Call for caution in the use of bibliometric data

Bibliometric information was extracted on the week of December 7 2015 for US institutions from Scopus and the WoS for a 3-year period. The 'affiliation search' option was used for Scopus, while the 'organization-enhanced' (code OG) was used for the WoS. For each database we established the difference in the number of publications for each university using both approaches and expressed it in terms of percentage. All types of documents published were considered for this analysis. data retrieved using Scopus showed noteworthy differences for the 20 institutions, ranging up to 6.7%. A closer look at the data from these institutions revealed minor problems with misspelled addresses, although most of the cases did not show any problem that could be highlighted. Conversely, it is worth noting that the WoS showed significantly lower differences between both search strategies. Only Cornell University and the Massachusetts Institute of Technology (MIT) showed differences during the 3 years assessed, reaching up to 1.2% during 2014. In the case of MIT, the 1% difference corresponded to documents that were authored by researchers that did not register MIT as their affiliation, even though MIT appears in all cases as the 'Organization-Enhanced Name.' At the end of the day, the better knowledge we have about the method by which data are collected, the smaller the risk of misuse or misinterpretation.