What a bibliometric study actually covers
Bibliometric analysis treats a body of scientific literature as data. Rather than summarising papers one by one, it measures how much is being published, who is publishing it, where it appears, how often it is cited and how the underlying ideas connect. Donthu and colleagues (2021) describe the method as two complementary families of techniques. Performance analysis quantifies the contribution of authors, institutions, countries and journals. Science mapping reveals the relationships between those actors and between the concepts they write about.
A well designed study combines both. Counting alone tells a reader that a topic is growing; mapping explains which sub-themes are driving that growth and which intellectual traditions they draw on. Our bibliometric analysis services are built to deliver both layers in a form that slots straight into a thesis chapter or a journal manuscript.
Is bibliometric analysis the right method for your question?
The approach works best when the literature is large enough to support statistics and your question concerns the shape of a field rather than the effect size of a single intervention. Typical signals that it will add value include:
- Volume: several hundred or more relevant publications, too many to synthesise by close reading alone.
- Structure questions: you want to know which research streams exist, how they relate and how they have shifted over time.
- Positioning: a supervisor, funder or committee expects evidence that your topic sits in an active, under-explored corner of the field.
- Transparency: you are writing a systematic or scoping review and want a reproducible, PRISMA 2020 aligned overview (Page et al., 2021) before qualitative synthesis.
Three layers of results
Performance indicators
Annual output, compound annual growth, total and mean citations, h-index for authors and sources, and the most productive countries, institutions and journals.
Science maps
Co-authorship, keyword co-occurrence, co-citation and bibliographic coupling networks, plus thematic maps and thematic evolution across periods.
Classic laws
In the Comprehensive package we test Lotka's law of author productivity and apply Bradford's law to identify the core journal zone.
We deliberately avoid running every possible network just because the software allows it. A typical project selects three or four analyses that map directly onto your research questions, for instance one question about collaboration, one about themes and one about intellectual foundations. That keeps the paper focused and makes each figure earn its place in the manuscript.
How the project runs
Scope and search design
We agree the core concepts, synonyms, year range, document types and languages with you, then iterate the query until the known key papers are retrieved without flooding the set with noise.
Data retrieval
Full records with cited references are exported from Scopus, Web of Science or open sources depending on the package, using your institution's legitimate access wherever the database licence requires it.
Cleaning and merging
Records are deduplicated by DOI and title, and author, affiliation and keyword variants are harmonised through a thesaurus. Every exclusion is counted so the PRISMA flow is accurate.
Analysis and visualisation
Descriptive statistics and thematic maps are produced in Bibliometrix (Aria and Cuccurullo, 2017); networks are built in VOSviewer and exported at journal resolution.
Hand-over
You receive the full file set and a technical report, with 30 days of free revisions within the package scope.
Methodological choices we document
Small parameter decisions can change a map considerably, and reviewers increasingly ask authors to justify them. We record each choice and the reasoning behind it:
| Decision | Common options | Why it matters |
|---|---|---|
| Data source | Scopus, Web of Science, or both merged | Coverage differs by discipline; merging reduces single-database bias |
| Unit of analysis | Authors, sources, countries, keywords, cited references | Each unit answers a different question, so we run only what your research question needs |
| Counting method | Full or fractional counting | Fractional counting stops hyper-authored papers from dominating collaboration maps |
| Inclusion threshold | Minimum occurrences or citations | Keeps maps legible, usually in the range of 100 to 300 nodes |
| Normalisation | Association strength | The VOSviewer default described by van Eck and Waltman (2010) |
| Time slicing | Event-based or equal-size periods | Allows a fair comparison of themes across phases of the field |
From our files to your methods section
A replicable methods section names the database and indexes searched, the exact query, the retrieval date, the filters applied, the number of records at each stage, the software and version, and every threshold and normalisation setting. Our technical report lists all of these. In the Advanced and Comprehensive packages it is written in English so you can adapt it to your own voice.
The results and discussion are a different matter. Explaining why a cluster matters, linking trends to theory and drawing implications is scholarly work that belongs to you. Our report describes what the data shows, for example which cluster is largest or which terms appear most recently, without making those arguments for you. If anything in the report is unclear, we explain the technical step behind it during the 30-day revision window.
Where our work stops
Read our full research integrity policy, browse bibliometric analysis examples, compare packages on the pricing page or start a project through the order form.