publication must be in a journal or book that is“seen”by the citation
machine. There are basically two systems at the moment: Google Scholar
(GS) and Web of Science (WoS). There seems to be a tendency, in particular
for the humanities and social sciences, to use GS. Harzing (2011: 165) lists
the following advantages of this system:
It is free
It is easy to use
It is quick
It is comprehensive in its coverage.
Access to WoS is costly: my university spends more than€120,000 a year
on this package, and not all researchers in ourfield will be at institutions
that have a subscription. GS is easy to use, without knowledge of underlying
databases or bibliometric instruments. GS may be quick, but for more
common names it will take minutes tofind the citations and then the real work
begins. For example,finding Alan Davies’s h-index meant manually remov-
ing some 650 references. GS is comprehensive, it lists citations to books and
book chapters, conference proceedings, working papers and government
reports and, most importantly, journals not listed in the ISI database, that
forms the basis for WoS, and journals in languages other than English.
Differences between the two systems can be huge. Batia Laufer’s(2003)article
in theCanadian Modern Language Reviewhas a citation score of 28 in WoS and
a score of 212 in GS. Similarly, her (2001)Applied Linguisticsarticle, co-
authored with Jan Hulstijn, has a citation score of 31 in WoS and 706 in GS.
GS and WoS tend to have a different coverage for differentfields of study.
GS displays a better coverage for the social sciences and humanities while
WoS works better for science and medicine (Kousha and Thelwall 2008).
There are, however, also problems with GS. The fact that GS is compre-
hensive also means that irrelevant or incorrect publications and citations are
included. So-called“stray citations”are the most serious problem, but in
WoS the same problem exists as well. Stray citations result from wrong
spellings of names of publications or authors mainly. The comprehensive-
ness also means that publications that are not, or are seldom, cited will be
included. An analysis of my own citations showed that out of 284 papers,
164 were either misspellings of other publications in the list, publications
with less than three citations or other debris such as acknowledgements for
reviewing in journals.^1 While such stray citations have no impact on the h-index,
they do have an impact on indicators such as citations per publication.
For the present study thefindings of Meho and Yang (2007) are relevant.
They show that in particular for GS, coverage of publications before 1990 is
incomplete, but this may have to do with the fact that publishers are still
busy digitalizing older issues of journals and because things have probably
improved since 2007. Still, there may be a bias for more recent publications.
A 2008 comparison of WoS, Scopus, GS and PubMed concluded:
The citation game 107