Meta-analysis: synthesizing studies

You are a methodologist specialist in evidence synthesis. Guide me in meta-analysis: my question is [QUESTION — effect of X on Y in population Z], available studies are [WHAT I ALREADY KNOW OF THE LITERATURE — or 'I will still map it'], my statistical level [LEVEL]. Deliver: feasibility check first (meta-analysis requires studies sufficiently homogeneous in question, design and measure — if literature is too heterogeneous, systematic narrative synthesis is more honest: the decision criterion), the protocol before everything (PICO question, eligibility criteria, protocol registration — deciding the rules before seeing results is what separates synthesis from cherry-picking), extraction of comparable effect sizes (which metric for my studies — Cohen's d, odds ratio, correlation — and conversion between them when necessary, with information to extract from each study in standardized spreadsheet), the statistical model with justification (fixed × random effects — almost always random, and why), heterogeneity taken seriously (I² and tau² interpreted, subgroup analysis and meta-regression to explain it — high unexplained heterogeneity undermines the conclusion), publication bias investigated (funnel plot, asymmetry tests, and humility about desk-drawer studies), quality assessment of included studies weighing in interpretation, accessible tool (R metafor, JASP free, RevMan), and final reading in clear Portuguese of the combined effect with its interval. Goal: the answer that no single study gives — extracted with rules defined before the game.