Disparities in document indexation in two databases (Scopus and Web of Science) among six subject domains, and the impact on journal-based metrics

A previous study (10.1007/s11192-020-03457-x) found a discrepancy between Elsevier's CiteScore and Clarivate Analytics' Journal Impact Factor (JIF) in library and information science (LIS) journals. One possibility to explain this discrepancy may lie in the number and type of documents used to calculate these journal-based metrics. Using the top quartile of Scopus-indexed journals from 2011 to 2018, we assessed the number of documents for each journal and year that were indexed in Scopus and in Web of Science (WoS) in six fields of study: LIS, discrete mathematics and combinatorics (DMC), medicine: epidemiology (ME), agriculture and biological sciences (ABS), social science: demography (SSD), and environmental engineering (EE). The number of documents in WoS was higher than those indexed in Scopus for four fields of study: LIS, ME, SSD and EE, with a difference of 1653, 3931, 635 and 197 documents, respectively. For DMC and ABS, Scopus listed more documents than WoS for the same years and journals, the differential being 7 and 1284, respectively. The greater indexing of documents in WoS than in Scopus in four fields of study may explain why the JIF of top-ranking LIS journals is different than their CiteScore. To verify this possibility, one category (DMC) was examined in detail. Of the 16 DMC journals examined, 91.1% were articles, while 8.9% of missing documents were corrections, an erratum, an editorial, an abstract report and in press articles. There were no significant differences between the citation patterns of the missing DMC journals' documents in Scopus and WoS. Citations to missing documents may impact the CiteScore and JIF and should thus be properly indexed.