Page 214 - Becoming Metric Wise

205
Indicators

research group; collaboration with colleagues from the same institute, but
at least some do not belong to the same research group; national collabo-
ration: all authors belong to the same country but at least two institutions
are involved; international collaboration: addresses in at least two coun-
tries are mentioned in the byline.
Similarly there are two types of sectorial collaborations, where sectors
are university, health sector (mainly hospitals), companies, other, including
public research centers, nonprofit organizations, and public administration.
A collaboration may involve scientists from the same sector, or from at least
two different sectors.
In gender studies one naturally studies if articles result from authors of
the same gender, and which, or from mixed teams.
Finally we note that besides research collaboration, one may also
distinguish publication collaboration. This happens when different teams
decide, after the fact, to pool their results.

7.2.3 Measures of Collaboration
(Egghe, 1991; Rousseau, 2011)
In this section we provide a review of measures that have been proposed
to measure the collaboration intensity in a set of publications. Consider a
set of N publications, none of which is written anonymously. We assume
that in total A different scientists are author or coauthor of at least one of
these papers. The number f j denotes the number of papers with j (co)-
authors. The index j belongs to the index set I 5 {1,2,.. ..A}. In most
practical cases, but not always, f A will be zero, but it is certain that f j , with
j . A is equal to 0.
We recall the following four collaboration indices, in order of
sophistication.
1. The degree of collaboration (Subramanyam, 1983)

f 1
DC 5 1 2 (7.1)
N
This is nothing but the fraction of coauthored articles.
2. The collaborative index (Lawani, 1980)
A
P
j51 jf j
CI 5 (7.2)
N
This is the average number of authors per publication.