Many new visualization techniques have
been developed to overcome this disadvantage (Wasserman & Faust, 1994).
We applied UCINET (Borgatti, Everett,
& Freeman, 2002), an analytical software, to plot the co-citation network.
This method uses normalized multi-dimensional scaling data, but it is less
sensitive to the high multidimensionality of the data (Pilkington & Meredith,
2009; Teichert & Shehu, 2010).
In our analysis we covered the
timeframe from 2001 through 2014.
We chose the year 2001 as our starting
point because the Agile Manifesto was
published in that year. To reveal changes
in the research streams of the agile dis-
course we split the 14-year timeframe
into two time periods of seven years
each (2001–2007 and 2008–2014). We
did not choose shorter sub-periods
because of the time delay between the
initial publication and its citations.
The seven-year range is sufficient to
reveal basic structures in the discourse
within the period, while also allowing for the observation of changes
across two periods (Nerur, Mahapatra,
& Mangalaraj, 2008). The steps we took
for our study are described in Figure 1.
We chose the Social Science Citation
Index (SSCI) as the database for identifying relevant publications because it
contains a very broad set of scientific
journals, books, conference proceedings,
and other scientific publications.
Identify Publications from SSCI
database and Bibliometric Descriptives
A search for keywords in the Social Science Citation Index database was used
in order to identify agile publications.
The keywords are names of specific agile
models, agile project management, and
their acronyms. We used project management as an explicit keyword to identify
those publications that discuss project
management–related topics in the context of agile and vice versa. Figure 2 illustrates the keywords and search logic. 1
“Agile” is not a specified term that
only appears in the software development domain; it is also an important term in the astroparticle physics,
production, supply chain, and some
other fields. Therefore, we manually
excluded those articles that were not
related to software development or
the project management domain. We
also limited the research area to four
categories: “computer science,” “
business economics,” “operational research
management science,” and “information
science.” We only considered academic
articles from journals and conferences
because they better represent the
academic discourse. Other authors,
including Chuang et al. (2014) have
identified the most influential authors
and articles but their database included
both academic and practitioner publications. Practitioner publications typically
contain very few references and some
do not have any reference lists and focus
mainly on practical but not theoretical
issues. Some journals and conference
proceedings publish both scientific and
practitioner articles, such as Communications of the ACM, Agile Conference;
therefore, we paid special attention to
articles from these types of sources.
Figure 1: Steps used in DCA (adapted from McCain, 1990).
First 7-year period
Citation frequency count
Factor analysis and network plot
Second 7-year period
Identify publications by keywords in SSCI database: 2001–2014
Figure 2: Search keywords and logic.
Or Relationship between Keywords
Agile, extreme programming, XP, Scrum,
cockburn crystal, test-driven development,
dynamic software development method, DSDM,
feature driven development, FDD, adaptive
software development, ASD, pair programming,
test driven development, TDD, lean software
AND ( ) OR 1The search team is: TS 5 ((agile OR XP OR ("extreme programming") OR scrum OR ("cockburn crystal") OR dsdm
OR ("dynamic software development method") OR ("feature
driven development") OR fdd OR asd OR ("adaptive software
development") OR ("pair programming") OR ("lean software")
OR tdd OR ("test driven development") OR ("test–driven
development")) AND (“software development” Or “project