methods are biased for rare events. (Cochrane Handbook, chapter 10)

Tag basis: input profile inference; selection inference; sequencing inference; robustness inference. All tag values are inference unless a source is named.