NGELES GALLEGO
Departamento de Administracin de Empresas y Marketing, University of Seville, Spain
ABSTRACT
Many studies have examined the relations between individuals and organizations and their influence on innovativeness. Some have looked at networks that improve innovative activity at an organizational, departmental and individual level, using in the latter case the individuals egonet. This
study explores the way in which an individuals social capital and each of its three dimensions
affect innovativeness. Having assessed the entire network of a university department and calculated
the social capital of its members, their innovativeness was compared on the basis of their scientific
production. The results show that the positions of researchers in the network structure and network
quality are less important than the resources that they are able to access through their relations.
Keywords: individual level, innovativeness, network structure, university research, resources, social capital
INTRODUCTION
Volume 12, Issue 1, April 2010 INNOVATION: MANAGEMENT, POLICY & PRACTICE
105
SOCIAL
HYPOTHESES
Previous studies have shown the different ways in
which network structures affect innovation. In the
majority of cases, the structure of a particular relation (personal, social, institutional or businessbased cooperation) is considered the channel for
the satisfactory transmission of new information
and new knowledge on which innovation is based.
The structure of one or various relations is considered a proxy for the patterns of information flows,
or the transmission of information between the
network actors (Rodan & Galunic 2004). This
logic is grounded in the two principal approaches
to the way in which network position affects an
Volume 12, Issue 1, April 2010 INNOVATION: MANAGEMENT, POLICY & PRACTICE
107
METHOD
Unit of analysis
In order to study the influence of individual
social capital on the creation of new knowledge
in innovative organizations, we have opted to
Volume 12, Issue 1, April 2010 INNOVATION: MANAGEMENT, POLICY & PRACTICE
109
Data
Two types of data were collected, so as to obtain
relevant information to construct the intradepartmental network: attributive data that indicates the demographic and academic traits of the
actors (age, gender, year of doctorate, area of
knowledge, employment category i.e. university
lecturer, professor, etc.) and their level of scientific production; and relational data that is used to
110
construct the relational matrices between department members. Six types of ties between actors
were analysed. The first refers to scientific collaboration expressed in tangible results, for which
purpose co-authorships were analysed between
actors. A further type of more general scientific
collaboration that generates relations between scientists is participation in research groups that
have to apply for funding on a competitive basis.
The third relation is a joint presence on the same
university course. Students are divided into
groups to study each subject due to their high
numbers at the university and in the department,
each subject having a maximum of 11 groups.
This generates important teaching needs, for
which reason a course is normally taught by various professors, which creates a social relation
between them. The departmental facilities are
located in various buildings and most lecturers
share offices, normally over lengthy periods of
time. The fourth relation was the tie that arises
from having shared an office. The fifth relation is
that which arises between tutors of doctoral theses and doctorands, as it is shared by most of the
thesis supervisors or the tutors for the department
members. Finally, relations of kinship between
the actors have been analysed. Each relation was
placed in squared matrices (adjacency matrices)
which represented the number of ties that each
actor had maintained with others for each of the
six relations.
A general matrix (network) was selected to
analyse the social network derived from these
relations by calculating the sum of the adjacent
matrices of each of the six relations that had previously been dichotomised. Thus, the matrix that
is representative of the network is a square, with
93 93 elements, and values of 06 for each of
its cells. The cells in which the relation of a
researcher, i, is shown with another, j, have the
value xi,j, which indicates the number of former
relations i has maintained with j over the 15 years
under analysis. Most of the data sets were taken
from secondary information available on the university databases.
which implies collaboration in the same organizational unit under the same supervisor. The first
indicator (Qcoaut) was calculated by establishing
a cut-off point in the co-authorship relation, such
that a quality relation will be seen to exist
between researchers i and j, if they have collaborated on a publication in their research work
three times or more over the 15 years of the
study. The second indicator (Qsubj) was calculated in a similar way for the relation of joint participation on a course, but the cut-off point to
define it as a quality relation was that they had
shared teaching for four academic courses or
more.
Resources: The third dimension that is used to
define the social capital of the researchers refers to
their capacity to access resources within the network that are held by the other members of the
department. From among the most important
resources under consideration, two have been
chosen to represent the third variable of the
model. In the first place, it was thought that
those researchers who hold full professorships
would possess a greater stock of resources than
other departmental members. A full professorship
usually opens the door (and in all cases in the
department under study, although at different
points in time) to important posts in university
management (deaconates, departmental directors,
vice-rectorships...) or supervisory positions in
research teams as lead researchers. This means
that university professors manage a stock of
resources of all types including, as is well known,
those related to research. An indicator Rprof was
prepared that recorded, for each actor, the degree
of contact with other members of the network
holding full professorships. In second place, the
international projection of network members was
taken into account as a resource. Over recent
decades, the centre of attention of the social sciences in Spain and in other neighbouring countries has been undergoing important changes. It
has shifted away from a fundamentally national
reference point towards another centred at an
international level, which is leading to the pro-
Volume 12, Issue 1, April 2010 INNOVATION: MANAGEMENT, POLICY & PRACTICE
111
RESULTS
Structural equation models based on PLS (partial
least squares) are suitable for this investigation, as
the theoretical knowledge is still not fixed, its
purpose is of a causal-predictive nature and a
complex model has to be estimated (Barclay, Higgins & Thompson 1995; Chin, Marcolin &
Newsted 1996).
The presentation of the results follows the two
steps that these techniques require to arrive at
reliable conclusions: firstly, the validity and reliability of the scales of measurement has to be validated and, secondly, conclusions have to be
reached on the relations incorporated in the
model.
The first step in confirming the reliability of
the model is to analyse the indicators that constitute its variables or its constructs. The individual
reliability of each indicator depends on its factorial load in each of the variables considered.
Variables
Loading
Average variance
extracted
0.990
0.963
0.765
0.620
0.961
0.924
0.9633
0.9942
0.9881
0.9783
Qrelation
Qcoaut
Qsubj
0.8162
0.7208
Resource
Recprof
Recint
112
ANALYSIS
Composed reliability
Structure
Degree
Neigh
Closs
Effsize
AVE
0.9572
0.9680
Carmines and Zeller (1979) propose that an indicator is reliable if it has a load of over 0.7, which
is the case for all the indicators used in the measurement models, as may be observed from Table
1 for the model. The next step consists in testing
the reliability of the construct social capital
that is used as a measurement model; by calculating the composed reliability of the construct (c ),
which is a measure of its internal consistency.
Nunnaly (1978) proposes a minimum reliability
that may be accepted when composed reliability
(c ) is 0.7, although he recommends values of
over 0.8 to obtain a stronger level of reliability. In
the structural model, which appears in the second
column of Table 1, all the variables have values of
over 0.9, except in the case of the relational
dimension, which has a value of 0.765, which is
sufficient for exploratory phases of the research,
as occurs in this case. Average variance extracted
(AVE) (Fornell & Larcker 1981) attempts to
measure the convergent validity of the measurement model. It is recommended that the indicators of a construct should explain at least 50% of
their variance, which is the case for the three
dimensions included in the social capital in this
research. Finally, to test the measurement models
goodness of fit, its discriminant validity must be
studied so as to confirm that each variable is really different from the others, for which purpose
two complementary strategies are used. In the
first place, the AVE should be above the shared
variance between the construct and the other
constructs of the model (which is to say, the
square of the correlation between two constructs).
Table 2 shows the correlations between the
dimensions and the square root of the AVE have
been placed in the principal diagonal. It may be
seen that they are in all cases higher than the correlations corresponding to their rows and
columns. In second place, the cross-loadings of
the indicators in each dimension are analysed.
Barclay et al. (1995) suggest that, as the rows are
tested, an indicator should never have a higher
loading in any other construct or dimension
other than that which it is trying to measure and,
VALIDITY OF THE
MEASUREMENT MODEL USING
CORRELATIONS AND AVE
Structure
Qrelation
Resource
Structure
Qrelation
Resource
0.981
0.181
0.727
0.787
0.508
0.961
STR
QREL
RES
.962(**)
.995(**)
.989(**)
.980(**)
.198
.092
.571(**)
.811(**)
.258(*)
.153
.159
.120
.764(**)
.811(**)
.528(**)
.454(**)
.825(**)
.686(**)
.677(**)
.638(**)
.380(**)
.419(**)
.956(**)
.967(**)
Degree
Neigh
Closs
Effsize
Qcoaut
Qsubj
Recprof
Recint
Volume 12, Issue 1, April 2010 INNOVATION: MANAGEMENT, POLICY & PRACTICE
113
Structural
dimension
H1; 1= -0.043
H4; 4= 0.657***
Resource
dimension
H3; 3= 0.568***
Individual
innovativeness
H5; 5= 0.380***
Relational
dimension
H2; 2= 0.222**
SOCIAL CAPITAL
*** p>0.001; ** p>0.01
F IGURE 1. M ODEL
DISCUSSION
The research results have shown that social capital arising from the internal relations of an intraorganizational network (in this case, a university
department) is associated with both the capacity
of individuals in the network to generate new
knowledge and their innovativeness. Based on an
understanding of social capital as a multidimensional construct, evidences were found that the
structural dimension, the relational dimension
and the resource dimension directly or indirectly
affect individual innovativeness. But the most
interesting contribution of this study is that it
114
highlights the role played by the resource dimension, which had not been explicitly mentioned
until a few years ago (Batjargal 2003) in the
dynamic of how an individuals own relations
affect their innovativeness. The capacity of an
individual to access and to mobilise valuable
resources not only directly affects the individuals
innovativeness, but it serves as a catalyser to
obtain benefits from a favourable position in the
intra-organizational network (in the structure of
their relations) and to establish quality relations
with other members of the network. In particular, the structural dimension, according to the
results, does not have a direct influence on individual innovativeness. An actor needs to be connected or to have indirect access to other actors
with resources in order to improve its performance. Previous studies have demonstrated that
contradictory effects exist in some contexts.
Gargiulo and Benassi (2000) pointed out that
dense networks fail to respond well to changes.
Ahuja (2000) stressed, in an analytical context,
that structural holes have a negative influence on
innovation. Bell (2005) analyzed executive relations and confirmed that greater centrality
improves innovation, but was unable to demonstrate as much for institutional relations.
ACKNOWLEDGEMENT
This research was supported with the SEJ200614369 grant of the Ministerio de Ciencia y Tecnologa (Ministry of Science and Tecnology) of
Spain.
References
Adler P and Kwon S (2002) Social capital: Prospects
for a new concept, Academy of Management
Review 27(1): 17-40.
Ahuja G (2000) Collaboration networks, structural
holes, and innovation: a longitudinal study,
Administrative Science Quarterly 45(3): 425-456.
Barclay D, Higgins C and Thompson R (1995) The
partial least squares (PLS) approach to causal
modelling: Personal computer adoption and use as
an illustration, Technology Studies 2(2): 285-309.
Batjargal B (2003) Social capital and entrepreneurial
performance in Russia: a longitudinal study,
Organization Studies 24(4):534-556.
Bell GG (2005) Clusters, networks, and firm innovativeness, Strategic Management Journal 26:
287-295.
Borgatti SP, Everett M and Freeman LC (2002)
Ucinet 6 for Windows: Software for social network
analysis. Analytic Technologies, Natick MA.
Borgatti SP, Jones C and Everett MG (1998)
Network measures of social capital, Connections
21(2): 27-36.
Bourdieu P (1986) The forms of capital, in Richardson JG (Ed.), Handbook of theory and research for
the sociology of education. Greenwood, New York.
Burt RS (1992) Structural holes: The social structure
of competition. Harvard University Press,
Cambridge MA.
Burt RS (1997) The contingent value of social
capital, Administrative Science Quarterly 42(2):
339-365.
Carmines EG and Zeller RA (1979) Reliability and
validity assessment. Sage, Beverly Hills CA.
Carr C, Castilhos M, Davis D, Snyder M and
Stecher B (1982) Costbenefit analysis in
educational evaluation, Studies in Educational
Evaluation 8: 75-85.
Volume 12, Issue 1, April 2010 INNOVATION: MANAGEMENT, POLICY & PRACTICE
115
S P E C I A L I S S U E S N O W AVA I L A B L E
Social Capital in a Rural Context Special Issue of Rural Society
Volume 17 Issue 3 ~ December 2007 ~ iv+140 pages ~ ISBN 978-0-9775742-7-8
Q-Squared in Poverty Analysis in Decision-Making Special Issue of Multiple Research Approaches
Volume 2 Issue 2 ~ October 2008 ~ 160 pages ~ ISBN 978-1-921348-26-6
eContent Management Pty Ltd, PO Box 1027, Maleny QLD 4552, Australia
Tel.: +61-7-5435-2900; Fax. +61-7-5435-2911; subscriptions@e-contentmanagement.com
www.e-contentmanagement.com
Volume 12, Issue 1, April 2010 INNOVATION: MANAGEMENT, POLICY & PRACTICE
117
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.