Co-Citation Analysis
Also known as: co-citation mapping, historiograph, direct citation, citation pair analysis
Co-citation analysis is a method that identifies the intellectual structure of a research domain by examining how frequently pairs of documents are cited together in other publications. When two papers are frequently cited together in the literature, they are considered co-cited, indicating they are conceptually related or influential within the same research community. Developed by Henry Small in 1973, co-citation analysis maps the 'invisible colleges' of science—networks of researchers working on related problems—and reveals how knowledge domains evolve over time.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
+29 more
When to use it
Use co-citation analysis to map the intellectual structure of a mature research field, identify influential papers and authors, detect paradigm shifts or emerging research fronts, assess the intellectual foundation of a discipline, or construct historiographies of scientific concepts. It is particularly valuable when you need evidence of how the scholarly community itself perceives intellectual relationships (rather than relying on keywords or topical labels). Co-citation is ideal for longitudinal studies tracking how fields evolve. It is less useful for very new domains (insufficient citing documents) or for real-time identification of emerging work (requires accumulation of citing literature).
Strengths & limitations
- Community-validated: co-citation reflects how the entire research community views relationships, not editorial classifications.
- Robust to terminology drift: discipline-specific language changes do not affect co-citation, which relies on bibliographic data.
- Longitudinal analysis: comparing co-citation structures across decades reveals field evolution, paradigm shifts, and generational patterns.
- Identifies influential works: papers with high co-citation frequency are those the field collectively deems important.
- Scalable: computational methods allow analysis of thousands of citing documents and hundreds of focal papers.
- Requires time accumulation: newly published papers cannot be analyzed via co-citation until other papers cite them—typically 1–3 years after publication.
- Citation bias affects results: self-citation, citation cartels, and disciplinary citation practices (some fields cite more heavily) distort co-citation patterns.
- Does not capture negative citations: if papers A and B are frequently cited together because they contradict each other, co-citation will misrepresent them as intellectually aligned.
- Depends on citing literature availability: research using journals or books not indexed in citation databases are invisible to co-citation analysis.
Frequently asked
How is co-citation analysis different from bibliographic coupling?
Bibliographic coupling is forward-looking: papers A and B share references (both cite the same sources). Co-citation analysis is backward-looking: other papers cite both A and B together. Coupling reveals papers that addressed related problems; co-citation reveals papers the community treats as related. Coupling is useful for discovering newly published papers before citations accumulate; co-citation requires years of citing literature but is more robust to temporal changes in terminology.
What co-citation frequency threshold should I use to define a meaningful link?
Thresholds depend on field and dataset size. In highly cited fields (e.g., molecular biology, medicine), use 5–10 co-citations minimum. In less-cited fields (e.g., education, humanities), use 2–3. For focal sets of 30–50 papers over 5–10 years, 3–5 co-citations is typical. Always inspect the co-citation frequency distribution first: if 95% of pairs have <3 co-citations, you may have too few citing documents; if >50% have >10, your focal set may be too small or highly overlapping.
Should I include self-citations in co-citation analysis?
Self-citations inflate the coupling between papers by the same author or team. For unbiased co-citation analysis, exclude self-citations (author A cannot be co-cited with author B if only author A's papers cite both). However, if your goal is to assess an author's or institution's influence (including self-promotion), include self-citations and report them separately. Most published co-citation analyses exclude self-citations or weight them at 0.5× the value of external citations.
Can co-citation analysis be used on very new research areas (< 5 years old)?
Co-citation analysis is slow to activate for new areas because it requires other papers to cite the focal set. For areas <5 years old, you may have only 10–30% of the eventual citing literature. Use co-citation for descriptive mapping of what exists now, but combine it with keyword or semantic analysis for real-time emerging work detection. Alternatively, use keyword co-occurrence, which requires no citation lag and is sensitive to early terminology.
Sources
- Small, H. (1973). Co-citation in the scientific literature: A new measure of the relationship between two documents. Journal of the American Society for Information Science, 24(4), 265–269. DOI: 10.1002/asi.4630240406 ↗
- Small, H., & Griffiths, B. C. (1974). The structure of scientific literatures I: Identifying and graphing specialties. Science Studies, 4(1), 17–40. DOI: 10.1177/030631277400400102 ↗
How to cite this page
ScholarGate. (2026, June 4). Co-Citation Analysis. ScholarGate. https://scholargate.app/en/bibliometrics/co-citation-analysis
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Bibliographic CouplingBibliometrics↔ compare
- Journal Co-Citation AnalysisBibliometrics↔ compare
- Keyword Co-Occurrence AnalysisBibliometrics↔ compare
- Research Front IdentificationBibliometrics↔ compare
- Science MappingBibliometrics↔ compare