Skip to contents

Audits the geographic distribution of a corpus, tabulating the representation of countries, regions, or World Bank income tiers. The analysis can be weighted by work count, total citations, first-authorship, or corresponding authorship. Each work counts once per distinct known group, with optional fractional credit.

The result includes a Gini coefficient for concentration and a coverage metric (proportion of works with known geography).

Usage

sm_audit_geographic(
  corpus,
  by = c("country", "region", "income_tier"),
  weight = c("count", "citations", "first-author", "corresponding"),
  method = c("full", "fractional"),
  call = rlang::caller_env()
)

# S3 method for class 'sm_audit_geographic'
print(x, ...)

Arguments

corpus

An sm_corpus object.

by

Character. Grouping variable: "country" (ISO 3166-1 alpha-2 codes from authorships), "region" (from institutions table), or "income_tier" (World Bank income classification from institutions).

weight

Character. How to weight each work: "count" (one per work and group, before applying method), "citations" (weighted by cited_by_count), "first-author" (only first-author affiliations), "corresponding" (only corresponding-author affiliations).

method

Counting rule: "full" gives each work one unit per distinct known group; "fractional" splits one unit equally among its known groups. Citation credit is split by the same shares.

call

Caller environment for error reporting.

x

An audit object to print.

...

Ignored.

Value

An sm_audit_geographic S3 object containing:

distribution

Tibble with columns group, count, citations, pct, credit and weighted_citations. Counts and raw citations include each work once per group under both methods. Percentages and the Gini use credit, or weighted citations for citation weighting. Count percentages describe work-group pairs under full counting and works under fractional counting.

gini

Gini coefficient of the distribution.

coverage

Proportion of works with at least one known geography.

by

The grouping variable used.

weight

The weighting method used.

method

The counting rule used.

x invisibly (print methods).

Details

Coverage uses all corpus works as its denominator and counts works with at least one known group after the authorship filter. Missing geography is excluded from credit allocation; incomplete metadata can therefore change both coverage and the shares assigned to known groups.

Examples

corpus <- sm_example_corpus()
geo <- sm_audit_geographic(corpus)
print(geo)
#> 
#> ── <sm_audit_geographic> ───────────────────────────────────────────────────────
#> Grouping: country
#> Weighting: count
#> Counting: full
#> Coverage: 100% of works have known geography
#> Gini coefficient: 0.126
#> 
#> 
#> ── Distribution (top 10) 
#> DE: 63 (9.17%) [954 cit.]
#> FI: 46 (6.7%) [708 cit.]
#> AU: 44 (6.4%) [646 cit.]
#> NO: 38 (5.53%) [472 cit.]
#> GB: 37 (5.39%) [544 cit.]
#> IT: 37 (5.39%) [540 cit.]
#> DK: 36 (5.24%) [500 cit.]
#> BR: 33 (4.8%) [533 cit.]
#> CA: 33 (4.8%) [568 cit.]
#> CH: 33 (4.8%) [474 cit.]
#> ... and 10 more
#> 
#> 
#> ── Limitations 
#> • Country codes are derived from author affiliation metadata, which may be
#> incomplete or inaccurate.
#> • Full counting gives a work one unit in each known group; fractional counting
#> splits one unit among those groups.
#> • Region and income tier classifications rely on institutional metadata which
#> may not be populated for all works.
#> • Geographic representation does not capture diaspora researchers or
#> researchers with affiliations in multiple countries.
#> • The Gini coefficient measures concentration but does not account for
#> population size or research funding differences.