Metrics
Primary Source Coverage: Calculation and Limitations
Primary source coverage is the share of source entries pointing to original documents or datasets, counted monthly.
- Author
- Xingge
- Published
- 2026-11-09
- Updated
- 2026-11-09
- Level
- L2 Reproducible
- Version
- 1.0
Term details
- Name
- Primary Source Coverage: Calculation and Limitations
- Type
- Metrics
- ID
- primary-source-coverage-en
Primary Source Coverage: Calculation and Limitations
Primary source coverage measures the share of source entries on a page that point to original documents, policy texts, or public datasets. It is calculated as primary source entries divided by sourced entries and counted monthly.
Definition and calculation standard
The denominator is the number of entries on the page that carry a source label, and the numerator is the number of entries whose source points to an original text. Entries pointing to secondhand paraphrases, aggregator articles, or on-site cross-references are excluded from the numerator.
| Item | Value |
|---|---|
| Metric name | Primary source coverage |
| Formula | Primary source entries divided by sourced entries |
| Unit | Percentage |
| Recording frequency | Monthly |
| Judgment basis | Source type list (original, dataset, paraphrase, on-site cross-reference) |
| Required fields | Source name, publisher, link, type judgment |
Collection and judgment
- List all source entries. Page by page, register every source in the source block separately.
- Judge the source type. Classify into four types: original text, public dataset, secondhand paraphrase, and on-site cross-reference.
- Calculate the ratio. Only the first two types enter the numerator; the rest count in the denominator.
- Record replacement plans. For secondhand paraphrases, mark whether the original can be traced and add a to-do item when it can.
- Aggregate monthly. Keep page-level detail so specific entries can be located.
Data sources and record fields
Data comes from the source register, with fields including page address, source name, publisher, date, link, type judgment, whether the original can be traced, and last verification date.
Source type judgment must stay consistent with the evidence level standard, so the same material is not judged as two types.
Limitations and misreadings
- Primary sources can themselves decay, so coverage may be inflated or understated over time; read it together with the broken evidence rate.
- Some topics lack public primary material, so coverage has a structural ceiling.
- Judgment relies on human identification; paraphrases dressed up as originals need manual tracing.
- Coverage does not measure source quality, only the structural level.
Relationship to other metrics
Coverage, the broken evidence rate, and evidence freshness together describe the evidence structure: whether primary sources are reachable, whether they still exist, and whether they are expired.
Related sections: metric index, method index, glossary index, research index, checklist index.