Content uploaded by Duane P. Truex
Author content
All content in this area was uploaded by Duane P. Truex on May 16, 2014
Content may be subject to copyright.
Association for Information Systems
AIS Electronic Library (AISeL)
ECIS 2011 Proceedings European Conference on Information Systems
(ECIS)
10-6-2011
NETWORKS OF INNOVATION IN IS
RESEARCH: AN EXPLORATION OF THE
RELATIONSHIP BETWEEN CO-
AUTHORSHIP NETWORKS AND H-FAMILY
INDICES
Hirotoshi Takeda
Michael Cuellar
Duane Truex
Richard Vidgen
This material is brought to you by the European Conference on Information Systems (ECIS) at AIS Electronic Library (AISeL). It has been accepted
for inclusion in ECIS 2011 Proceedings by an authorized administrator of AIS Electronic Library (AISeL). For more information, please contact
elibrary@aisnet.org.
Recommended Citation
Takeda, Hirotoshi; Cuellar, Michael; Truex, Duane; and Vidgen, Richard, "NETWORKS OF INNOVATION IN IS RESEARCH:
AN EXPLORATION OF THE RELATIONSHIP BETWEEN CO-AUTHORSHIP NETWORKS AND H-FAMILY INDICES"
(2011). ECIS 2011 Proceedings. Paper 146.
http://aisel.aisnet.org/ecis2011/146
1
NETWORKS OF INNOVATION IN IS RESEARCH: AN
EXPLORATION OF THE RELATIONSHIP BETWEEN CO-
AUTHORSHIP NETWORKS AND H-FAMILY INDICES
Takeda, Hirotoshi, Georgia State University, Robinson College of Business, CIS Department
Room 913, 9th Floor, 35 Broad Street, Atlanta, GA, USA, University of Paris Dauphine,
Paris, France and North Carolina Central University, 1801 Fayetteville Street, Durham,
NC, USA htakeda@cis.gsu.edu
Cuellar, Michael, North Carolina Central University, 1801 Fayetteville Street, Durham, NC,
USA mcuellar@nccu.edu
Truex, Duane, Georgia State University, Robinson College of Business, CIS Department, 35
Broad Street, Atlanta, GA USA, dtruex@gsu.edu and Mittuniversitetet (MID SWEDEN
UNIVERSITY), Division of Information and Communication Systems, Department of
Information Technology and Media, Sundsvall, Sweden
Vidgen, Richard, University of New South Wales, Australian School of Business, Sydney,
Australia, r.vidgen@unsw.edu.au
Abstract
Assessing the work of scholars is of great importance in the life of academic institutions, disciplines
and scholars. Research suggests that that the notion of ‘scholarly influence’ should be substituted for
current approaches towards judging scholarship (Truex et al. 2009). This paper seeks to examine the
nature of the construct ‘scholarly influence’ by reconceptualising the activity of academic research as
a social process of peer production enacted through networks of innovation. It combines techniques
used to assess ‘ideational influence’– a measure of the productivity and the uptake of an author’s
ideas – and techniques used in social network analysis to assess ‘social influence’ – patterns of social
interaction measured as co-authored publications in journals and conferences. The analysis suggests
that social and ideational influence appear to be inter-related; those with high citation indices are
also well connected. Rather than argue causality we have proposed that the two are mutually
reinforcing and that an assessment of researcher impact should take account of both when looking for
indicators that might have predictive power. Given that citations are backward looking it is possible
that measures of social influence, such as closeness to highly ranked scholars as evidenced by co-
authorship networks, will provide a useful forward looking indicator. Promotion boards might
consider social network and citations when considering a researcher in the round.
Keywords: Peer production, open innovation, scholarly influence, social network analysis (SNA),
ideational influence, social influence, H-family indices, bibliometrics.
2
1 Introduction
In both research oriented and teaching oriented institutions, the evaluation of scholarly output has been
of key importance to retention, promotion, and tenure decisions. This raises the question of how do we
evaluate and compare the scholarship and influence of one researcher to another, both in one‟s own
field and between fields of study. Scholar evaluation is a pragmatic concern because deans and
resource awarding bodies consider such comparisons in awarding grants and other resources. The key
question is “what is the best method of doing so?” The received methodology has been to count
publications in “quality journals”, with the appropriate number comprising the cut-off for a
determination of “quality scholarly research.” Voices have been raised critiquing the subjectivity and
„fairness‟ of this method as being a biased and inaccurate representation of scholarly output (Chua et
al. 2002; Singh et al. 2007; Truex et al. 2009; Walstrom et al. 1995).
Following MacDonald and Kam (2007) we argue that all of the extant approaches are insufficient to
the task of assessing and measuring scholarly output. To provide a better method of evaluating
scholarly contribution, the concept of influence has been proposed as an alternative evaluation
mechanism in replacement of “quality”. While any measure or composite measure is still going to
have some built-in bias, we contend that a socially constructed methodology based on co-authorship
and bibliometric measures assesses academic „influence‟ better than an evaluation based on the
opinions of a handful of scholars. Accordingly, the use of several of the Hirsch family of indices, the h
(Hirsch 2005), the hc (Sidiropoulos et al. 2006), and the g indices (Egghe et al. 2006) has been
proposed as a means of creating a profile of measures of influence ((Truex et al. 2009; Truex III et al.
2009)). Using these metrics, a profile of a scholar‟s influence can be assessed and then compared with
other “benchmark” scholars to make a determination of the value of the scholar‟s intellectual
contributions. Such a methodology shifts the focus from the publication venue to the uptake of the
scholar‟s ideas.
Counting publications favours the view of the academic as lone researcher, working independently of
other academics and relying on publication outlet reputation to promote the uptake of ideas. In
practice, much academic research is produced through co-authorships whereby publications are peer-
produced. Benckler and Nissenbaum (2006) argue that peer production is a model of social production
and that it is orthogonal to “contract- and market-based, managerial-firm based and state-based
production” (p. 400). Peer production is typified by two characteristics: decentralization - the authority
to act is at the discretion of the individual rather than a central organizer – and the use of social cues
rather than prices or managerial direction to motivate and coordinate participation in the peer
production network (ibid, p. 400). We further argue that one measure of an academic‟s influence is the
strength of their network, i.e., the social capital they possess in their academic community. By
leveraging their existing network capital (through building new and maintaining existing co-
authorship relationships) the academic researcher has access to further peer-production opportunities.
The current research is a continuation of a line of meta-research in the IS field that is interested in the
nature of the IS field, how IS research is output, and how IS researchers are evaluated. Meta-research
in IS has a long history in the IS field reaching back to the beginnings of the IS field. Mason and
Mitroff (Mason et al. 1973) came up with a IS research agenda back in 1973. The first ICIS
conference had two papers on IS meta research (Hamilton et al. 1980; Keen 1980) and meta-research
continues to this day (Basden 2010). The current research is continuing the evaluation of scholars in IS
meta-research.
All this suggests that to evaluate scholarly influence we need to consider both „ideational influence‟– a
measure of a scholar‟s productivity (measured in research publications), the degree to which others
refer to his work (citations analysis), and the concept of social influence, i.e., the social capital that a
scholar accrues through participation in the peer-production of scholarly outputs. In this paper we
hypothesize that social influence and ideational influence are associated and mutually reinforcing.
3
The rest of the paper proceeds as follows: in the next section we examine the nature of social influence
in scholarly research, then methods of assessing social influence including social network analysis
which we will propose as a means of assessing social influence as expressed in cohort networks in the
IS community. We then describe the methodology used to create an IS scholars social network. Then
we analyze the social network of prominent IS scholars and show how social influence may drive the
ideational influence of scholars. In doing so, we will describe the data collection and analysis methods
used in this study. We then present and discuss the findings and limitations of this research, before
finally, considering the implications and how they may inform future research.
2 Theory Development
2.1 The Concept of Social Influence Defined
The development of scientific knowledge is a social phenomena (Bhaskar 1997; Pinch et al. 1984).
Kuhn (1996) argues that paradigms are established or changed when the community of scientists
determine that the existing paradigm should be replaced. This replacement process, which he terms a
revolution, is a political process enacted within the field. Latour (1987) proposes that scientific
knowledge develops based on a dialogic interaction between those who make claims and those who
support or refute them. Eventually, this builds to a critical mass to where the proposed knowledge is
accepted as “true” or “blackboxed” (Latour, 1987) so that it is no longer contested but accepted and
then not subject to further dispute without additional evidence. Similarly, Pinch and Bijker (1984)
describe an approach to the sociology of knowledge they call the “Empirical Program of Relativism”.
There are three stages to this approach. First, the interpretive flexibility of results; that results are open
to more than one explanation. Second, the interpretive flexibility is limited by social constraints
imposed when a field gains consensus on an issue. Third, the social constraint is arrived at through
social-cultural interaction. From the critical realist position, Archer (1988) provides a methodology for
analyzing cultural social structure change. She argued that cultural social structures are changed or
maintained through the use of social interaction. Bourdieu (1984, 1985) developed a similar concept of
„capital‟ as cultural capital, social capital, and economic capital. In Bourdieu‟s framework social
capital arises from the networks of shared interests and influence (Bourdieu 1984; Bourdieu 1985).
One common aspect to all of these approaches is the concept of social interaction. In interpreting
findings or developing theories, scientists interact with each other to test these theories either formally
through the publication process or informally through interactions at conferences and other meetings.
These interactions mould and shape the ideas of those interacting and eventually help foster the
consensus that determines what the field regards as “truth”. The informal interactions sometimes lead
to formalization of relationships in terms of co-authoring or forming virtual research teams.
Now, in this process of interaction, some scholars are more persuasive than others in terms of
influencing others as to the validity of their ideas. Others are less capable of that type of influence. The
differences in these levels of influence arise through differential social skills, affinity between
scholars, commonality of thought, etc. This ability to influence others through the processes of social
interaction we term “Social Influence.” A scholar may be said to have higher social influence if he/she
is able to change the thought of other scholars through social interactions. Ideational influence on the
other hand is in view when the influence is exercised strictly through their published works.
2.2 The Concept of Ideational Influence Defined
The main way that researchers output their research findings are by publishing. These are results of
research and ideas that are being put out to the intellectual market. While the publication count will
measure the ideational output of a researcher, the impact on the field is not seen by just the publication
count. However, the uptake of the ideational output can be seen by using citation counts. Thus,
“ideational influence” is defined not only by the output of ideas by a researcher but the output of ideas
and the subsequent uptake of the researchers ideas by the field.
4
2.3 Operationalizing Social Influence
Since social interaction takes place in largely informal situations, operationalization would seem to be
difficult. However as this interaction often formalizes into partnerships, we can use these partnerships
to assist in operationalizing the concept of social influence. These partnerships such as doctoral
student-advisor and co-researchers are often difficult to collect data on. Therefore, we suggest that
both relationships will likely result in co-authored citable research artefacts (conference and journal
papers, panels, and edited collections). As advisors take their students through the process of learning
how to conduct research, the advisor teaches the student proper methods and also introduces them to
the field‟s literature and interprets it with him/her. This is a position of great influence. The student
often shows the advisor new streams of literature or performs innovative research that contributes new
knowledge. Thus the student is reciprocally influential to the advisor. Similarly, the relationship
between research partners has significant communication between them exchanging ideas and
interpreting the findings. Accordingly each exerts influence on each other. One of the results of these
processes of interaction is jointly produced co-authored research artefacts reporting their collaboration.
Their joint vested interest in seeing the fruits of their joint research labours in the best possible venue
further cements the relationship. These publications, therefore, represent the result of joint activity
between them and can serve as a proxy for the social influence that occurred between them.
2.4 Social Influence and Ideational Influence
As shown above the workings of social influence and ideational influence are intertwined. Social
influence represents the influence that occurs through the social interactions while ideational influence
represents the uptake of his idea in the field. Social influence is assessed via co-authorship while
ideational influence is assessed via citation analysis. We expect to see that those authors that have high
social influence numbers (via centrality measures) will have high citation analysis numbers (via h-
index). We argue for this on the basis that an author who has high centrality will have grater
opportunity for co-authorship and thus be able to publish more papers and the potential to attract
higher citation numbers.
From the preceding discussion in which we saw that scholars exercised social influence on each other
through the interaction that occurs around a joint research endeavour, there results a knowledge of
each other‟s results and publications. Given this common research interest and knowledge of each
other‟s publication record, we can surmise that this will result in the research partners citing each
other. Given this idea, we propose the following:
Proposition: Social influence and Ideational influence are positively related
2.5 Assessing Social Influence
In evaluating social influence, we need a methodology that allows us to assess scholar relatedness
expressed in co-authorship in such a way to determine influence on the ideas of each other. We
propose using methods commonly used in social network analysis (SNA) to assess these relationships.
In SNA, formal (e.g., such as the co-authoring relationship described above) and informal (e.g., who
you have dinner with when attending conferences) relationships between researchers can be mapped
(Vidgen et al. 2007). One of the results from SNA is the analysis of centrality, or the types and
quantity of connections that one member of the network has to other members of the network. By
examining the centrality measures of the various members of the community, we can arrive at a profile
of measures that assess the social influence of the members of a research community. Proper
comparison of these profiles would allow evaluators to assess the social influence of the scholar and
along with the ideational measures provided by the Hirsch indices create an assessment of the
scholar‟s intellectual contribution. Such an assessment we argue would be superior to that provided by
the simple publication count metric. However, these measures are likely to be related: researchers with
high social influence are likely to have high social capital which will lead to higher levels of citation.
5
Although it might seem unlikely that the reverse is true, i.e., that high levels of citation will lead to
stronger social influence, this may well be the case if promotions and funding are allocated on the
basis of citations and researchers with high citation counts become courted by other researchers.
Network analysis is being used fields from neurobiology to statistical physics. In our literature SNA is
increasingly featured in papers (Nerur et al. 2005; Polites et al. 2008; Takeda 2010), including
examination of social networking sites such as MySpace and Friendster (Howard 2008; Kleinberg
2008). SNA “can in fact uncover subtle, unrecognized relationships between journals, and thus can aid
in the development of more accurate classification schemes in the future” (Polites et al. 2008).
SNA can be used to map networks of co-citations (when one author cites another author‟s work) and
networks of co-authorship (when scholars co-author a research publication). We choose to use co-
authorship as a measure of social influence as it is a more direct and personal linkage between scholars
that requires two-way communication. The strength of the tie between co-authors is indicated by the
number of times that they have written together. In SNA graphical representations typically show
darker and thicker lines to represent higher frequency of co-authorship relationships.
When performing a SNA, several aspects of a network – nodes, edges, connectivity, distance, and
components (or clusters) – are identified and examined. A node is defined as a point on the network
(Barbasi et al. 1999; Coleman 1988; Kleinberg 2000; Travers et al. 1969). In co-authorship networks,
the authors are placed as the nodes. An edge of a network is defined as a line connecting two nodes
(Barbasi et al. 1999; Coleman 1988; Kleinberg 2000; Travers et al. 1969). An edge can be non-
directional, directional, or bidirectional. A co-authorship is represented as a non-directional edge made
between the author/nodes. Distance is the length of the shortest path, measured in links, between two
distinct nodes (Travers et al. 1969). Distance is measured by counting the minimal number of edges it
takes to go from one node to another node. Connectivity is a notion of how a node (an author) in the
network is connected to others via an edge. Depending on the research question, connectivity can be
measured by the pure number of edges coming out of any given node. The researcher may want to
discover how an author is connected so a weight/distance measure to other nodes may be incorporated.
Strength of edges and nodes may also be included in the measure. Connectivity may be shown on the
network by proximity of nodes – how authors form clusters. Proximity measures can show how many
authors are closely related to one author, or how many authors are within a given cluster (Albert et al.
2002; Barbasi et al. 1999; Barbasi et al. 2002; Henry et al. 2007; Vidgen et al. 2007).
In SNA there are three core measures of centrality (Freeman 1977): degree centrality, closeness
centrality, and betweenness centrality. Degree is simply a measure of the number of direct connections
that an author has, i.e., their co-authorships. The more co-authorships an author has then the higher
their degree. If author A co-authors three times with author B, once with C, and twice with D then A‟s
degree is six. In an undirected network, such as co-authorship, the in-degree and the out-degree are the
same. In a directed network, such as friendship, then in-degree and out-degree can be different (e.g., A
likes B but the friendship is not reciprocated). Although degree centrality is a useful measure of
popularity (and possibly promiscuity) it does not say anything much about how powerful an author‟s
position in the network is. However, keeping busy may bring its own rewards and we hypothesize:
H1: A scholar with a higher degree centrality will tend to have a higher ideational influence in
the field.
Betweenness reflects the power of an author‟s network position. An author might have less direct co-
authorship connections than other authors (degree) and not be as close to the centre (closeness) but
occupy a position of power through being a gatekeeper. An author with high betweenness who spans
otherwise unconnected constituencies can exercise control over the flow of information and ideas,
either improving network flow or impeding it. Betweenness is calculated using the geodesic – the
shortest path between two authors. The more times an author is on the geodesic path then the higher
that author‟s betweenness score. In a co-authorship network, betweenness can be thought of as
brokerage, for example a single author might connect a cluster of authors working in IS strategy
research with a separate cluster of researchers interested in business process management research.
6
H2: A scholar with a higher betweenness centrality will tend to have a higher ideational
influence in the field.
Closeness is concerned with how quickly an author can access all the other authors in the network. A
measure of „farness‟ is calculated using the geodesic distance (i.e., the shortest path) between pairs of
authors and closeness is then simply the inverse of the farness score. The author with the highest
closeness score therefore has the shortest paths to all the other authors in the network and this
“closeness” will help them hear information and new ideas quickly and be able to disseminate their
own ideas quickly and efficiently.
H3: A scholar with a higher closeness centrality will tend to have a higher ideational influence
in the field.
2.6 Assessing Ideational Influence
We measure ideational influence by using the Hirsch family of indices. An author‟s Hirsch index h
(h-index) is defined as being the number of papers “h if h of his/her Np papers have at least h citations
each, and the other (Np - h) papers have no more than h citations each” (Hirsch 2005). To increase
one‟s h-index one must continue to publish (productivity), and those publications must garner multiple
citations (influence). Hence the h-family indices assess both productivity and influence (Hirsch 2005).
3 Research Method
To test our hypotheses, we ran a correlation analysis of the social influence against the ideational
influence for a set of the field‟s top ranked scholars. The list was taken from the 100 scholars detailed
in Truex, et al (2009, Table 8) in which they derived the h-family indices for a list of IS scholars
reported in Lowry (2007). Lowry‟s work used MISQ, ISR, and the IS articles of Management Science.
Truex et al. (2009) augmented Lowry‟s original with a list of European scholars derived from EJIS,
ISJ, and JSIS. We do NOT claim that this table is a definitive list of „The‟ top IS scholars, but that it is
a representative list of scholars from both the US and Europe who have high ideational influence, i.e.,
h-index. For the purposes of this research by choosing such a list, we have a list of authors who have a
large number of papers each and thus a larger number of co-authors.
3.1 Measures
We operationalized the concept of social influence by use of three SNA centrality measures: degree,
betweenness, and closeness were used as part of a profile describing the scholar‟s social network. In
this research the social network is defined by the set of authors who have co-authored research
publications in the IS field. The network takes the set of all papers submitted to IS journals and
conferences (including panels and edited collections) in which there is co-authorship of a citable
scholarly production. Sole-authored publications are excluded from the network. SNA‟s main metric is
distance: how many co-author connections does it take to reach another author. For example, if A and
B co-author and B and C co-author and C and D co-author, the distance between A and B is one,
between A and C is two (via B) and between A and D is three (via B and C). To operationalize
ideational influence, we used the Hirsch statistic as described above for each of the scholars.
3.2 Data
The sources of data for the research are shown in Table 1. Data was collected from a range of sources
(conference websites, AIS) as Endnote databases (DBs). The DBs were consolidated into a single file
of citable research artefacts, including conference papers, journal papers, edited collections, and
panels, giving a total of more than 18,000 publications. The DBs were then exported to text and a
program was written to break the records into their constituent parts for loading into a database. A
number of the Endnote records were inaccurate, for example, fields were missing and names mistyped.
7
Closer inspection revealed that an author could be entered into Endnote in different ways, e.g., with a
single initial, with first and middle initials, with variations such as „Bob‟ for „Robert‟ and with various
misspellings. For example, if a researcher‟s name were John Quincy Public, we would find entries
such as John Quincy Public, J. Q. Public, John Q. Public, John Public, J. Quincy Public, J. Public and
John Q. P. For the high-scoring 100 h-index authors we cleaned the data in the database by searching
on parts of their family name and then combining the variants into a single author code.
The h-index was computed using the „Publish or Perish‟ (Harzing 2010) tool by the authors of that
paper. PoP1 is a tool that can be used to measure bibliometric properties of a researcher, including the
h-index. The researcher name needs to be input to the PoP tool. While the tool is very proficient at
finding the bibliometric measures, we did find similar data errors mentioned above during the data
collection for the SNA. These were manually corrected for our research. The h-index measures were
collected in May of 2008.
All the publications in the database that matched the high-scoring 100 authors that had two or more
authors were extracted from the database for input to the social networking analysis software,
UCINET. Not all authors had co-authored with others in the top 100, resulting in 84 authors being
extracted. We then extracted the main component, arriving at a population of 78 authors (six authors
were not connected to the main group of authors), which forms the basis for the subsequent analysis.
Journals
Conferences
CAIS - Communications of the Association
for Information Systems (1999 – 2010)
EJIS - European Journal of Information
Systems (1993 – 2007)
ISJ - Information Systems Journal (1991 –
2010)
ISR - Information Systems Research (1990
– 2009)
JAIS - Journal of the Association for
Information Systems (2000 – 2010)
JITTA - Journal of Information Technology
Theory and Application (1999 – 2010)
JMIS - Journal of Management Information
Systems (1984 – 2009)
JSIS - The Journal of Strategic Information
Systems (1991 – 2009)
MISQ - Management Information Systems
Quarterly (1977 – 2010)
SJIS - Scandinavian Journal of Information
Systems (1989 – 2009)
ACIS – Australian Conference on Information Systems (2001 – 2008)
AISHCI - AIS Transactions on Human-Computer Interaction (2009)
AMCIS – Americas Conference on Information Systems (1998 – 2009)
BLED – Bled Conference on E-Commerce (2001 – 2009)
CONFIRM – International Conference on Information Resources Management
(2008)
ECIS – European Conference on Information Systems (1993 -2009).
GLOBDEV – ICT and Global Development (2008)
ICDSS – International Conference on Decision Support Systems (2007)
ICIS – International Conference on Information Systems (1994 – 2009)
IRWITPM - International Research Workshop on IT Project Management (2006
– 2009)
MCIS - Mediterranean Conference on Information Systems (2007 – 2008)
MWAIS - Midwest AIS Conference (2006 – 2009)
PAJAIS - Pacific Asia Journal of the Association for Information Systems (2009)
PACIS - Pacific Asia Conference on Information Systems (1993 – 2009)
RELCASI - Revista Latinoamericana Y Del Caribe De La Associacion De
Sistemas De Informacion (2008 – 2009)
SIGHCI - Special Interest Group on Human Computer Interaction Conference
(2003 – 2009)
Table 1. Publication sources (alphabetical list)
4 Results
The social network for the 78 scholars is shown in Figure 1. The 78 scholars are shown in Appendix
A, together with their scores on the three measures of centrality: degree, closeness, betweenness, and
their h-index score. Inspection of the table reveals those IS researcher with the highest citations and
social influence.
A clique is a subset of authors in which every author in the clique is connected to every other author in
the clique. Graphically, it is a subset of nodes in which every edge exists between every possible pair
of nodes. In addition a clique cannot be a part of any other larger clique. In Figure 1 the actors
1 The PoP tool may be downloaded from www.harzing.com. The tool can be used by individual researchers to find their h, hc,
and g indices.
8
represented as black symbols form a clique of 13, in which each of the authors has co-authored a
research publication with all 12 of the other authors (Table 2).
BaskervilleRL
DavisGB
HirschheimRA
JarvenpaaSL
KraemerKL
WatsonRT
ZmudRW
DavisFD
GeorgeJF
IvesB
KemererCF
ValacichJS
WhinstonA
Table 2. The central group
A clique of 13 is unusually large. The existence of this clique is evidence of a powerful central group
in the network (Scott 2000). These authors score also the highest on closeness centrality and are
clearly at the very heart of the network.
Figure 1. Social network of the 78 high-scoring authors (produced in Netdraw)
5 Discussion
The Hirsch index is a power function (Egghe et al. 2006) and by implication the g and hc being similar
would also be power functions and hence the variables are, unsurprisingly, heavily left-skewed,
requiring a non-linear transformation (using a log10 function) to create a more nearly normal
distribution. Referring to the table of correlations (Table 3) for the transformed variables we see that
all three measures of centrality are positive and statistically significant supporting our hypotheses H1,
H2, and H3.
The degree of correlation between the h-index and the centrality measures is low-moderate with
betweenness highest at .306. The degree of determination (r2) for betweenness and closeness are .094
and .085 respectively and thus relatively low (Cohen et al. 2003). This would indicate that while social
influence has an influence on ideational influence and/or vice versa, there are many other influences
on them that drive the creation of these types of influence.
9
There is a clear core to the „top 100‟ researchers – thirteen researchers who have each worked with
each of the other twelve researchers in the clique (and in some cases, many times). One might argue
that such an inner core would exert influence over journals (for example, the group is well represented
by senior journal editorships) and to some extent be self-reinforcing. This is not to suggest a cause and
effect relationship or a conspiracy theory, rather it is a virtuous circle of co-authorship and citations
that would be expected of the leading and influential researchers in a field. However, a potential
problem with a clique is that the authorship in the journals of our target list is centred in the clique.
This makes it harder for others to break into the major IS journals and creates a rigid authorship
structure in the IS field.
Correlations
h-index
closeness
betweenness
degree
h-index
Pearson Correlation
1
.291**
.306**
.242*
Sig. (2-tailed)
.010
.007
.033
N
78
78
78
78
closeness
Pearson Correlation
.291**
1
.749**
.749**
Sig. (2-tailed)
.010
.000
.000
N
78
78
78
78
betweenness
Pearson Correlation
.306**
.749**
1
.659**
Sig. (2-tailed)
.007
.000
.000
N
78
78
78
78
degree
Pearson Correlation
.242*
.749**
.659**
1
Sig. (2-tailed)
.033
.000
.000
N
78
78
78
78
**. Correlation is significant at the 0.01 level (2-tailed).
*. Correlation is significant at the 0.05 level (2-tailed).
Table 3. Correlation Matrix
We have proposed that researcher impact be assessed using social and ideational influence. For
example, a promotion committee interested in researchers having national and international networks
might consider both citation counts and co-authorship networks. It seems likely that those researchers
with a strong social network (measured objectively using SNA) will likely be more successful in the
future: ideas do not circulate in a vacuum – they circulate in social networks. Clearly, there are more
networks at play than simply co-authorship but this is one form of social network that all researchers,
to a greater or lesser extent, are part of. Further, we would expect there to be a complex relationship
between social influence and citation index. Researchers who are well connected in the co-authorship
network will gain citations through their associations with other central players, while researchers with
high citation indices may be attractive to other researchers when they are considering who to forge co-
authorship links with. The combination of social and ideational influence may well be self-reinforcing
and institutions would do well to consider both in making appointments and promotions. It is therefore
not surprising that the link between citation and centrality has been demonstrated in a correlation that
has a medium size effect (.306) with high significance (p<0.01).
5.1 Contributions
We have positioned academic research as the peer-production of knowledge through networks of open
innovation. Thus, the position a researcher occupies in their academic community is an indicator of
10
their ability to be involved in influential research projects (their social influence). We further argue
that the success of an academic‟s scholarly output should be measured by the uptake work of that
work by others (their ideational influence) as measured by the H-family of indices.
We see eight contributions of this work. First, this paper is the first to define the concept of social
influence and propose it as a means of assessing intellectual contribution to the field. In doing so, it,
second, attempts to understand the connection between social influence (using SNA) and ideational
influence (using the h-index). Third, to the authors‟ knowledge, this is the first attempt at creating the
SNA using such a large number of IS venues, of the core top 100 IS researchers. Fourth, the research
results show that there exists a „circle of co-authorship and citations‟ in the top IS researchers. Fifth, it
continues a line of research trying to better understand and develop measures for the construct
„Scholarly Influence‟. Sixth, the topic is relevant to others outside our own field so if we make inroads
to the understanding of the relationship between the social and ideational measures we have reason to
anticipate that the work will have broad applicability beyond the field of IS research. Seventh, in
confirming the structural relationship between social connections and ideational influence it is the only
work we know that provides empirical and theoretical grounding for why a social network matters.
Finally eighth, following Davis (1971), where a Sociology of the Interesting is juxtaposed with a
Theory of Knowledge to create a typology of twelve sets of propositions of „interestingness‟, our work
challenges the notion that scholarly influence is a monolithic construct primarily associated with the
generation of publications in a very limited set of premier journals and that the interaction of social
and ideational influence is multifaceted and complex.
5.2 Limitations
We see four limitations in this exploratory research. First, our current research, being an initial attempt
at the SNA of IS scholars, took a small slice of the IS researcher pie. We started with a hybrid top 100
IS researcher list, which is incomplete and can be argued as not representing the IS field. Second, the
SNA was created using publications that were limited to those in table 1. While table 1 includes many
top IS research venues, one can argue that using this list does not take into account lesser known
venues. Third, the research only incorporates three SNA measures. Finally, the ideational influence
was measured using the h-index. Whenever a bibliometric measure is used citations are seen in the
lens of the measure used, including inherent biases that are part of the measure. While quantifying of
author citations is necessary for making comparisons in this research, this is only one view and it is
shaped by through the lens of the bibliometric measure used. While these four measures were used as
surrogates for measuring social and ideational influence, there are many other SNA and bibliometrics
that could be used. It is possible that further insights may be attained using other SNA measures and
bibliometrics.
5.3 Future Research
While we are satisfied with the size of our database for creation of co-authorship data for the SNA, we
realize that our work still needs to integrate many other outlets of research. We are continually adding
more IS venues as they become available. With the addition of other venues, we hope to create a better
picture of the IS social network. The sole dependency on research artefact co-authorship for creating
an SNA can be expanded. Other forms of relationships such as research groups, institutional
affiliation, and advisor-advisee relationships can be used for creating an SNA. Having this database of
authorship now affords an opportunity to add author characteristics including place of PhD investiture,
PhD advisor, faculty memberships, gender, and so on, allowing for the generation of many different
kinds of authorship networks. With the addition of abstracts and keywords and using latent semantic
analysis techniques we can explore the evolution of topics and research cohorts over an author‟s
publishing career. These various types of social influence networks can also be compared to h-family
indices and other „ideational‟ influence measures.
11
6 Conclusion
This research suggests that social and ideational influence appear to be inter-related; those with high
citation indices are also well-connected. Rather than argue that one causes the other we have proposed
that the two are mutually reinforcing and that an assessment of researcher impact should take account
of both when looking for indicators that might have predictive power. Promotions boards might
consider social network position and citations jointly when considering a researcher in the round.
Given that citations are backward looking it is possible that measures of social influence, as evidenced
by co-authorship networks, will provide a useful forward-looking indicator.
References
Albert, R., and Barabasi, A.-L. "Statistical Mechanics of Complex Networks," Reviews of Modern Physics (74),
January 2002.
Andreev, P., Feller, J., Finnegan, P., and Moretz, J. "Conceptualizing the Commons-based Peer Production of
Software: an activity theoretic analysis," International Conference on Information Systems, St. Louis, MO,
2010.
Archer, M.S. Culture and Agency, The Place of Culture in Social Theory Cambridge University Press,
Cambridge, 1988.
Barbasi, A.-L., and Albert, R. "Emergence of Scaling in Random Networks," Science (286:5439), October 15
1999, pp 509-512.
Barbasi, A.L., Jeong, H., Neda, Z., Ravasz, E., Schubert, A., and Vicsek, T. "Evolution of the social network of
scientific collaborations," Physics A. (311) 2002, pp 590-614.
Basden, A. "On using spheres of meaning to define and dignify the IS discipline," International Journal of
Information Management (30:1) 2010, pp pp 13-20.
Benckler, Y., and Nissenbaum, H. "Commons-based Peer Production and Virtue," The Journal of Political
Philosophy (14:4) 2006, pp 394-419.
Bhaskar, R. A Realist Theory of Science, (2nd ed.) Verso, London, 1997.
Bourdieu, P. Homo Academicus, (1st ed.) Stanford University Press, 1984.
Bourdieu, P. "The Forms of Capital," in: Handbook of Theory and Research for the Sociology of
Education, J. Richardson (ed.), Greenwood Press, NY, 1985, pp. 241-258.
Chesbrough, H. Open innovation: the new imperative for creating and profiting from technology Harvard
Business School Press, Boston, MA, 2003.
Chua, C., Cao, L., Cousins, K., and Straub, D. "Measuring Researcher-Production in Information
Systems," Journal of the Association of Information Systems (3) 2002, pp 145-215.
Cohen, P., Cohen, J., West, S.G., and Aiken, L.S. Applied Multiple Regression/Correlation Analysis for the
Behavior Sciences, (3rd ed.) Lawrence Erlbaum Associates, Publishers, Mahwah, New Jersey, 2003.
Coleman, J.S. "Social Capital in the Creation of Human Capital," The American Journal of Sociology (94) 1988,
pp S95-S120.
Davis, M.S. "That's interesting: Towards a phenomenology of sociology and a sociology of phenomenology,"
Philosophy of the Social Sciences (1:4) 1971, pp 309-344.
Egghe, L., and Rousseau, R. "An informetric model for the Hirsch-index," Scientometrics (69:1) 2006, pp 121-
129.
Freeman, L.C. "A Set of Measures of Centrality Based on Betweenness," Sociometry (40:1) 1977, pp 35-41.
Hamilton, S., and Ives, B. "Communications of MIS research: An analysis of journal stratifications,"
Proceedings of the First International Conference on Information Systems Association for Information
Systems, Philadelphia, Pennsylvania, 1980.
Harzing, A.-W. "Publish or Perish," 2010.
Henry, N., Goodell, H., Elmqvist, N., and Fekete, J.-D. "20 Years of Four HCI Conferences: A Visual
Exploration," International Journal of Human-Computer Interaction (23:3) 2007, pp 239-285.
Hirsch, J.E. "An index to quantify an individual's scientific research output," Proceedings of the
National Academy of Sciences of the United States of America (102:46) 2005, pp 16569-16572.
12
Howard, B. "Analyzing Online Social Networks," Communications of the ACM (51:11), November 2008, pp
14-16.
Keen, P.G.W. "MIS Research: Reference Disciplines and a Cumulative Tradition," Proceedings of the First
International Conference on Information Systems, Association for Information Systems, Philadelphia, PA,
1980, pp. pp. 9-18.
Kleinberg, J. "The Small-World Phenomenon: An Algorithmic Perspective," Annual ACM Symposium on
Theory of Computing, 2000, pp. 163-170.
Kleinberg, J. "The Convergence of Social and Technological Networks," Communications of the ACM (51:11),
November 2008.
Kuhn, T.S. The Structure of Scientific Revolutions, (3rd ed.) The University of Chicago Press, Chicago, 1996.
Latour, B. Science in Action - How to follow scientist and engineers through society Harvard University Press,
Cambridge, MA, 1987.
Lowry, P.B., Karuga, G.G., and Richardson, V.J. "Assessing Leading Institutions, Faculty, And
Articles in Premier Information Systems Research Journals," Communications of the Association
for Information Systems (20) 2007, pp 142-203.
MacDonald, S., and Kam, J. "Ring a Ring o' Roses: Quality Journals and Gamesmanship in
Management Studies," Journal of Management Studies (44:4), June 2007, pp 640-655.
Mason, R.O., and Mitroff, I.I. "A Program for Research on Management Information Systems," Management
Science (19:6) 1973, pp pp 475-487.
Nerur, S., Sikora, R., Mangalaraj, G., and Balijepally, V. "Assessing the Relative Influence of Journals
in a Citation Network," Communications of the ACM (48:11) 2005, pp 71-74.
Pinch, T.J., and Bijker, W.E. "The Social Construction of Facts and Artegacts: or How the Sociology of Science
and the Sociology of Technology might Benefit Each Other," Social Studies of Science) 1984, pp pp 399-
441.
Polites, G.L., and Watson, R.T. "The Centrality and Prestige of CACM," Communications of the ACM (51:1),
January 2008, pp 95-100.
Scott, J. Social Network Analysis Sage Publications, London, 2000.
Sidiropoulos, A., Katsaros, D., and Manolopoulos, Y. "Generalized h-index for Disclosing Latent
Facts in Citation Networks," arXiv:cs.DL/o606066 (1), 07/13/2006 2006.
Singh, G., Haddad, K.M., and Chow, C.W. "Are articles in “top” management journals necessarily of higher
quality?," Journal of Management Inquiry (16:4) 2007, pp 319-331.
Takeda, H. "A Social Network Analysis of the IS Field: A Co-Authorship Network Study," Southern AIS, AIS,
Atlanta, 2010.
Travers, J., and Milgram, S. "An Experimental Study of the Small World Problem," Sociometry (32:4),
December 1969, pp 425-443.
Truex, D., Cuellar, M., and Takeda, H. "Assessing Scholarly Influence: Using the Hirsch Indices to Reframe the
Discourse," Journal of the Association for Information Systems (10:7) 2009.
Truex III, D.P., Cuellar, M.J., and Takeda, H. "Assessing Scholarly Influence: Using the Hirsch
Indices to Reframe the Discourse," Journal of the Association of Information Systems (10:7) 2009,
pp 560-594.
Vidgen, R., Henneberg, S., and Naude, P. "What sort of community is the European Conference on Information
Systems? A social network analysis 1993-2005," European Journal of Information Systems (16) 2007, pp
5-19.
Walstrom, K.A., Hardgrave, B.C., and Wilson, R.L. "Forums for Management Information Systems Scholars,"
Communications of the ACM (38:3) 1995, pp 93-107.
13
Appendix A: The 78 Scholars analyzed using Hirsch index and Centrality Measures
author
h-
index
author
degree
author
closeness
author
betweenness
WhinstonA
42
GalliersRD
44
DavisGB
65.15
DavisGB
24.54
BenbasatI
41
JarvenpaaSL
42
WatsonRT
63.53
WatsonRT
20.34
BrynjolfssonE
40
ValacichJS
40
BaskervilleRL
59.63
SaundersC
11.32
GroverV
40
WatsonRT
39
IvesB
58.87
IvesB
9.47
BankerRD
38
DavisGB
36
HirschheimRA
58.66
BaskervilleRL
9.24
NunamakerJF
37
LyytinenK
31
JarvenpaaSL
58.66
LyytinenK
8.11
HirschheimRA
36
WhinstonA
30
ZmudRW
57.79
ValacichJS
7.97
OrlikowskiWJ
36
NewellS
28
ValacichJS
57.03
ZmudRW
7.42
HuberGP
35
HirschheimRA
27
DavisFD
56.39
WhinstonA
7.00
JarvenpaaSL
35
ZmudRW
27
WhinstonA
56.17
GeorgeJF
6.37
KraemerKL
35
BaskervilleRL
26
KraemerKL
55.84
JarvenpaaSL
5.90
RobeyD
35
IvesB
26
GeorgeJF
55.74
HirschheimRA
5.83
StraubD
35
JiangJJ
24
KemererCF
55.52
GalliersRD
4.66
ZmudRW
34
SambamurthyV
23
LyytinenK
55.30
DavisFD
4.54
WillcocksLP
33
KleinG
22
GalliersRD
53.68
RobeyD
4.51
DennisAR
32
DavisFD
20
SaundersC
49.24
GroverV
4.19
IgbariaM
32
GeorgeJF
20
SambamurthyV
47.51
OKeefeRM
3.56
KingWR
32
DennisAR
19
RobeyD
46.75
KemererCF
3.50
IvesB
30
SaundersC
18
BenbasatI
45.67
MukhopadhyayT
3.40
LyytinenK
30
NunamakerJF
17
KleinHK
45.67
DennisAR
3.31
ValacichJS
29
KemererCF
16
WillcocksLP
45.67
WalshamG
2.81
WatsonRT
29
KleinHK
16
LandF
44.81
BenbasatI
2.67
CiborraCU
28
LeidnerDE
16
LeidnerDE
44.81
BrynjolfssonE
2.64
PooleMS
28
SwanJA
16
MyersMD
44.81
GoodhueDL
2.60
KemererCF
27
BaruaA
15
MukhopadhyayT
44.74
JiangJJ
2.60
AlaviM
26
GroverV
15
MathiassenL
43.29
WebsterJ
2.60
ChenC
26
KraemerKL
15
PittLF
42.47
SambamurthyV
2.25
DavisGB
26
MyersMD
14
EarlMJ
42.29
KettingerWJ
1.50
KauffmanRJ
26
GefenD
13
WeiKK
41.56
BostromRP
1.24
VogelDR
26
StraubD
13
StraubD
41.45
KraemerKL
1.22
GalliersRD
25
BenbasatI
12
GroverV
41.21
StraubD
1.18
GefenD
25
MathiassenL
11
VenkateshV
41.10
GiaglisGM
1.16
HittLM
25
AgarwalR
10
WigandR
41.10
MathiassenL
1.01
NorthcraftGB
25
BrynjolfssonE
10
LeeAS
41.06
NunamakerJF
0.75
AgarwalR
24
RobeyD
10
BostromRP
40.63
WillcocksLP
0.67
BostromRP
24
AlaviM
9
AgarwalR
40.41
MyersMD
0.51
DavisFD
24
KauffmanRJ
9
WalshamG
40.30
LeidnerDE
0.41
MukhopadhyayT
24
KettingerWJ
8
NewellS
40.13
WeiKK
0.32
ThompsonSH
24
PittLF
8
BrynjolfssonE
39.85
KauffmanRJ
0.29
WalshamG
24
VogelDR
8
GuimaraesT
39.81
KleinHK
0.21
WatsonHJ
24
WillcocksLP
8
DennisAR
39.74
VogelDR
0.17
VenkateshV
23
MukhopadhyayT
7
PooleMS
39.61
AgarwalR
0.17
KeilM
22
WeiKK
7
GoodhueDL
39.37
VenkateshV
0.16
14
author
h-
index
author
degree
author
closeness
author
betweenness
MathiassenL
22
WigandR
7
IgbariaM
39.16
NewellS
0.15
PaulRJ
22
BankerRD
6
WetherbeJC
39.16
PaulRJ
0.13
SambamurthyV
22
GuimaraesT
6
NunamakerJF
39.09
LeeAS
0.13
SwanJA
22
LandF
6
WebsterJ
39.00
SwansonEB
0.10
WebsterJ
22
VenkateshV
6
KauffmanRJ
38.87
CiborraCU
0.08
BaruaA
21
EarlMJ
5
AlaviM
38.83
LandF
0.05
ConnollyT
21
HittLM
5
KingWR
38.51
PittLF
0.04
GuimaraesT
21
IgbariaM
5
SwansonEB
38.51
AlaviM
0.04
KettingerWJ
21
KeilM
5
BankerRD
38.12
BaruaA
0.03
MingersJC
21
LeeAS
5
OKeefeRM
37.77
BankerRD
0.02
WeillP
21
BostromRP
4
CiborraCU
37.75
PooleMS
0.02
WetherbeJC
21
WalshamG
4
KeilM
37.58
SwanJA
0.02
BaskervilleRL
20
CiborraCU
2
ConnollyT
36.71
GuimaraesT
0.02
KekreS
20
ConnollyT
2
SwanJA
36.54
AlterS
0.00
OKeefeRM
20
GiaglisGM
2
BaruaA
36.49
ChenC
0.00
RamamurthyK
20
GoodhueDL
2
HuberGP
36.23
ConnollyT
0.00
SwansonEB
20
KingWR
2
NorthcraftGB
36.06
EarlMJ
0.00
WeiKK
20
NorthcraftGB
2
WeillP
35.91
GefenD
0.00
EarlMJ
19
OKeefeRM
2
ThompsonSH
35.30
HittLM
0.00
GeorgeJF
19
PaulRJ
2
JiangJJ
33.90
HuberGP
0.00
GoodhueDL
19
PooleMS
2
KleinG
33.25
IgbariaM
0.00
KleinG
19
SwansonEB
2
GefenD
32.79
KeilM
0.00
LeidnerDE
19
WebsterJ
2
RamamurthyK
32.38
KekreS
0.00
MartocchioJ
19
WetherbeJC
2
VogelDR
32.23
KingWR
0.00
MyersMD
19
AlterS
1
OrlikowskiWJ
31.41
KleinG
0.00
NewellS
19
ChenC
1
KekreS
30.74
MartocchioJ
0.00
AlterS
18
HuberGP
1
KettingerWJ
30.15
MingersJC
0.00
GiaglisGM
18
KekreS
1
GiaglisGM
28.83
NorthcraftGB
0.00
JiangJJ
18
MartocchioJ
1
MingersJC
28.57
OrlikowskiWJ
0.00
KleinHK
18
MingersJC
1
WatsonHJ
28.29
RamamurthyK
0.00
LeeAS
18
OrlikowskiWJ
1
MartocchioJ
28.16
ThompsonSH
0.00
PittLF
18
RamamurthyK
1
HittLM
28.05
WatsonHJ
0.00
SaundersC
18
ThompsonSH
1
AlterS
27.99
WeillP
0.00
LandF
17
WatsonHJ
1
PaulRJ
25.30
WetherbeJC
0.00
WigandR
17
WeillP
1
ChenC
25.19
WigandR
0.00
