Page 197 - Becoming Metric Wise

187
Journal Citation Analysis

3. Generally speaking review journals have a higher impact factor than
journals which publish few reviews. In fact, these two types of jour-
nals are incomparable. Because reviews tend to receive more citations,
journals tend to increase the number of reviews they publish, see e.g.
(Colebunders et al., 2014) for some medical fields.
4. Databases do not treat all fields, all regions and all languages in the
same way.
5. Databases are not error-free.
6. Impact factors can be manipulated by editors.
a. One manipulation method is forcing authors to cite recent articles
published in their journal (Wilhite & Fong, 2012). These dealings
can be made futile if one removes journal self-citations from the
calculation of a journal’s impact factor. Yet, highly specialized fields
have sometimes just one main journal and removing journal self-
citations from the calculation of the impact factor of these journals
would marginalize them even more. As most authors consider
coercive citation as inappropriate, cf. Chapter 3, Publishing in
Scientific Journals, and prestige reducing (for the editor as well as
for the journal) (Wilhite & Fong, 2012) one might hope that this
practice would eliminate itself.
b. A more subtle form of impact factor manipulation is by using cita-
tion cartels. This means that a group (a cartel) of editors agrees to
preferentially cite each other’s journals. This is most easily done in
editorials, but also coercive citations can be used in this way. As a
prospective author is asked to cite other journals this is less conspic-
uous. An example of such a cartel was revealed by Davis (2012).
7. Impact factors are, essentially, an average number of citations.
However, the distribution of citations to individual publications is
skewed and, hence, the impact factor can sometimes yield misleading
results. An extreme example of this occurred for the journal Acta
Crystallographica A. Whereas this journal typically has a JIF of about 2,
it rose to, respectively, 49.926 and 54.333 in the years 2009 2010,
almost completely due to one single, very highly cited paper.
8. In calculating the JIF Clarivate Analytics does not seem to check if an
item mentioned in a reference list really refers to an existing article:
they just check the name of the journal (is it the journal for which a
JIF is calculated?) and the publication date of the article as mentioned
(is the publication date mentioned one of the 2 years required in the
calculation of the JIF; not if this date is correct) (Vanclay, 2012).