BackgroundSystematic reviews and meta-analyses are central to evidence-based nursing, but their conclusions can be distorted by inappropriate effect measures, unit-of-analysis errors, poorly estimated heterogeneity, selective outcome reporting, and overreliance on statistical significance.AimTo provide nursing researchers, educators, and postgraduate students with a practical and current account of the statistical decisions required to plan, conduct, interpret, and report systematic reviews and meta-analyses.ApproachThis methodological review synthesizes authoritative guidance and key statistical literature, including PRISMA 2020 and the Cochrane Handbook for Systematic Reviews of Interventions version 6.5. It is not a systematic review of a clinical intervention and therefore does not report invented search counts or pooled effects.Key contentThe review covers formulation of estimands; selection and transformation of effect measures; common-effect and random-effects models; restricted maximum likelihood estimation; Hartung–Knapp–Sidik–Jonkman confidence intervals; heterogeneity, prediction intervals, subgroup analysis, and meta-regression; cluster, crossover, multi-arm, and repeated-measures designs; sparse events; missing statistics; dependent effects; small-study effects; risk of bias; certainty of evidence; and reproducible reporting.Implications for nursingNursing evidence frequently involves complex behavioural interventions, diverse settings, patient-reported outcomes, cluster allocation, event rates, and multiple follow-up times. Statistical plans should therefore be outcome-specific, design-aware, prespecified, and linked to clinical interpretability.ConclusionA credible nursing meta-analysis is not defined by whether a pooled p value is statistically significant. Credibility depends on a transparent question, compatible effect estimates, correct handling of dependence, justified synthesis models, uncertainty-aware interpretation, and complete reporting.