Page 170 - Becoming Metric Wise

162 Becoming Metric-Wise

Concretely: the data in the WoS were augmented by all references used in
the ISSI conference proceedings. Next we provide a scheme to go beyond
the standard synchronous and diachronous impact factor calculations.

6.6.1 A General Framework for Calculating Impact
(Frandsen & Rousseau, 2005): A First Proposal

Calculations of the impact of groups of articles can be made more robust
when allowing for publication and citation windows consisting of more
than 1 year. If the analyzed unit, for instance a research group, only pub-
lishes a small amount of articles a year, then an analysis based on several
years is clearly called for.
We now introduce the following notation:
n p represents the length of the publication window.
n c represents the length of the citation window.
Y p is the first (oldest) year of the publication period.
Y c is the first (oldest) year of the citation period.
Using these notations we denote the general impact factor of a set S
of articles by IF S (n p ,n c ,Y p ,Y c ), and define it by:
n p 21
P P n c 21
i50 k50 CITðY c 1 k; Y p 1 iÞ
IF S n p ; n c ; Y p ; Y c 5 (6.7)
n p 21
P
PUBðY p 1 iÞ
i50
Formula (6.7) is similar to the diachronous impact factor. Sums run
from the first year in the publication period and in the citation period till
the last one (the most-recent one). Yet, it may also represent a synchro-
nous Garfield-Sher impact factor for journal J as follows: IF J (2,1,Y-2,Y).
0
The diachronous JDIF 2 (Y) of journal J is then, in this notation, IF J (1,2,
1
Y,Y). Similarly JDIF n (Y) for journal J is IF J (1,n,Y,Y 1 1). The immediacy
index of journal J becomes: IF J (1,1,Y,Y).
Table 6.5 shows the data used for the calculation of IF S (2,3,Y,Y),
while Table 6.6 shows those needed for the calculation of IF S (3,3,Y,
Y 1 2). These tables are taken from (Frandsen & Rousseau, 2005).
Some of these generalized impact factors were used in Moed et al.
(1985a) and de Bruin et al. (1993). The approach explained here can be
used if all citation and publication data are known till and including the
year Y 1 n c 2 1. An obvious disadvantage of this approach is that publica-
tion years are treated differently, in terms of used citation years. The next
approach tries to remediate this by treating publication years on an equal
footing.