Page 174 - Becoming Metric Wise

166 Becoming Metric-Wise

more than 11 times less than 31.427, the highest impact factor in the sub-
ject category Neurosciences. For comparison, the latter category also has a
journal with impact factor of 2.833, which is ranked 123rd. This example
illustrates that impact factors cannot be compared across fields, because dif-
ferent fields have different citation habits (number of references, age of
references) and are not always covered as well in international databases.
Hence, if one wants to compare impact factors across fields, normalization
is called for. Normalizing JIFs with respect to a field or domain is done in
two fundamentally different ways: normalization on the cited side or
normalization on the citing side (Moed, 2010; Zitt & Small, 2008). The
normalization process goes through the following steps.
1. One determines an average number of citations per article (in other
words: an impact factor is calculated).
2. One determines the field or domain to which the journal belongs.
This is a problem in itself on which we do not dwell further now.
3. The actual normalization step:
a. Normalization on the cited side. In this approach a journal’s field is
determined by some classification scheme. Then one determines
the average number of citations in the journals’ field or domain and
applies a correction based on this average. The simplest way to do
this is by taking the ratio of the journal’s impact and that of its field.
b. Normalization on the citing side. Here a journal’s field is deter-
mined by the journal itself (see further when we discuss the audi-
ence factor and the source-normalized impact per paper (SNIP)
indicator). One determines the average number of references in
the journal’s field or domain (a kind of citation potential). Then a
correction factor is determined based on this citation potential.
Normalization on the cited side is done in the MOCR (Mean Observed
Citation Rate) (Braun & Gla ¨nzel, 1990), the earlier CWTS crown indicator
(van Raan, 2004a) and the MNCS (Mean Normalized Citation Score) indi-
cator (Waltman et al., 2011a,b). Normalization on the citing side is done
when calculating the audience factor (Zitt & Small, 2008) or the SNIP index,
in its original and revised version (Moed, 2010; Waltman et al., 2013). Most
of these indicators are discussed further on in this chapter or in this book.

6.7.3 Meta-Journal Indicators (Egghe and Rousseau, 1996a,b)
Indicators such as the JIFs presented above are actually indicators for sets
of publications and corresponding citations. So, it is also possible to