--- name: evidence-meta-analysis description: "Plan or interpret meta-analysis with compatible studies, effect measures, pooling models, heterogeneity, bias checks and sensitivity analyses." --- ## Selection and handoffs Plan or interpret statistical synthesis of compatible studies. **Select for:** Plan or interpret meta-analysis with compatible studies, effect measures, pooling models, heterogeneity, bias checks and sensitivity analyses. **Expected work:** Compatibility check; d/g/OR/RR/correlation; Fixed/random models; I-squared; Forest/funnel; Egger; Subgroup/sensitivity; Software options **Expected output:** Output format A plan, an explanation of the user's results, or a written methods and results section. **Scope boundaries:** Do not activate solely because the family was used earlier or for unrelated chat.; Use the adjacent specialist when its distinct task is requested: evidence-systematic-review. Keep: numerical pooling and diagnostics differ from the protocol/search/screen/extract/report lifecycle, which may remain narrative.; Statistical thresholds, model defaults and test matrices are contextual heuristics, not universal rules or executed analyses. **Related skills:** [evidence-systematic-review](../evidence-systematic-review/SKILL.md) Apply this skill only to the selected task. Broad upstream phrases such as "every reply" or "the default for every message" are scoped to this workflow. They do not activate it for unrelated chat. User instructions and the [integration contract](../../docs/INTEGRATION.md) govern combined work. Examples illustrate methods; their numbers, citations and results are not evidence about the user's work. ## Preserved source instructions ## Unified plugin integration Read [the integration contract](../../docs/INTEGRATION.md) when combining workflows. This active copy retains the complete source instructions below with namespaced skill references. Original bytes and upstream metadata remain in `../../resources/`. Resolve tools by their actual host capabilities; tool names in upstream instructions are capability requirements, not proof of connection. Use the task workspace for generated outputs. Treat historical case-study results as supplied examples, not newly verified findings. Family skill aliases (read the linked SKILL.md before use): - `citation-chaining` → `evidence-citation-chaining` (`../evidence-citation-chaining/SKILL.md`) - `citations-and-references` → `evidence-citations-and-references` (`../evidence-citations-and-references/SKILL.md`) - `claim-checker` → `evidence-claim-checker` (`../evidence-claim-checker/SKILL.md`) - `compare-studies` → `evidence-compare-studies` (`../evidence-compare-studies/SKILL.md`) - `data-extraction` → `evidence-data-extraction` (`../evidence-data-extraction/SKILL.md`) - `database-guide` → `evidence-database-guide` (`../evidence-database-guide/SKILL.md`) - `debate-and-argument-evidence` → `evidence-debate-and-argument-evidence` (`../evidence-debate-and-argument-evidence/SKILL.md`) - `decision-evidence` → `evidence-decision-evidence` (`../evidence-decision-evidence/SKILL.md`) - `evidence-answer` → `evidence-evidence-answer` (`../evidence-evidence-answer/SKILL.md`) - `evidence-based-writing` → `evidence-evidence-based-writing` (`../evidence-evidence-based-writing/SKILL.md`) - `evidence-help` → `evidence-evidence-help` (`../evidence-evidence-help/SKILL.md`) - `evidence-meter` → `evidence-evidence-meter` (`../evidence-evidence-meter/SKILL.md`) - `field-guides` → `evidence-field-guides` (`../evidence-field-guides/SKILL.md`) - `find-studies` → `evidence-find-studies` (`../evidence-find-studies/SKILL.md`) - `literature-review` → `evidence-literature-review` (`../evidence-literature-review/SKILL.md`) - `meta-analysis` → `evidence-meta-analysis` (`../evidence-meta-analysis/SKILL.md`) - `open-access-full-text` → `evidence-open-access-full-text` (`../evidence-open-access-full-text/SKILL.md`) - `paper-summary` → `evidence-paper-summary` (`../evidence-paper-summary/SKILL.md`) - `plain-language-research` → `evidence-plain-language-research` (`../evidence-plain-language-research/SKILL.md`) - `research-evidence-finder` → `evidence-research-evidence-finder` (`../evidence-research-evidence-finder/SKILL.md`) - `research-gaps` → `evidence-research-gaps` (`../evidence-research-gaps/SKILL.md`) - `research-proposal-background` → `evidence-research-proposal-background` (`../evidence-research-proposal-background/SKILL.md`) - `research-question` → `evidence-research-question` (`../evidence-research-question/SKILL.md`) - `search-strings` → `evidence-search-strings` (`../evidence-search-strings/SKILL.md`) - `source-verification` → `evidence-source-verification` (`../evidence-source-verification/SKILL.md`) - `statistics-explained` → `evidence-statistics-explained` (`../evidence-statistics-explained/SKILL.md`) - `study-appraisal` → `evidence-study-appraisal` (`../evidence-study-appraisal/SKILL.md`) - `study-designs-and-evidence-levels` → `evidence-study-designs-and-evidence-levels` (`../evidence-study-designs-and-evidence-levels/SKILL.md`) - `systematic-review` → `evidence-systematic-review` (`../evidence-systematic-review/SKILL.md`) --- # Meta-Analysis ## When it is appropriate Studies ask the same question, with similar populations, interventions and outcomes, and report enough statistics to compute effect sizes. ## Core concepts - **Effect sizes:** standardized mean difference (Cohen's d or Hedges' g), odds ratio, risk ratio, correlation. - **Fixed effect:** assumes one true effect. **Random effects:** assumes true effects vary across studies; usually the safer default. - **Heterogeneity:** I squared about 25% low, 50% moderate, 75% high. High heterogeneity means explore why before trusting one pooled number. - **Forest plot:** each study's effect and CI plus the pooled estimate. - **Funnel plot and Egger's test:** check for small-study and publication bias. - **Subgroup and sensitivity analyses:** test whether results change by group or when removing low-quality studies. ## Software R (metafor, meta), Stata, JASP, Comprehensive Meta-Analysis, RevMan. ## Reading a meta-analysis critically Check how studies were found, their quality, the heterogeneity, the pooled effect size (not just significance) and whether bias was assessed. ## Output format A plan, an explanation of the user's results, or a written methods and results section.