CWTS Leiden Ranking: Bibliometric Transparency Without Composite Scores

The Leiden Approach

The CWTS Leiden Ranking, produced by the Centre for Science and Technology Studies at Leiden University in the Netherlands, takes a fundamentally different approach from most university rankings. It provides a set of bibliometric indicators for universities without combining them into a composite score. The indicators include the total number of publications, the number and proportion of highly cited publications, measures of scientific collaboration, and gender balance among authors. Users can select a subset of indicators and produce their own ranking based on their priorities. The underlying data comes from the Web of Science database, and the methodology is documented with academic rigour, including detailed explanations of how publications are assigned to institutions and how fractional counting handles multi-authored papers.

This approach prioritises transparency over simplicity. A traditional league table hides its methodological choices behind a single number; the Leiden Ranking makes those choices explicit and gives the user control over them. For researchers and research administrators, this is valuable because it allows them to examine specific aspects of research performance without the distortions introduced by composite scoring. For example, a university can be compared on its proportion of highly cited publications in a specific field, with fractional counting to account for co-authorship. This level of granularity is unavailable in most global rankings. However, the Leiden Ranking is exclusively research-focused and provides no indicators on teaching quality, student experience, or graduate outcomes. It should be understood as a research performance tool, not a general-purpose university ranking.

Interpreting Leiden Indicators Responsibly

The Leiden Ranking's transparency makes it less prone to misinterpretation than many other rankings, but responsible use still requires understanding what the indicators mean. The proportion of highly cited publications, for instance, depends on the field's citation norms and the time window selected. A 10 percent threshold for the top 10 percent of cited publications will look different from a 1 percent threshold for the top 1 percent. The ranking allows users to adjust these parameters, which is powerful but also means that results can be sensitive to choices. Two users with different parameter settings could reach different conclusions about the same university. This is not a flaw; it reflects the genuine complexity of research assessment.

Another important consideration is the database coverage. The Leiden Ranking uses Web of Science, which has known biases in its coverage of non-English-language journals, humanities publications, and some applied and professional fields. Universities that publish extensively in these areas will have less of their research captured by the Leiden indicators. The ranking itself is transparent about these limitations, and it encourages users to read the methodology documentation before interpreting results. For students, the Leiden Ranking is most useful at the doctoral level, where research performance indicators are directly relevant to supervisor and department selection. For undergraduate applicants, other rankings or data sources that include teaching and student experience measures are likely more useful.