Cluster RCT in your meta-analysis, no ICC reported? You borrow one. Bhat & Beretvas (Res Synth Methods) simulated pooling ICCs: REML vs RVE barely mattered; the variance formula did. Only the Fisher-transformation weight never gave a substantially biased pooled ICC.

When two RoB 2 raters disagree, often it isn't judgment. They read different sentences: one stopped at "analysed by ITT," the other found the dropout table. In GLASS, every RoB call sits next to the quote it rests on. You argue about text, not memory. A human still decides.

RoB 2 papers usually tell us raters disagree. D'Souza et al. (Cochrane Evid Synth Methods) show what one team did about it: 49 results from 34 RCTs of remote alcohol/drug therapies, piloted together, assessed in pairs, every assumption written into a review-specific rulebook.

An NMA has no single certainty rating. Each comparison gets its own: rate direct and indirect evidence, take the higher if they agree, downgrade for incoherence if they don't, then judge imprecision on the network estimate. CoreSR does NMA GRADE natively, beside the R model.

After stroke/TIA, pooling 432 estimates says 74.1% of patients stay adherent. Standardize to one definition (pharmacy data, moderate cutoff, 9–15 mo) and it's 53.7%. Persistence: 76.8% → 48.2%. Jiang et al. (JCE): how adherence was measured drove more heterogeneity than who.

With 30 included studies, recall moves in 3.3-point steps: miss one and you're at 96.7%. So a screener's "99% recall" only means something next to the number of includes behind it. Small validation sets make recall look perfect or awful. Ask for the denominator.

Living reviews promise to stay current. Do they say how? Of 119 living evidence syntheses (JCE), 52% gave a monitoring plan and 5% stated stopping criteria. In 20 tracked across updates, PRISMA-LSR compliance fell from 57.5% to 42.3%. Fatigue, or was "living" never defined?

RoB 2, novices vs experts: 10 newly trained and 6 experienced raters assessed 94 results from 37 RCTs (Xiang et al., BMC Med Res Methodol). Novices agreed more, yet rated "low risk" more often in most domains. Agreement isn't accuracy. Shared training, or shared defaults?

Time-lag bias isn't just "positive trials publish first." Xing et al. (Res Synth Methods) add inverse, cyclical and localized patterns. In 19 falls trials the pooled estimate swung; a linear year slope was flat (P=0.351). Early trials skew small. Time lag or small-study effect?

Exercise + education + CBT vs usual care. Standard NMA gives each bundle its own node, so it can't say which part does the work. Component NMA estimates each component's effect under an additive model you can test against full NMA. CoreSR runs it in real R.