What bibliometric analysis is and what it can answer
Understand performance analysis and science mapping, how bibliometrics differs from a systematic review, and how to write bibliometric research questions.
The number of publications in almost every field has grown so fast that no researcher can read everything. A literature review of 50 carefully chosen papers is still essential, but it cannot show how a field of 5,000 papers is structured: which journals and countries dominate it, which ideas it builds on, which themes are rising and which are fading. Bibliometric analysis answers those questions by analysing publication metadata quantitatively.
What bibliometric analysis uses
Every record in Scopus or Web of Science comes with structured metadata: authors and their affiliations, title, abstract, author keywords and index keywords, source (journal), year, document type, the number of times it has been cited and, crucially, the list of references it cites. Bibliometric analysis counts, links and maps these fields. It does not read the full text; that remains your job when you interpret the results.
Two families of techniques
Following Donthu et al. (2021), most studies combine:
- Performance analysis: the productivity and impact of research constituents. How many documents per year? Which sources, authors, institutions and countries publish most and are cited most? What are their h-indices?
- Science mapping: the relationships between constituents. Which keywords appear together (conceptual structure)? Which references are cited together (intellectual structure)? Who collaborates with whom (social structure)?
Enrichment techniques such as clustering, thematic maps and trend analysis help interpret the maps.
Bibliometric analysis versus systematic literature review
| Systematic literature review | Bibliometric analysis | |
|---|---|---|
| Aim | Synthesise the findings of studies on a focused question | Map the structure and development of a broad field |
| Scope | Tens to a few hundred studies, screened in full text | Hundreds to thousands of records |
| Analysis | Qualitative synthesis or meta-analysis of results | Quantitative analysis of metadata |
| Typical output | What we know, the evidence, gaps | Who, where, what topics, how connected, what is emerging |
The two combine well. A common design runs a bibliometric analysis on the full search results to map the field, then a systematic review of the core documents (for example, the most cited documents in each cluster) to synthesise their content. If your question is about the size of an effect across studies, you need a meta-analysis instead; the meta-analysis tool on this site helps with that.
Writing bibliometric research questions
Good bibliometric questions map onto specific analyses, which keeps the results section focused:
| Research question | Analyses |
|---|---|
| RQ1. How has research on the topic developed over time, and where is it published? | Annual production, growth rate, most relevant sources, Bradford’s law |
| RQ2. Who are the most productive and influential authors, institutions and countries? | Most relevant and most cited authors, affiliations and countries, h-index |
| RQ3. What are the main research themes, and how have they evolved? | Co-occurrence network, thematic map, thematic evolution, trend topics |
| RQ4. What are the intellectual foundations of the field? | Most locally cited references, co-citation network, historiograph, RPYS |
| RQ5. How do researchers and countries collaborate? | Collaboration networks, world collaboration map, international co-authorship |
| RQ6. What are the emerging topics and future research directions? | Trend topics, keyword bursts, bibliographic coupling, niche and emerging themes |
The course collection
Throughout the course you work with the Studio’s demo collection: 418 documents on digital health (electronic health records, telemedicine, mobile health, artificial intelligence and related topics) published between 2008 and 2025. It is synthetic but behaves like a real Scopus export, so every analysis in the Studio works on it.
Explore the demo collection
- The link opens Bibliometrics Studio with the demo collection loaded.
- Open Main information: 418 documents, 2008–2025, 30 sources, 228 authors.
- Click through the groups in the menu (Overview, Sources, Authors, Documents, Conceptual, Intellectual and Social structure) to see the full set of analyses.
What bibliometrics cannot tell you
Citation counts measure attention, not quality: a paper can be highly cited because it is wrong. Database coverage is uneven across languages, regions and disciplines. Keyword maps depend on how authors label their work. And co-occurrence is not causation or agreement. Keep these limits in mind, mention them in your paper, and always interpret maps by reading the documents behind them.
Key takeaways
- Bibliometric analysis quantifies publication metadata: performance analysis plus science mapping.
- It maps broad fields; a systematic review synthesises focused evidence; they combine well.
- Write research questions that map onto specific analyses.
- Citations measure attention, not quality; interpret maps by reading the key documents.
Check your understanding
Answer each question to see the explanation. Finishing the quiz marks this lesson as complete.