Data check Data
The errors that don't crash R — they just ruin the meta-analysis
R runs, the forest comes out, and the result is wrong because an SE column went in as SD.
You give
The extraction spreadsheet (.xlsx or .csv), in your own column layout.
You get
A blocking report (.docx, which opens in Word, and .md) and your spreadsheet back with two new columns: flags (what to check on each row) and severity (blocks or only warns), plus the Report sheet (the report) and the Mapping sheet (which of your columns became which field).
How it works
# Data check $ scopetools run sanity Deterministic — no model. It maps your columns, runs per-row checks (impossible values, missing arm data, precision) and pool checks (unit and scale consistency across studies, duplicates, outliers) and grades each finding as blocking or warning.
Deterministic; it maps your columns (free layout), checks per row and per pool, grades blocking vs warning — and returns your own spreadsheet with flags/severity columns and Report/Mapping sheets added. No redoing your data in someone else's format.
What it costs
3 credits per spreadsheet (US$0.15), whatever the number of rows.
You see the price before it runs — a failed run gives the credits back.
Credits are shared across every MetaScope tool — buy once, spend anywhere.
Limits
It checks the numbers and how they fit together, not whether a study belongs in the review: that stays with the reviewers.
Frequently asked questions
- Do I need a specific template?
- No; it maps your own column names (n_e/mean_e/sd_e and variants).
- What is “blocking” vs “warning”?
- Blocking means do not pool that row yet.
- When should I run it?
- Before the forest plot, not after.
Rows clean? Run the forest plot in PlotScope — no R required.
