RANKING METHODOLOGY

How ResearchRank calculates and displays rankings

Every rank is tied to a defined population, ranking basis and reproducible data source. Rank-based certificates are issued only when that context is available.

1. Data source and entity matching

ResearchRank processes scholarly metadata through its research-data service. Author, institution and publisher identifiers are used to reduce name ambiguity. Profile metrics include indexed works, citations, h-index and i10-index when supplied by the source.

2. Exact metric rank

For supported entity contexts, rank is calculated as one plus the number of entities in the same population with a strictly higher cited-by count. Researchers with equal cited-by counts therefore share the same competition rank. Country and institution filters are applied before the count.

3. Subject / topic rankings

For researchers, the strongest ResearchRank topics attached to the researcher profile are evaluated as independent citation-impact populations. Other subject hierarchy pages may use matched-work grouping where a direct entity filter is not available. The certificate always records the actual ranking basis used.

4. Percentile calculation

Percentile position = rank position ÷ evaluated population × 100. ResearchRank Top 1%, Top 2%, Top 3%, Top 5% and Top 10% labels are assigned only when the calculated position falls inside the relevant threshold.

5. Composite analytics score

Citation impact35%

h-index25%

Research output20%

i10-index10%

2-year mean citedness10%

The composite score is an analytical aid normalized within a candidate set. It is not silently substituted for the stated ranking basis.

6. Multiple ranking scopes

A researcher may simultaneously hold global, country, institution and subject/topic ranks. State/region and gender ranks are separate ResearchRank verified cohorts and are never inferred from names or other proxies.

7. Calculation audit trail

Competition rankrank = 1 + count(metric > researcher metric)

Tied metric values share the same rank. No random tie-breaking is used.

Percentile positionpercentile = rank / population × 100

The smallest qualifying threshold is used: ≤1% Top 1%, ≤2% Top 2%, ≤3% Top 3%, ≤5% Top 5%, ≤10% Top 10%.

Populationpopulation = ResearchRank meta.count after filters

The same country, institution or topic filter is applied to both the target rank and population count.

Profile certificatesrules = works / citations / h-index thresholds

These do not claim a percentile rank unless a valid rank population is available.

8. Duplicate-profile controls

Search results are de-duplicated by source identifier. For researcher profiles, ResearchRank additionally checks probable duplicates using ORCID, normalized name, shared last-known institution and similar output counts. Possible duplicates are shown for review but are not silently merged, because merging distinct researchers would corrupt ranking metrics.

9. Responsible interpretation

Open bibliographic data can contain coverage, affiliation and author-disambiguation limitations. ResearchRank therefore displays the ranking basis and verification data with each certificate. A calculation can be mathematically exact for the returned source population while the underlying scholarly metadata may still contain source-level errors.