Leiden Ranking

Credibility highlights

Leiden Ranking limits itself to research performance based on publications and citations, instead of mixing teaching, reputation surveys, income, and other dimensions into a single composite score. CWTS explicitly argues against a generic “best university” concept and against arbitrary composite indicators.
You can choose indicators (impact, collaboration, open access, gender) and choose how to sort results, which makes the ranking easier to audit than “one number” systems.

Field and year normalization: Leiden’s impact indicators include field normalized citation measures (TNCS, MNCS) and top x percent cited measures, which reduces bias from different citation cultures across disciplines.
Fractional counting option for impact: fractional counting reduces inflation from large, highly collaborative fields and supports fairer cross-field comparisons, CWTS states it as the preferred method for impact indicators.
Stability intervals: Leiden provides uncertainty ranges (95% bootstrap stability intervals) so you can see when small rank differences are noise. Many rankings do not give this.

Data source: Web of Science (Clarivate) indices used include Science Citation Index Expanded, Social Sciences Citation Index, and Arts and Humanities Citation Index.
CWTS describes how it enriches data, including citation matching, geocoding, open access labeling, author disambiguation, and careful institutional name unification. This transparency supports credibility, because errors often come from affiliation matching.

Main credibility limits you should account for

Leiden excludes books and conference proceedings, which can materially underrepresent output and impact in computer science, engineering, and parts of social sciences and humanities. CWTS calls this an important limitation.
Leiden’s “core publications” restrict to English language publications and to “core journals” deemed suitable for citation analysis. This improves comparability, but can penalize universities with strong local language publishing or practice oriented outlets.

CWTS explains that university delimitation varies by country and that it often follows local practice, splitting or combining systems differently (for example, collegiate and system universities, and recent reorganizations in France). Those boundary decisions can shift publication counts and proportions.

Size dependent indicators (counts) favor large institutions, size independent indicators (proportions) can favor smaller, specialized institutions. Leiden documents this distinction, but users still misread “rank” as “quality overall” if they do not choose carefully.
Even with normalization, citation based indicators reflect visibility and citation practices, not teaching quality, societal impact, or contribution to local economies. CWTS explicitly states Leiden does not cover teaching and other dimensions.

Gender indicators rely on name based gender inference and cannot classify a meaningful share of authorships, CWTS reports around 70% coverage. Use for broad patterns, avoid fine grained comparisons.

How to use Leiden Ranking in a credible way for evaluation or strategy

Prefer size independent indicators when you compare differently sized universities, for example PP(top 10%) or MNCS.
Turn on fractional counting for impact when you compare across fields with different collaboration intensity.
Treat small rank gaps as uncertain. Check stability intervals before you interpret changes year to year.
Run the comparison within the relevant field (for example, Social sciences and humanities) rather than “All sciences” if your institution has a distinctive profile
Use Leiden as one component in a dashboard:
- Research: Leiden indicators (impact, collaboration, OA)
- Teaching: separate validated teaching metrics
- Societal impact: patents, policy citations, clinical guidelines, regional engagement metrics
  This aligns with CWTS’s responsible use principle that “best university” claims do not make sense across dimensions.

Bottom line

Leiden Ranking earns credibility for research comparison because CWTS documents its data source, applies field normalization, supports fractional counting, and exposes uncertainty via stability intervals.
Your confidence should drop when you evaluate fields where books and proceedings dominate, when local language publishing matters, or when institutional boundary choices create non-comparable units.

18 Jan 2026