PaperlayerView exact revision on arXiv

Source-linked AI summary

The Journal Coverage of Web of Science, Scopus and Dimensions: A Comparative Analysis

Vivek Kumar Singh, Prashasti Singh, Mousumi Karmakar, Jacqueline Leta, Philipp Mayr

arXiv:2011.00223v2cs.DL

TL;DR

Direct journal-coverage comparisons of Web of Science, Scopus and Dimensions were limited. This study compares their journal lists and country publication records; coverage differs substantially, with Dimensions covering 82.22% more unique journals than Web of Science.

  • Problem

    Direct comparison of Dimensions’ journal coverage with Web of Science and Scopus was lacking, leaving limited evidence on how exhaustive or selective it is relative to them.

    Source: Discussion · p. 14

  • Method

    The study compared the databases’ master journal lists and publication records for 20 highly productive countries.

    Source: Data & Methodology · p. 9

  • Results

    82.22% more unique journals are covered by Dimensions than by Web of Science; database coverage also varies across countries’ research outputs and subject areas.

    Source: Conclusion · p. 16

  • Takeaways & Limitations

    Different database choices may produce different outcomes in country-level bibliometric evaluations because research output volumes, ranks and global shares vary.

    Source: Conclusion · p. 16

    The comparison covers journals only; books and conference publications were outside its scope.

    Source: Limitations · p. 16

Abstract

from arXiv · show

Traditionally, Web of Science and Scopus have been the two most widely used databases for bibliometric analyses. However, during the last few years some new scholarly databases, such as Dimensions, have come up. Several previous studies have compared different databases, either through a direct comparison of article coverage or by comparing the citations across the databases. This article attempts to compare the journal coverage of the three databases: Web of Science, Scopus and Dimensions. The most recent master journal lists of the three databases have been used for the purpose of identifying the overlapping and unique journals covered in the databases. The results indicate that the databases have significantly different journal coverage, with the Web of Science being most selective and Dimensions being the most exhaustive. About 99.11% and 96.61% of the journals indexed in Web of Science are also indexed in Scopus and Dimensions, respectively. Scopus has 96.42% of its indexed journals also covered by Dimensions. Dimensions database has the most exhaustive coverage, with 82.22% more journals covered as compared to Web of Science and 48.17% more journals covered as compared to Scopus. We also analysed the research outputs for 20 highly productive countries for the 2010-2019 period, as indexed in the three databases, and identified database-induced variations in research output volume, rank and global share of different countries. In addition to variations in overall coverage of research output from different countries, the three databases appear to have differential coverage of different disciplines.

A brief overview of the three scholarly databases … Dimensions

The overview traces Web of Science, Scopus, and Dimensions from their founding dates to their distinct coverage models and data offerings. Web of Science and Scopus curate selective publication indexes, while Dimensions combines publications and citations with linked research data from multiple sources.

A brief overview of the three scholarly databases

  • The databases were created in 1964, 2004, and 2018, respectively, making Dimensions the newest of the three.Web of Science is the oldest, followed by Scopus.Source: A brief overview of the three… · p. 3

Web of Science

  • Web of Science began as the Science Citation Index in 1964, added social-science and humanities indexes, and launched as Web of Science in 1997.The Emerging Sources Citation Index launched in 2015 to provide early visibility for titles being evaluated for the classical indexes (Somoza‐Fernández, Rodríguez‐Gairín, & Urbano, 2018).Source: Web of Science · p. 3
  • In 2020, Web of Science Core Collection covered more than 74.8 million scholarly data and datasets and 1.5 billion cited references across 254 subject-disciplines.Its SCIE, SSCI, and AHCI indexes are selective, with journal indexing decisions made at regular intervals.Source: Web of Science · p. 3

Scopus

  • Scopus, launched in 2004, covered about 23,452 active journal titles, 120,000 conferences, and 206,000 books in its October 2019 guide.It reported about 77.8 million core records; its master journal list contains more entries than the active-title count, including inactive journals.Source: Scopus · p. 4
  • Scopus offers search, discovery, and analysis tools for finding publications and tracking citations by country, affiliation, and research area.Downloaded publication data usually contains 43 fields.Source: Scopus · p. 4

Dimensions

  • Dimensions’ May 2020 journal list contained more than 74,000 entries, alongside publications, citations, clinical trials, patents, policy documents, and altmetric data.Its indexed journal records grew from more than 50,000 in April 2019; by September 2019 it contained 109 million publications and about 1.1 billion citations.Source: Dimensions · p. 4
  • Dimensions combines publication and citation data with grants, clinical trials, patents, policy documents, and altmetric information.Its 2018 launch version contained about 90 million publications and more than 873 million citation links.Source: Dimensions · p. 4
  • Dimensions uses Crossref and PubMed as its data spine, then enhances records with information from sources including DOAJ, Open Citations, clinical-trial registries, and policy data.It offers three access forms—Dimensions, Dimensions Plus, and Dimensions Analytics—and publication downloads usually contain 71 fields.Source: Dimensions · pp. 4–5

Related Work · Studies on direct coverage comparison of databases · Studies on comparing citation counts in databases

Earlier studies compared databases either by examining which journals or articles they indexed or by comparing citation counts for the same publications. Their findings varied across databases, disciplines, samples, and measures.

Related Work

  • Prior studies examined both database coverage and citation differences, providing context for the comparisons reviewed here.Direct coverage comparisons considered journals or articles indexed in different databases; citation comparisons examined differences for a given sample of articles.Source: Related Work · p. 5

Studies on direct coverage comparison of databases

  • Early studies found Scopus broader in STM and oncology, while journal overlap, country-level effects, and shared-core citation impact varied across samples (Mayr & Walter, 2007; Gavel & Iselid, 2008; Lopez-Illescas et al., 2008, 2009; Vieira & Gomes, 2009).Lopez-Illescas et al. (2008) reported Scopus covered 90% more oncological journals than Web of Science; Vieira & Gomes (2009) found about two-thirds of documents in either database appeared in both.Source: Studies on direct coverage comparison… · p. 5
  • Scopus covered more journals but with lower impact and more recent articles, while WoS and Scopus coverage could bias evaluation toward sciences and English-language journals (Chadegani et al., 2013; Mongeon & Paul-Hus, 2016).Mongeon & Paul-Hus compared 13,605 WoS and 20,346 Scopus journals with 63,013 in Ulrich’s directory.Source: Studies on direct coverage comparison… · p. 6
  • Database coverage varied by discipline and population: Norway’s Scopus and WoS totals were 72% and 69%, respectively, and both covered humanities less than medicine and technology (Aksnes & Siversten, 2019).Other studies reported stable quarterly publication and citation growth across WoS, Scopus, and Google Scholar, and found PubMed focused on life sciences and biomedicine (Harzing & Alakangas, 2016; AlRyalat, Malkawi & Momani, 2019).Source: Studies on direct coverage comparison… · p. 6

Studies on comparing citation counts in databases

  • Citation comparisons found valuable citations missing from WoS, about 20% more coverage in Scopus than WoS for biomedical research, and substantial h-index variation across databases (Yang & Meho, 2006; Falagas et al., 2008; Bar-Illan, 2008).Google Scholar citation retrieval was also reported to have technical problems or inconsistent accuracy, and h-index and citation counts did not vary linearly (Yang & Meho, 2006; Falagas et al., 2008; Bar-Illan, 2008).Source: Studies on comparing citation counts… · p. 6
  • Scopus yielded 14.7% more Health Sciences citations than WoS, while Google Scholar covered management better than WoS, which the authors cautioned against using alone for impact measurement (Torres-Salinas, Lopez-Cózar & Jiménez-Contreras, 2009; Mingers & Lipitakis, 2010).The citation difference did not correspond to the databases’ coverage difference; Google Scholar’s broader management coverage was described as somewhat unreliable.Source: Studies on comparing citation counts… · pp. 6–7
  • Citation-resource rankings depended on evaluation level: Scopus led at the macro level, but WoS led at the micro level and retrieved the most unique South African environmental-sciences items (Adriaanse & Rensleigh, 2011, 2013).The 2013 study ranked Google Scholar second and Scopus third for unique items.Source: Studies on comparing citation counts… · p. 7
  • Google Scholar citation data largely subsumed WoS and Scopus and added unique citations, while WoS and Scopus missed more highly cited Social Sciences and Humanities documents than those in natural, life, and health sciences (De Winter & Dimitra Dodou, 2014; Martin-Martin et al., 2018; Martín-Martín, Orduna-Malea & López-Cózar, 2018).About half of Google Scholar’s unique citations were not from journals; its citations included 95% of WoS citations and 92% of Scopus citations.Source: Studies on comparing citation counts… · p. 7

Studies on comparison of newer scholarly databases · Data & Methodology

Prior comparisons found substantial differences in database coverage and suggested that comprehensiveness and selectivity can both be valuable. This study compares journal lists and country-level publication records across Web of Science, Scopus, and Dimensions, then examines output measures and subject-area distributions.

Studies on comparison of newer scholarly databases

  • Thelwall (2018) reported a 0.96 narrow-field correlation between Scopus and Dimensions citation counts in 2012, and 97% of Scopus articles with DOIs were found in Dimensions.The study concluded that Dimensions could plausibly support general citation analyses as an alternative to Scopus and Web of Science.Source: Studies on comparison of newer… · p. 7
  • Harzing (2019) found Crossref and Dimensions coverage similar to or better than Scopus and Web of Science but below Google Scholar and Microsoft Academic; Huang et al. (2020) found asymmetric institutional DOI coverage.Huang et al. inferred that relying on one database could seriously disadvantage some institutions in assessments.Source: Studies on comparison of newer… · p. 8
  • Visser et al. (2019) found Dimensions had higher coverage than Web of Science and Scopus, covering about 78% of Scopus publication records; their 2020 comparison found Microsoft Academic perhaps most comprehensive.The 2020 study also found many journal articles exclusive to Scopus and an even larger number exclusive to Dimensions and Crossref, while nearly all Web of Science journal articles were covered by Scopus.Source: Studies on comparison of newer… · p. 8
  • Google Scholar covered almost 88% of citations; Microsoft Academic found 60% overall, including 82% of Scopus and 86% of Web of Science citations, while Dimensions covered 84% and 88%, respectively.For comprehensive citation counts, Martin-Martin et al. (2020) favored Google Scholar; when a complete list of citing sources is needed, they favored Microsoft Academic.Source: Studies on comparison of newer… · pp. 8–9

Data & Methodology

  • The study used journal master lists and 2010–2018 publication records for 20 selected countries; the lists contained 13,610 Web of Science, 40,385 Scopus, and 77,471 Dimensions entries.The Web of Science and Scopus lists were updated in June 2020; the curated Dimensions list was updated in May 2020.Source: Data & Methodology · p. 9
  • The 20 countries were selected by Web of Science publication counts, and comparisons used only article and review records; Dimensions apparently classifies journal reviews as articles.Records were retrieved through Web of Science and Scopus advanced searches and Dimensions API queries, with publication-year filters for 2010–2018.Source: Data & Methodology · pp. 9–10
  • The analysis matched journal lists to identify overlapping and unique journals, and computed country research-output ranks, global shares, and compound annual growth rates using article and review records.Journal-list preprocessing removed duplicate and incomplete entries before matching.Source: Data & Methodology · p. 10
  • Subject-area comparisons mapped Scopus’s 27 and Dimensions’s 22 subject areas to Web of Science’s five major areas, enabling a broad comparison despite differing classification systems.Dimensions uses article-level classification, unlike the source-based classifications of Web of Science and Scopus; the mapping was chosen as a manageable alternative for broad comparison.Source: Data & Methodology · p. 10

Results · Journal coverage of the three databases

Journal coverage differs substantially across Web of Science, Scopus, and Dimensions: Web of Science overlaps heavily with both other databases, while Dimensions has many journals not shared with either. The results also examine country-level research output and subject-area distributions across the databases.

Results

  • The results also analyze research output volume, relative rank, and global share for twenty selected countries, and compare their subject-area distributions across databases.Source: Results · p. 11

Journal coverage of the three databases

Analysing research output of different countries in the three databases

For 20 selected countries, article and review output during 2010–18 varied across Web of Science, Scopus and Dimensions in volume, rank, global share and growth rate. Dimensions generally reported higher output volumes and CAGR, but China and India were exceptions to the volume pattern.

Analysing research output of different countries in the three databases

  • USA’s global share was 27.24% in Web of Science, 23.04% in Scopus and 17.03% in Dimensions, while output volumes also varied by as much as 1 million records.The largest volume differences included USA between Web of Science and Dimensions and China between Web of Science and Scopus.Source: Analysing research output of different… · p. 12
  • India ranked 9th in Web of Science, 6th in Scopus and 7th in Dimensions; Web of Science and Scopus showed greater agreement in country rank and global share than either did with Dimensions.India’s global share was 4.00% in Web of Science, 4.72% in Scopus and 2.74% in Dimensions.Source: Analysing research output of different… · p. 12
  • Dimensions output exceeded Scopus, which exceeded Web of Science, for most countries; China and India were exceptions, with Scopus output significantly higher than Dimensions.The authors suggest possible explanations include Dimensions’ preferential coverage of developed-country journals or differences in countries’ subject-area research strengths.Source: Analysing research output of different… · pp. 12–13
  • Dimensions showed higher CAGR than Web of Science and Scopus for most countries, though rates differed across databases—for example, India’s CAGR was 6.51% in Web of Science and 10.59% in Dimensions.Other examples include USA at 2.63% in Web of Science versus 4.03% in Dimensions, and China at 8.85% in Scopus versus 13.17% in Web of Science.Source: Analysing research output of different… · p. 13

Subject area distribution of research output in the three databases

The three databases differ in the subject-area distribution of research output, both across the twenty countries combined and within individual countries. Dimensions has larger Social Sciences and Arts & Humanities shares, while country-level patterns and database coverage vary by discipline.

Subject area distribution of research output in the three databases

  • 14.4% of Dimensions output is in Social Sciences, versus 8.7% in Scopus and 6.7% in Web of Science; its Arts & Humanities share is also highest.The Arts & Humanities shares are 3.5% for Dimensions, 2.7% for Scopus and 1.3% for Web of Science. All three databases have Life Sciences as their largest area, exceeding 40% of output; Web of Science and Scopus generally have higher Physical Sciences and Technology shares. The authors suggest Dimensions’ broader coverage in Social Sciences and Arts & Humanities could help explain why China and India have lower output volumes in Dimensions.Source: Subject area distribution of research… · p. 13
  • Most countries have 40–50% of publications in Life Sciences, with China, Russia and Taiwan as exceptions.Germany, Japan, France and India have higher Physical Sciences proportions; China, India, South Korea and Taiwan have substantial Technology output. China and India have negligible Arts & Humanities output and very low Social Sciences output across all three databases.Source: Subject area distribution of research… · pp. 13–14
  • Research-output proportions by subject area vary across databases even for the same country, indicating differences in coverage beyond overall journal coverage.The analysis compares the subject-area distributions of all twenty countries individually across Web of Science, Scopus and Dimensions.Source: Subject area distribution of research… · pp. 13–14

Discussion

The databases differ substantially in journal coverage, with Web of Science most selective and Dimensions broadest. Their differing coverage also corresponds to variations in countries’ research outputs and in subject-area representation.

Discussion

  • Using master journal lists updated through June 2020, the study provides an up-to-date comparison of Web of Science and Scopus and the first direct journal-coverage comparison of Dimensions with both.The authors also analyze the subject-area distribution of coverage across the three databases.Source: Discussion · p. 14
  • About 34% of Scopus journals overlap with Web of Science and about 96% with Dimensions; Web of Science is most selective and Dimensions has the broadest coverage.None of the databases completely covers another’s journals; each has unique journals, with smaller unique sets for Web of Science and larger ones for Scopus and Dimensions. Journal overlap need not match article overlap because databases’ coverage has changed over time and journals publish different numbers of articles.Source: Discussion · pp. 14–15
  • The twenty selected countries vary across databases in research-output volume, relative rank, and global share, and some—notably China and India—have higher output in Scopus than Dimensions.These differences indicate that database journal coverage varies across geographies.Source: Discussion · p. 15
  • Dimensions has a higher proportion of Social Sciences and Arts & Humanities output than Web of Science and Scopus, indicating better coverage of journals in those areas.The authors report that Web of Science and Scopus have more or less similar coverage across subject areas.Source: Discussion · pp. 14–16

Conclusion · Limitations

The comparison finds substantial differences in journal coverage among Web of Science, Scopus and Dimensions, with consequences for country-level bibliometric results and disciplinary coverage. The study also notes limitations in journal matching and scope.

Conclusion

  • The study complements article-level database comparisons with an analysis of journals commonly and uniquely covered by Web of Science, Scopus and Dimensions.Source: Conclusion · p. 16
  • Dimensions covers 82.22% more unique journals than Web of Science and 48.17% more than Scopus; Scopus indexes 66.07% more unique journals than Web of Science and Dimensions.Web of Science has the most selective coverage, while Dimensions has the most exhaustive; almost all Web of Science journals are also covered by Scopus and Dimensions.Source: Conclusion · p. 16
  • Different database coverage produces variations in countries’ research-output volume, rank and global share, potentially changing country-level bibliometric evaluations.Source: Conclusion · p. 16
  • Web of Science and Scopus mainly cover Life Sciences, Physical Sciences and Technology, while Dimensions has better Social Sciences and Arts & Humanities coverage.Source: Conclusion · p. 16
  • Database choice can depend on purpose; Dimensions is presented as promising for its exhaustive coverage and filters for expert-curated or nationally recognized journal lists.Source: Conclusion · p. 16

Limitations

  • Restrictive matching may miss journals spelled differently across lists, although manual sample scans suggest the overall impact is likely very small.Source: Limitations · p. 16
  • The study excludes books, conferences and Web of Science’s ESCI, leaving their database coverage comparison as possible future work.Source: Limitations · p. 16

APPENDIX

The appendix details how the three master journal lists were cleaned to remove incomplete, duplicate, and non-journal entries. After preprocessing, Web of Science contained 13,610 entries, Scopus 39,758, and Dimensions 73,966.

Pre-processing step 1

  • Entries missing both ISSN fields or duplicated on both ISSN and e-ISSN were removed, leaving 13,610 Web of Science, 39,851 Scopus, and 74,705 Dimensions entries.Source: APPENDIX · p. 20

Pre-processing step 2

  • Removing entries with repeated ISSN or e-ISSN values left 39,758 Scopus entries and 74,593 Dimensions entries; Web of Science had no such entries.Source: APPENDIX · p. 20

Pre-processing step 3

  • Dimensions entries for preprint sources and conferences were removed, reducing its list to 73,966 journal entries.The cleanup removed 7 preprint-source entries and 617 conference entries.Source: APPENDIX · p. 20

Pre-processing

  • The final preprocessed lists contained 13,610 Web of Science entries, 39,758 Scopus entries, and 73,966 Dimensions entries.Source: APPENDIX · p. 20

Matching:

The matching process first compared ISSN and e-ISSN fields, then used exact and cosine-similarity title matching on remaining records, excluding records matched at earlier steps. Publisher information was also used to validate title matches.

Matching:

  • Matching proceeded from ISSN and e-ISSN comparisons to exact-title and then cosine-similarity matching of remaining records.Records matched in one step were excluded from subsequent matching computations.Source: Matching · pp. 20–21
  • ISSN-field matching identified 12,744 Web of Science–Scopus, 11,305 Web of Science–Dimensions, and 23,579 Scopus–Dimensions records.Source: Matching · p. 21
  • Modified e-ISSN matching and subsequent ISSN-to-modified-e-ISSN matching produced additional matches across the three database pairs.Modified e-ISSN sets yielded 12 Web of Science–Scopus, 1,084 Web of Science–Dimensions, and 8 Scopus–Dimensions matches; remaining ISSN sets matched to modified e-ISSN sets yielded 413, 648, and 43, respectively.Source: Matching · p. 21
  • Further e-ISSN comparisons and cross-matching addressed records with ISSN and e-ISSN values in different fields across database lists.Remaining ISSN sets compared on e-ISSN yielded 164 Web of Science–Scopus, 763 Web of Science–Dimensions, and 12,246 Scopus–Dimensions matches; modified e-ISSN sets compared with remaining ISSN sets yielded 1, 3, and 239. Cross-matching the ISSN and e-ISSN fields yielded 120, 259, and 999 matches, respectively.Source: Matching · p. 21
  • Exact title matching retained 30 Web of Science–Scopus, 144 Web of Science–Dimensions, and 120 Scopus–Dimensions records after publisher checks.The initial title-match counts were 42, 180, and 188, respectively; records with different publisher information were discarded.Source: Matching · p. 22
  • Cosine similarity of 0.9 or higher identified 5 Web of Science–Scopus, 19 Web of Science–Dimensions, and 26 Scopus–Dimensions matches with the same publisher.These came from 5, 22, and 56 inexact title matches, respectively.Source: Matching · p. 22

TABLES

The tables document database coverage, article-type distribution, subject-area mapping, and country-level research-output measures. They specify the scope and organization of these comparisons, but the supplied passages do not provide table values for comparing outcomes.

TABLES

  • Table 1 reports the number of journals and publication records indexed in the three databases.Its note lists SCIE (9,397 journals), SSCI (3,497 journals), and AHCI (1,843 journals), with database lists updated in May or June 2020.Source: Table 1 · p. 23, TABLES · p. 23
  • Table 3 outlines how Scopus and Dimensions subject areas are mapped to Web of Science major subject areas.Source: Table 3 · p. 24
  • Table 4 reports research output rank, volume, and global share for 20 selected countries over 2010–18, including only the document types ‘article’ and ‘review’.Source: Table 4 · p. 24
  • Table 5 presents country-level publication records by database for 20 countries during 2010–18.The accompanying fragment identifies country, database, year, and CAGR fields.Source: Table 5 · p. 25, TABLES · p. 25

FIGURES

The figures show journal-coverage overlap across the three databases and compare research outputs for 20 countries during 2010–18, including distributions by WoS major subject area.

FIGURES

  • Figure 1 depicts pair-wise journal-coverage overlap among the three databases.Source: Figure 1 · p. 27
  • Figure 2 depicts journal-coverage overlap across all three databases.Source: Figure 2 · p. 27
  • Figure 3 compares article-and-review research output for 20 countries across the three databases in 2010–18.Source: Figure 3 · p. 28
  • Figure 4 shows the distribution of the 20 countries’ combined article-and-review output across databases by WoS major subject area.Source: Figure 4 · p. 28
  • Figure 5 shows the distribution of the 20 countries’ individual article-and-review outputs across databases by WoS major subject area.Source: Figure 5 · p. 29

Related papers

Loading 2011.00223v2…