CWTS Leiden Ranking and the Leiden Manifesto: Methodology for Responsible University Research Assessment

The CWTS Leiden Ranking and the Leiden Manifesto are frequently mentioned in the same breath, but they are two different things produced by the same institute for two different purposes. The Leiden Ranking, published by Leiden University’s Centre for Science and Technology Studies (CWTS), is an actual bibliometric data product: it ranks universities using field-normalized citation and collaboration indicators computed from a defined dataset. The Leiden Manifesto is a ten-principle methodological statement, published in Nature in 2015, arguing for how metrics like the ones in the Ranking should — and should not — be used. This guide covers the Ranking’s actual methodology, the Manifesto’s principles, how CWTS itself reconciles producing one while advocating for the other, and how both relate to the broader responsible-metrics movement anchored by DORA and CoARA.

What the CWTS Leiden Ranking actually measures

The Leiden Ranking now exists in two editions, both maintained by CWTS:

• Traditional Edition — built on Web of Science data, the longer-running of the two editions (over two decades of iterations).
• Open Edition — launched in 2024, computed entirely from OpenAlex data in collaboration with the Curtin Open Knowledge Initiative (COKI), Sesame Open Science, and OurResearch, as a fully open and reproducible counterpart to the Traditional Edition. Each annual edition is tied to a specific OpenAlex snapshot (the 2024 edition used the August 2024 snapshot; the 2025 edition used the August 2025 snapshot).

Both editions are built around the same core design choices, documented on CWTS’s own methodology pages (traditional.leidenranking.com and open.leidenranking.com):

• Field normalization. Raw citation counts are not comparable across fields — a highly-cited biomedical paper and a highly-cited mathematics paper accumulate citations at very different rates. The Ranking normalizes for this using a field classification scheme distinguishing roughly 4,000 fields, and for publication year and document type.
• Mean Normalized Citation Score (MNCS). The average citation impact of a university’s publications relative to the field/year/document-type average. An MNCS of 2 means a university’s publications are cited twice the field average.
• Proportion of top-10% publications (PP(top 10%)). The share of a university’s publications that fall in the top 10% most-cited publications in the same field, year, and document type. CWTS and independent commentators generally treat this as a more robust indicator than MNCS, since it is less sensitive to a small number of extremely highly-cited outlier papers.
• Full counting vs. fractional counting. Full counting gives a co-authored publication full weight at every contributing university; fractional counting splits the weight of a collaborative publication across contributing universities. CWTS’s own guidance favors fractional counting for fairer field comparisons, since fields with heavier international collaboration (e.g. high-energy physics) would otherwise be systematically advantaged under full counting.

Beyond citation-impact indicators, the Ranking also publishes collaboration indicators (share of publications with international or industry co-authorship) and — in the Open Edition specifically — indicators drawing on OpenAlex’s open-access status metadata. Because methodology, snapshot dates, and even field-classification details change between annual editions, always cite the specific edition and year when referencing a Leiden Ranking figure, and link to CWTS’s current methodology page rather than restating exact index values from memory — they are revised annually.

The Leiden Manifesto’s ten principles

The Leiden Manifesto for Research Metrics was published by Diana Hicks, Paul Wouters, Ludo Waltman, Sarah de Rijcke, and Ismael Rafols in Nature in April 2015 (Hicks et al., “Bibliometrics: The Leiden Manifesto for research metrics,” Nature 520:429-431). It distills roughly a decade of scientometric research into ten guidelines for using quantitative indicators in research evaluation responsibly, summarized on CASRAI’s Leiden Manifesto dictionary entry:

1. Quantitative evaluation should support, not replace, qualitative expert assessment.
2. Measure performance against the research missions of the institution, group, or researcher.
3. Protect excellence in locally relevant research.
4. Keep data collection and analytical processes open, transparent, and simple.
5. Allow those evaluated to verify data and analysis.
6. Account for variation by field in publication and citation practices.
7. Base assessment of individual researchers on a qualitative judgement of their portfolio.
8. Avoid misplaced concreteness and false precision.
9. Recognise the systemic effects of assessment and indicators on researcher behavior.
10. Scrutinise indicators regularly and update them.

None of this is signable in the way DORA is — it is a set of methodological principles, cited widely across university rankings literature, research-assessment policy, and (as covered below) by CWTS’s own documentation of the Ranking it produces.

Why CWTS produces both the Ranking and the Manifesto

It is a fair question whether an institute that literally sells and maintains a university ranking product can also credibly co-author the field’s leading critique of ranking misuse. CWTS’s own position, reflected in the Leiden Ranking’s methodology documentation, is that the two are complementary rather than contradictory: the Manifesto’s principles (field normalization, transparency of method, avoiding false precision, recognizing systemic effects) are largely the design choices CWTS built into the Ranking itself. Field-normalized indicators like MNCS and PP(top 10%) exist specifically because raw, non-normalized citation counts violate Manifesto principle 6 (accounting for field variation). Publishing full methodology documentation and allowing users to choose full vs. fractional counting reflects principle 4 (openness and transparency). CWTS explicitly discourages using the Ranking to produce a single composite “best university” score or ordinal position-by-position league table, which tracks principle 8 (avoiding false precision) and principle 1 (supporting, not replacing, qualitative judgement).

In practice this means the Ranking is a data source built to Manifesto-consistent design principles, not proof that ranking a university at all is Manifesto-compliant by default — how an institution or a journalist subsequently uses or reports a Leiden Ranking position is a separate question from how CWTS computed it.

How this relates to DORA and CoARA

Research administrators comparing responsible-metrics frameworks should keep four distinct things straight:

• CWTS Leiden Ranking — an actual bibliometric data product; a set of computed university-level indicators.
• Leiden Manifesto — a ten-principle methodological statement about how to use indicators like the Ranking’s responsibly. Not an organization, not signable.
• DORA (San Francisco Declaration on Research Assessment) — a signable declaration, originating in 2012, focused heavily on eliminating journal-level metrics (particularly the Journal Impact Factor) from funding, hiring, and promotion decisions. Institutions and individuals become DORA signatories.
• CoARA (Coalition for Advancing Research Assessment) — a coalition of research-performing and research-funding organizations that formally commit, via the CoARA Agreement, to a broader, longer-term reform of institutional assessment practices across Europe and beyond.

All three responsible-metrics efforts (Manifesto, DORA, CoARA) converge on the same underlying critique the Leiden Ranking’s own methodology tries to design around: single, non-normalized, journal-level or composite metrics distort research behavior and misrepresent quality. For a direct comparison of the two signable frameworks, see CASRAI’s DORA vs CoARA comparison and the DORA and CoARA in Practice guide covering how institutions have actually rewritten promotion and tenure criteria.

Using the Leiden Ranking responsibly: practical guidance

For a research administrator citing, defending, or advising against reliance on a Leiden Ranking position, the Manifesto’s principles translate into concrete practice:

• Prefer PP(top 10%) over MNCS or raw citation counts when the goal is comparing research quality rather than volume — it is less distorted by a handful of extremely highly-cited papers and more directly reflects the “top decile” framing most institutional strategy documents actually care about.
• Never use a Leiden Ranking position to assess an individual researcher. The Ranking is computed at the institutional level from thousands of publications; applying an institution-level normalized indicator to a single person’s portfolio is exactly the “misplaced concreteness” the Manifesto warns against (principle 8), and CWTS does not design or market the Ranking for that use.
• State which edition, year, and counting method (Traditional or Open; full or fractional) a cited figure comes from — the two editions use different underlying databases (Web of Science vs. OpenAlex) and can produce meaningfully different indicator values and rankings for the same institution in the same year.
• Treat small movements in position as noise, not signal. A university moving a handful of ordinal ranks year over year on a list of 1,000+ institutions is rarely a meaningful change in underlying research performance; the underlying indicator values (MNCS, PP(top 10%)) are more informative than the ordinal rank itself.
• Use it alongside, not instead of, discipline-specific and mission-specific evidence. A single field-normalized university-wide indicator cannot substitute for the qualitative expert review the Manifesto’s first principle calls for in a tenure dossier, program review, or funding decision — see CASRAI’s Research Excellence guide for how national assessment frameworks combine metrics with expert panel review.

Related tools and methodology pages

Research administrators building out a full picture of how bibliometric assessment tools relate to each other may also find these useful: Is Web of Science Citation Data Reliable for Research Assessment? (the database underlying the Leiden Ranking’s Traditional Edition), VOSviewer (CWTS’s own bibliometric mapping and visualization tool), Snowball Metrics (a separate, institution-led standardization effort for cross-institution benchmarking indicators), and Scopus vs. Web of Science vs. OpenAlex for background on the underlying citation databases.

Frequently asked questions

Is the CWTS Leiden Ranking the same thing as the Leiden Manifesto?

No. The Leiden Ranking is a computed bibliometric ranking of universities. The Leiden Manifesto is a ten-principle methodological statement about how metrics — including rankings like the Leiden Ranking — should be used responsibly. Both come out of CWTS at Leiden University, which is the source of the shared name and the frequent confusion.

How is the CWTS Leiden Ranking calculated?

Using field-normalized bibliometric indicators — chiefly the Mean Normalized Citation Score (MNCS) and the proportion of publications in the top 10% most-cited (PP(top 10%)) — computed against roughly 4,000 field classifications and either full or fractional counting of co-authored publications. The Traditional Edition uses Web of Science data; the Open Edition (launched 2024) uses OpenAlex data.

What is the difference between the Leiden Ranking Traditional and Open Editions?

The Traditional Edition is built on Web of Science, a subscription citation database, and has run for over two decades. The Open Edition, launched in 2024 in collaboration with the Curtin Open Knowledge Initiative, Sesame Open Science, and OurResearch, is computed entirely from OpenAlex’s open data, making its methodology and underlying data fully reproducible by third parties.

Can the Leiden Ranking be used to evaluate an individual researcher?

No — it is an institution-level indicator computed from an entire university’s publication output. Applying it to a single researcher’s portfolio is precisely the kind of false-precision misuse the Leiden Manifesto’s principles warn against.

How does the Leiden Manifesto relate to DORA and CoARA?

All three are part of the same responsible-research-metrics movement, but they are different kinds of instruments. The Leiden Manifesto is a set of methodological principles with no membership mechanism. DORA is a declaration organizations and individuals sign, focused heavily on removing journal-level metrics from evaluation. CoARA is a coalition whose members commit, via a formal agreement, to a broader multi-year reform of institutional assessment practice.