Practical guides · Page 16

Calculate a Risk Difference with a Newcombe Interval

Express an independent-group probability difference in percentage points and document its interval construction.

Check the binary denominator and interval method before reporting uncertainty around an observed proportion.

Distinguish reported absence of events from missing reporting and retain sparse-study data for an appropriate synthesis decision.

Capture the data, code, plan, environment, and outstanding decisions needed to reproduce a defined analysis release.

Reconstruct one descriptive summary from non-overlapping groups without confusing combined variability with a pooled within-group SD.

Check complete-pair counts and assumptions before translating a correlation into an approximate confidence interval.

Look for curvature, influential points, subgroup structure, and repeated observations before interpreting r.

Check effect scales, standard errors, independence, and weighting before interpreting a pooled estimate.

Review extraction, clinical differences, and uncertainty before turning an I-squared statistic into a conclusion.

Explain why software can produce different quartiles and record a reproducible summary convention.

Convert record counts, reviewer effort, and pilot timings into a transparent workload estimate.

Calculate median-based spread while documenting scaling, ties, and limits of interpretation.