Mapping the margins: visibility of Humanities, Arts, and Social Sciences in Dimensions
Research data bites: 17.
Key takeaways:
Humanities, Arts and Social Sciences (HASS) account for a smaller share of indexed publications in Dimensions, but their growth rate exceeds that of most STEM fields.
The diversity of HASS publication types is well-represented in Dimensions, particularly chapters and monographs.
In an era where data drives research discovery, the question of comprehensive coverage in platforms like Dimensions.ai is crucial. We are often asked what coverage Dimensions has in terms of area of research, which is a fair question to ask since bibliometric databases have historically underrepresented the Humanities, Arts, and Social Sciences (HASS). However, Dimensions has an inclusive approach to publication indexing; it incorporates content from data platforms like CrossRef, PubMed, and arXiv, as well as publisher-submitted records.
In the next few bites we will look at the Dimensions’ coverage; today we start with research disciplines. In this bite, we will use: the Fields of Research (FoR), the publication classification, and Dimensions on GBQ (as usual, code at the end); and use the Compound Annual Growth Rate (CAGR) to measure growth.
Fields of Research coverage
As in my previous analysis, I will use the following classification based on the ANZSRC FoR for the 5 broad disciplines covering STEM and HASS:
Biological Sciences and Biotechnology (BSB),
Engineering, Information and Computing Sciences (EIC),
Mathematics, Physics, Chemistry and Earth Sciences (MPCE),
Humanities and Creative Arts (HCA),
Social, Behavioural and Economic Sciences (SBE).
In Dimensions, HCA and SBE are the two smallest disciplines, with respectively 18.3% and 14.2% of all publications between 2013 and 2024. However, in terms of growth, SBE has the highest CAGR (8.2%) while HCA follows closely (7.5%). Both outpace BSB (5.2%) and MPCE (4.3%). The only STEM area with comparable growth is EIC (7.2%).
The higher growth rates in SBE and HCA suggest a potential shift in research focus or improved data capture in these disciplines. But since the growth appears steady over time, it’s more likely that these fields are genuinely producing more publications.
Looking at the STEM vs SHAPE comparison below—where STEM stands for Science, Technology, Engineering, and Mathematics, and SHAPE represents Social Sciences, Humanities, and the Arts for People and the Economy—we see that STEM accounts for 76.2% of all publications indexed in Dimensions, while SHAPE makes up nearly a quarter at 23.7%. However, STEM’s growth rate is just 5.5% (the COVID-19 publication surge has now leveled off), while SHAPE is growing at 7.9% (with a smaller pandemic-related bump, less pronounced in this view).
By document type
Concerns about discipline coverage often highlight differences in publication types. HASS disciplines are typically associated with a higher volume of books, chapters, and monographs, while fields like Engineering and Computer Science tend to produce more conference proceedings. Dimensions accounts for this diversity by indexing a broad range of publication types, including: articles, books (including edited volumes), chapters, monographs (usually focused on a single subject and author), preprints, proceedings, and seminars (via Cassyni).
The figure below shows both the count and percentage of publications by type for each disciplinary group.
HCA and SBE have article proportions comparable to STEM fields: HCA falls between EIC and MPCE, while SBE mirrors MPCE.
EIC has the lowest article percentage due to its high number of proceedings.
HCA has the highest share of chapters and monographs, confirming expectations.
These trends suggest that while HASS disciplines may favour different publication types, Dimensions’ model does accommodate this diversity reasonably well.
Conclusion
While Humanities, Arts, and Social Sciences remain smaller in terms of total publication count within Dimensions, their growth trajectory is strong—often faster than traditional STEM fields. The diversity of types, from chapters to monographs, is well captured, contradicting assumptions that HASS output is systematically excluded from bibliometric databases.
This robust coverage supported my colleagues Juergen and Kathryn to co-author a report with the Academy of Social Sciences (AcSS), showing—among other findings—that the social sciences are essential to international collaboration and tackling shared global challenges. Briony, Juergen, Daniel, and I also co-authored a preprint with the British Academy on the importance of SHAPE to the UK research ecosystem. It shows that UK SHAPE research not only exceeds STEM in global impact metrics but also plays a critical role in addressing societal challenges alongside technological advancement.
In the next bite, we will move from thematic coverage to examine the geographic distribution of journals—and then turn to patterns in language diversity.
Code
Discipline trends
First the 5-discipline classification:
SELECT pubs.year, disc.Discipline, COUNT(DISTINCT pubs.id) AS publication_count
FROM `dimensions-ai.data_analytics.publications` pubs
CROSS JOIN UNNEST(pubs.category_for.first_level.full) AS ffl
LEFT JOIN `disciplines_ffl` disc
ON ffl.code = CAST(disc.ffl AS STRING)
WHERE pubs.year BETWEEN 2013 AND 2024
GROUP BY pubs.year, disc.Discipline The STEM vs SHAPE:
SELECT
pubs.year,
CASE
WHEN CAST(ffl.code AS INT64) IN (33, 35, 36, 38, 39, 43, 44, 47, 48, 50, 52) THEN "Social Sciences, Humanities and the Arts for People and the Economy (SHAPE)"
WHEN CAST(ffl.code AS INT64) IN (30, 31, 32, 34, 37, 40, 41, 42, 46, 49, 51) THEN "Science, Technology, Engineering, and Mathematics (STEM)"
ELSE NULL
END AS Discipline,
COUNT(DISTINCT pubs.id) AS publication_count
FROM `dimensions-ai.data_analytics.publications` pubs
CROSS JOIN UNNEST(pubs.category_for.first_level.full) AS ffl
WHERE pubs.year BETWEEN 2013 AND 2024
GROUP BY pubs.year, DisciplineDocument type
SELECT type, disc.Discipline, COUNT(DISTINCT pubs.id) AS publication_count
FROM `dimensions-ai.data_analytics.publications` pubs
CROSS JOIN UNNEST(pubs.category_for.first_level.full) AS ffl
LEFT JOIN `disciplines_ffl` disc
ON ffl.code = CAST(disc.ffl AS STRING)
WHERE pubs.year BETWEEN 2013 AND 2024
GROUP BY type, disc.Discipline