Introduction
Collaboration affects citation patterns, with domestically produced research generally receiving fewer citations than internationally co-authored work. Research itself is becoming increasingly international, with collaborations rising from 8% in the 1990s to over 25% today. Performance assessments that overlook these trends may fail to distinguish whether higher impact results from research quality or collaboration patterns.
Category Normalized Citation Impact (CNCI) is a widely used method for assessing research impact. This ‘standard’ CNCI approach compares the accumulated citation count for a document to other documents of the same type, published in the same year and research area.
Collab-CNCI applies one additional step to normalize citation counts also for the collaboration type, distinguishing between domestic and international research. Next to Collab-CNCI, InCites includes a suite of collaboration indicators and filters to:
- Evaluate impact by gaining a balanced view of research collaboration influence versus research quality, revealing true impact drivers.
- Balance collaboration partners by maintaining strong domestic and international networks to ensure sustained, resilient research impact over time.
- Align policies and funding to prioritize and target funds where collaborations yield the highest impact.
Watch the below video to learn about Collab-CNCI:
Collaboration Types
Collab-CNCI compares citation impact to global expected citations, normalized by subject, year, document type and collaboration type. Each publication falls into one of five collaboration types, determined by:
- The number of distinct countries in the author addresses
- If all addresses are from the same country, the number of organizations involved
Overview of Collaboration types
-
Domestic – Single: publications where all authors are affiliated with the same institution in a single country.
- More formally, all author addresses share the same country and are affiliated with the same institution
- Both unified and non-unified organizations are considered:
- If an organization is non-unified, the published name from the address is used.
- If an address is resolved, the unified organization name is used.
- If an address resolves to multiple unified organizations with a parent-child relationship, only the child organizations are counted.
- Domestic – Multiple: Publications where all authors are affiliated with multiple institutions within the same country.
- International – Bilateral: Publications with author addresses from exactly two distinct countries.
- International – Trilateral: Publications with author addresses from exactly three distinct countries.
- International – Quadrilateral +: Publications with author addresses from four or more distinct countries.
Counting countries and organizations
- All addresses are considered, including those not resolved to a unified organization entity.
- The country is identified directly from the address field, not from the unified organization’s entity country.
- For domestic publications (i.e., addresses from only one country), the number of organizations is counted. Both unified and non-unified organizations are considered:
- If an address isn’t resolved, the published organization name from the address is used.
- If an address resolves to multiple unified organizations and there is a parent-child relationship, only the child organizations are counted.
- The collaboration type does not depend on the number of authors. Publications by a single author can still be classified as Domestic – Multiple, or International.
Interim data limitations (unverified organizations)
- A small fraction (~0.15%) of publications may be incorrectly categorized as “Domestic - Single” instead of “Domestic - Multiple”, if an organization’s parent-child relationship is not yet manually reviewed.
- The next data release (end of March) will correct these. A smaller subset of French organizations may take longer.
- A link to the list of affected organizations is provided Unverified Organizations.
Detailed examples
One author but International - Trilateral
- Example: WOS:000611933300001 has 1 author with 3 addresses in 3 countries.
- Classified as International - Trilateral because the same author is collaborating via multiple affiliations in multiple countries.
Mixing resolved and unresolved addresses
- Example: WOS:001325039300012 has 2 addresses in the same country. One address is resolved to Harvard University, the other is unresolved and has published organization name Wentworth Institute of Technology.
- Because both are recognized as distinct organizations, the paper is Domesti - Multiple.
Address country vs organization entity country
- Example: WOS:000551549800007 has 3 countries in the addresses, Canada, South Korea and USA. One of the resolved organizations is Toyota Motor Corporation with publication address country USA, but having Japan as the organization entity country.
- Because countries of the organization entities are not counted, the paper is International - Trilateral
Parent-Child Relationships
- Example: WOS:000606519500004 has addresses all resolving to “Harvard University” and “Harvard Medical School.” Since Harvard Medical School is a child of Harvard University, it counts only the child (Harvard Medical School), resulting in Domestic - Single.
- Example: WOS:000581916800002 has 4 authors, one address, resolved to 4 organizations: University of California System, a parent of University of California San Francisco, a parent of UCSF Medical Center, a parent of UCSF Helen Diller Family Comprehensive Cancer Center. Because of the parent-child relationships, only the last one counts resulting in Domestic – Single.
- Example: WOS:000770361800073 has addresses resolving to “Harvard University”, “Harvard Medical School” and “Brigham & Women’s Hospital”. The last two are both children of Harvard University but not parent-child of each other, so it’s Domestic - Multiple.
- Example: WOS:000086021100001 with 1 author, 1 address, but resolving to two independent institutions, Sorbonne University and Ecole Normale Superiere (ENS), hence Domestic – Multiple.
Collab-CNCI and other Collaboration Indicators
The Collaboration Category Normalized Citation Impact (Collab-CNCI) of a document is calculated by dividing the actual count of citing items by the expected citation rate for documents with the same document type, year of publication, subject area and collaboration type. When a document is assigned to more than one subject area an average of the ratios of the actual to expected citations is used.
The Collab-CNCI of a set of documents, for example the collected works of an individual, institution or country/region, is the average of the Collab-CNCI values for all the documents in the set.
A Collab-CNCI value of 1 represents performance at par with world average. Values above 1 mean that the publication set is cited more than expected, above average.
Next to the new Collab-CNCI indicator, we also introduce the Collab-CNCI indicator for each collaboration type, along with aggregated Domestic and International indicators. We’re also adding document count and percentage indicators for each type, showing both the number and proportion of publications in each category.
-
Collab-CNCI by Collaboration Type
- E.g., Collab-CNCI Domestic, Collab-CNCI Domestic Single, Collab-CNCI International Bilateral, etc.
- Enables deeper analysis by each distinct collaboration type.
-
Document Counts and Percentages
- Domestic Documents, International Documents, Domestic Single Documents, International Bilateral Documents, etc.
- Each has a complementary percentage indicator: e.g., % Domestic Single Documents, % International Bilateral Documents, and so on.
Old vs. new collaboration concepts
InCites historically offered three concepts: International Collaboration, Domestic Collaboration, and Organization Only Collaboration. These are still available but considered deprecated in favor of the new collaboration categories. The differences are summarized here:
| Legacy Concept | Definition | Comparison to New Types |
| International Collaborations | At least 2 authors and at least 2 distinct countries. |
Corresponds to the combination of International - Bilateral, -Trilateral, and -Quadrilateral+, but requires ≥ 2 authors. The new approach also classifies single-author multi-country papers as international. Instead use the indicators International Documents and % International Documents. Optionally combine with the Authors per Document filter |
| Domestic Collaborations | At least 2 authors, all from the same country, with at least 2 distinct published organization names. | Similar to Domestic-Multiple, but requires ≥ 2 authors and uses published organization names from the address and does not account for multiple resolved organizations in a single address. Hence it may undercount multiple institutions. |
| Organization Only Collaborations | At least 2 authors all from the same resolved organization (unresolved addresses are disregarded; parent-child not collapsed). | Conceptually close to Domestic-Single but excludes addresses not unified and does not collapse parent-child relationships properly. For accurate single-institution metrics, use the new Domestic Single Documents and optionally combine with “≥2 authors” filter. |
Recommendation: Use the new collaboration types (Domestic-Single, Domestic-Multiple, International-Bilateral/Trilateral/Quadrilateral+) or the aggregated Domestic/International categories for more precise and up-to-date analysis.
All Collaboration Indicators
| Collaboration Indicators | Descriptions |
| Collab-CNCI | Normalized citation impact adjusted for subject, year, document type and collaboration type. |
| Domestic Documents | Number of publications for all domestic publications (both single- and multiple institutions within one country). |
| International Documents | Number of publications for all international collaborations (between institutions in different countries). |
| % Domestic Documents | Percentage of publications for all domestic publications (both single- and multiple institutions within one country). |
| % International Documents | Percentage of publications for all international collaborations (between institutions in different countries). |
| Collab-CNCI Domestic | Normalized citation impact for all domestic publications (both single- and multiple institutions within one country), adjusted for subject, year, document type and collaboration type. |
| Collab-CNCI International | Normalized citation impact for all international collaborations (between institutions in different countries), adjusted for subject, year, document type and collaboration type. |
| Domestic Single Documents | Number of publications from single-institution research within one country. |
| Domestic Multiple Documents | Number of publications from collaborations among multiple institutions within one country. |
| International Bilateral Documents | Number of publications from collaborations between institutions in two different countries. |
| International Trilateral Documents | Number of publications from collaborations between institutions in three countries. |
| International Quadrilateral+ Documents | Number of publications from collaborations between institutions in four or more countries. |
| % Domestic Single Documents | Percentage of publications from single-institution research within one country. |
| % Domestic Multiple Documents | Percentage of publications from collaborations among multiple institutions within one country. |
| % International Bilateral Documents | Percentage of publications from collaborations between institutions in two different countries. |
| % International Trilateral Documents | Percentage of publications from collaborations between institutions in three countries. |
| % International Quadrilateral+ Documents | Percentage of publications from collaborations between institutions in four or more countries. |
| Collab-CNCI Domestic Single | Normalized citation impact for single-institution research, adjusted for subject, year, document type and collaboration type. |
| Collab-CNCI Domestic Multiple | Normalized citation impact for collaborations among multiple institutions within one country, adjusted for subject, year, document type and collaboration type. |
| Collab-CNCI International Bilateral | Normalized citation impact for collaborations between institutions in two different countries, adjusted for subject, year, document type and collaboration type. |
| Collab-CNCI International Trilateral | Normalized citation impact for collaborations between institutions in three countries, adjusted for subject, year, document type and collaboration type. |
| Collab-CNCI International Quadrilateral+ | Normalized citation impact for collaborations between institutions in four or more countries, adjusted for subject, year, document type and collaboration type. |
| Industry Collaborations | Number of publications with at least 2 organizations and one of them listing its organization type as corporate or global corporate. |
| % Industry Collaborations | Percentage of publications with at least 2 organizations and one of them listing its organization type as corporate or global corporate. |
| Organization only Collaborations | Deprecated, use 'Domestic Single' instead. Number of publications with at least 2 authors, from the same resolved organization. Unresolved organizations are disregarded, and when an address is resolved to multiple organizations, the relationship between them is also disregarded. When using Domestic Single instead, optionally combine with filter on 'Authors per Document' set to minimum 2. |
| % Organization only Collaborations | Deprecated, use '%Domestic Single' instead. Percentage of publications with at least 2 authors, from the same resolved organization. Unresolved organizations are disregarded, and when an address is resolved to multiple organizations, the relationship between them is also disregarded. When using Domestic Single instead, optionally combine with filter on 'Authors per Document' set to minimum 2. |
| Domestic Collaborations | Deprecated, instead use 'Domestic Multiple Documents'. Number of publications for all domestic publications with at least 2 authors and at least 2 distinct organizations, all in the same country. The published organization name is used, not the resolved organization name(s). When using Domestic Multiple instead, optionally combine with filter on 'Authors per Document' set to minimum 2. |
| International Collaborations | Deprecated, instead use 'International Documents'. Number of publications for all international collaborations with at least 2 authors and at least 2 different countries. When using 'International Documents', optionally combine with filter on 'Authors per Document' set to minimum 2. |
| % Domestic Collaborations | Deprecated, instead use '% Domestic Multiple Documents'. Percentage of publications for all domestic publications with at least 2 authors and at least 2 distinct addresses, all in the same country. When using Domestic Multiple instead, optionally combine with filter on 'Authors per Document' set to minimum 2. |
| % International Collaborations | Deprecated, instead use '% International Documents. Percentage of publications for all international collaborations with at least 2 authors and at least 2 different countries. When using 'International Documents', optionally combine with filter on 'Authors per Document' set to minimum 2. |
Industry Collaborations and % Industry Collaborations: the number/percentage papers with at least one organization listing its organization type as corporate or global corporate. When analyzing organizations, the rule is different. In this case, when the listed organization is a corporate or global corporate, the publications from the listed organization will be excluded from the calculation. This means that only publications where the listed corporate / global corporate, also collaborates with another corporate / global corporate are counted as industry collaboration papers.
Best practices for collaboration analysis
Combine filters & indicators
If you are interested in analyzing a specific collaboration type segment, it’s best to use the collaboration filter. This allows to also view other indicators for the filtered segment. For example, if you only want to study international collaborations, you could apply the collaboration filter. Filter for International – Bilateral, Trilateral, and Quadrilateral+. Add the Collab-CNCI indicator, and others like Documents in Top 10% or % Industry Collaborations to get a fuller picture.
If you are rather interested to compare different segments against each other, use the Collab-CNCI, document counts and percentage documents indicators per collaboration type, and view them next to each other.
Apply ‘Authors per Document’ filter
The new classifications do not require multiple authors. If you rather want to analyze collaborations between authors, rather than institutions, exclude single-author edge cases, use the Authors per Document filter set to a minimum of 2.
When to use CNCI or Collab-CNCI
Since CNCI and Collab-CNCI calculate citation impact differently, directly comparing their values isn’t meaningful. Instead, they provide complementary insights that allow for a deeper analysis. A more useful approach is to examine how each metric changes over time rather than looking at absolute values.
For example, CNCI is expected to rise as the number of international publications grows. However, analyzing the Collab-CNCI trend over time can reveal whether this increase is due to research excellence or simply a shift in collaboration patterns.
Collab-CNCI also adjusts for expected trends, such as higher citation impact for international collaborations and lower impact for domestic ones. Because international collaborations generally attract more citations, this adjustment can lower overall CNCI averages when looking at institutions or regions. By comparing the trends of both metrics while considering the balance of international vs. domestic publications, we can better understand the factors driving changes in citation impact.
A general guideline is to use CNCI for broad comparisons across fields and countries, and Collab-CNCI for a more detailed and nuanced view of collaboration types.
Focus on Domestic and International research
While international collaboration is widely recognized as a driver of higher citation impact, strong domestic collaborations remain foundational for building a sustained research environment. Large multinational (quadrilateral-plus) collaborations, while impactful, can be difficult to establish and maintain in the long run. Therefore, domestic and bi- or trilateral international collaborations often represent a more practical and reliable strategy for fostering long-term growth in citation impact.
Impact profile, not only metrics
To gain a complete picture, look beyond average CNCI and Collab-CNCI values.
The Impact Profile shows the overall distribution, highlighting for instance, the number of uncited documents. To display the Impact Profile for a specific collaboration segment, do use the Collaboration Type filter, in combination with the Collab-CNCI indicator.
Use Cases
Evaluate international collaboration impact
When examining international collaborations, CNCI can overstate research impact as such publications typically receive more citations. Collab-CNCI adjusts for collaboration type, offering a balanced view of research quality. Comparing countries or organizations using both metrics reveals whether higher impact stems from research excellence or collaboration patterns, enabling informed, strategic improvements.
For example, this graph shows Japan's CNCI and Collab-CNCI over time for international publications. While CNCI suggests rising research impact, Collab-CNCI remains stable and slightly below global average, indicating that collaboration, not just research quality, drives the higher impact.
Benchmark domestic research against nations
The Locations module provides valuable insights into how domestic collaborations influence research impact over time. Use the Collab-CNCI Domestic-Single (single institution) or Domestic-Multiple (multi-institution, same country) indicators to track trends across countries. These insights can help decision-makers refining policies and strengthen the national research ecosystem.
For example, this graph tracks Collab-CNCI Domestic-Single across countries, highlighting single-institution research impact over time. Singapore and the USA lead, while France and Japan remain below the global average.
Belgium, though below 1 in overall CNCI, performs above average in domestic-single research, showing stronger impact compared to domestic research in the world.
Benchmark collaboration trends and strategies
Relying solely on CNCI can misrepresent research impact when collaboration patterns vary. Collab-CNCI provides deeper insights. Benchmarking institutions using Collab-CNCI per collaboration type, helps identify strengths and refine policies. For example, an institution with strong domestic impact but fewer international collaborations may have a lower CNCI, while Collab-CNCI reveals its true excellence.
For example, Sichuan University’s performance might appear undervalued when looking at CNCI because the percentage of international collaborations is much higher for other institutions. Collab-CNCI corrects for this, ranking Sichuan University on second place for these organizations.
Identify the partnerships with the strongest impact
Use the Organizations module to list international academic institutions engaged in bilateral collaborations with your institution. Apply the bilateral collaboration filter alongside Collab-CNCI and Top 10% documents for a detailed assessment of both research productivity and impact.