Page 232 - Becoming Metric Wise

223
Indicators

7.8 SUCCESS INDICES AND SUCCESS MULTIPLIERS
(ROUSSEAU & ROUSSEAU, 2016)

7.8.1 The Success Index
Kosmulski (2011) and Franceschini et al. (2012a) introduced the success
index. This indicator, or actually family of indicators, is constructed as
follows. One considers a set of articles and collects for each of these the
number of citations received over a given citation window W. In a first
step, a binary score (zero or one) is determined for each of these articles:
the score is one if the citations received by a particular article reach a
certain threshold value and it is zero otherwise. This threshold can be
determined in a variety of ways (which is why we say that the success
index is actually a family of indicators). In a next step the success index of
this set of articles with respect to a particular threshold is defined as the
number of publications that has reached the threshold, or stated other-
wise: the sum of all binary scores. Among other proposals, the following
thresholds could be used (Kosmulski, 2011; Franceschini et al., 2012a,b):
• The number of references (each publication’s citations is compared
with its own number of references). This is the original proposal by
Kosmulski (2011).
• The mean or the median number of references in articles published in
the same journal and year as the article under consideration.
• The mean or the median number of citations received by articles pub-
lished in the same journal and year as the article under consideration,
where citations are gathered over the same period W (Franceschini
et al., 2012a,b). Kosmulski referred to this proposal as a modesty
index, as it would reward publication of high-impact articles in lower
impact journals (Kosmulski, 2012).
Clearly, the number of possible thresholds is limitless. One may, for
instance, define a threshold by only considering citations in journals belonging
to the first quartile in one of the JCR categories, or citations received from
authors with a high h-index. Such approaches would operationalize the idea
of “quality citations.” Alternatively, one may consider only recent references.
In the beginning of this chapter we noted that an index can be defined
as a number, derived from a mathematical expression, characterizing a prop-
erty of a dataset. In the case of a success index, the data set is a set of articles
and their citations, and its value is used as a proxy for the “success”—
the term visibility or popularity would be more to the point—of this dataset
and the corresponding entity (scientist, department, journal).