Page 185 - Becoming Metric Wise

177
Journal Citation Analysis

selected publication. Then p j is the proportion of these publications that
have at least one active reference. The underlying idea, according to
Waltman et al. (2013) for using these new variables is that without
including p j , the source normalization mechanism may fail to completely
correct for differences between low and high citation density fields.
We further note the following points related to the DCP calculation.
1. Only peer-reviewed publications of the type article, conference paper
and review are considered (as was the case for the original SNIP).
2. The subject field of a journal consists of all publications in the year of
analysis that refer to a publication in the journal in the three preceding
years (not 8 anymore). Moreover, duplicate publications are now
allowed (they were not).
3. The multiplication by 1/3 ensures that the average SNIP-value for all
journals in the database is close to one.
The authors make the following comments on the selection of citing
journals:
• journals that have changed titles are considered as one (this is typically
not the case in the WoS).
• trade journals are excluded.
• journals that did not publish continuously during four consecutive
years are excluded.
• journals with less than 20% of the publications in the year of analysis
having at least one active reference are excluded.
We now have another look at the formulae, beginning with SNIP 5 RIP .
DCP
RIP 5 k/m, where k is the number of received citations and m is the
number of publications during the used citation and publication windows.
We already know that:
1 n
DCP 5 : (6.18)
3 1 1 1 1 ::: 1 1
p 1 r 1 p 2 r 2 p n r n
Combining these two formula yields:
n
3 X 1
SNIP 5 (6.19)
m p j r j
j51
Clearly SNIP decreases when any r (a number of references) increases.
This formula can be understood as follows: the revised SNIP is a jour-
nal’s average number of citations per publication (division by m), where each
citation is weighted inversely proportional to both the number of active
references in the citing publication, and the proportion of publications with