Sunteți pe pagina 1din 22

Strategic Management Journal

Strat. Mgmt. J., 29: 251272 (2008)

Published online 31 October 2007 in Wiley InterScience ( DOI: 10.1002/smj.666
Received 19 April 2002; Final revision received 3 October 2007


Iowa State University, Ames, Iowa, U.S.A.

This study examines the underexplored tensions and complementarities between bridging ties
and strong ties in innovation-seeking alliances. Bridging ties span structural holes to provide
innovation potential but lack integration capacity, and strong ties provide integration capacity
but lack innovation potential. We theoretically develop the idea thatnotwithstanding their
tensionsstrong ties complement bridging ties in enhancing alliance ambidexterity at the project
level. While bridging ties provide access to diverse, structural hole-spanning perspectives and
capabilities, strong ties help integrate them to realize an innovation. We also propose that their
effects and complementarities influence alliance ambidexterity because they facilitate knowledge
integration at the project level. Tests using data on 42 innovation-seeking project alliances
involving a major American services conglomerate and its alliance partners support the majority
of the proposed ideas. Implications for interfirm network configuration, strategic alliances, and
the broader strategy literature are also discussed. Copyright 2007 John Wiley & Sons, Ltd.

INTRODUCTION other hand, collaboratively exploiting that exper-

tise at the project level requires strong ties. The
While alliances create the potential for innova- greater the extent of bridging ties in an alliance, the
tion by providing access to a diverse variety of greater is both the diversity of accessible knowl-
capabilities and expertise (Stuart, 1998), they must edge, capabilities, and perspectives and the diffi-
be integrated at the project level to realize that culty of integrating them. In contrast, the presence
potential. This study examines a relatively under- of strong ties in such alliances eases knowledge
explored tension in the literature between bridging integration but lowers the likelihood of innovation
ties and strong ties in innovation-seeking project because strong ties are associated with redundant
alliances. The tension manifests itself in the fol- knowledge, perspectives, and capabilities. Obstfeld
lowing manner. On the one hand, the successful (2005) characterizes this as the tension between
accomplishment of novel projects requires the het- the idea problem versus the action problem.
erogeneity of capabilities and expertise provided A network of collaborators with strong ties has
by bridging ties among alliance partners, but on the greater capacity to implement innovative ideas, but
has inherently lower capacity to generate them; a
network that is rich in structural holes (i.e., greater
Keywords: alliance ambidexterity; structural holes; bridging ties) has greater capacity to generate new
knowledge integration; combinatorial innovation; Burt ideas, but has a lower capacity to implement them.
complements; network configuration The tradeoff, therefore, is that the potential for
*Correspondence to: Amrit Tiwana, College of Business, Iowa
State University, 2340 Gerdin Business Building, Ames, IA novelty is lost by strong ties, and the potential for
50011-1350, U.S.A. E-mail: integrating novel knowledge is lost by bridging

Copyright 2007 John Wiley & Sons, Ltd.

252 A. Tiwana

ties. An alliance that is high on both bridging (Darr, Argote, and Epple, 1995; Simonin, 1999),
ties and strong ties might then provide what Burt but without directly considering tie characteristics.
(1992) would describe as the ideal configuration. More recent work has shed light on the influ-
In other words, strong ties should complement ence of network characteristics on knowledge pro-
bridging ties. The objective of this paper is to cesses (Hansen, 1999; Levin and Cross, 2004), but
theoretically develop and test these underexplored with an implicit assertion about its effect on per-
ideas. The theoretical significance of these ideas formance. Furthermore, while the importance of
stems from three notable gaps in the strategy liter- knowledge integration is widely recognized in the
ature. social networks literature (Obstfeld, 2005) as well
A review of the past three decades (19762006) as the broader strategy literature (Grant, 1996a;
of research on social network configurations and Kogut and Zander, 1992), the majority of stud-
strategic alliances reveals three overarching gaps: ies have actually focused on knowledge transfer
(a) the inattention to tie characteristics, knowl- or knowledge acquisition. Knowledge integration,
edge processes, or performance, (b) the lack of however, is conceptually distinct and inherently
direct attention to the complementarities between more challenging than knowledge transfer (Carlile,
strong ties and structural holes, and (c) the pre- 2002). The implicit mediating knowledge integra-
dominant use of dyads or alliances rather than tion construct that theoretically links tie character-
projects (where much innovation work within mul- istics with alliance performance has therefore not
tifirm strategic alliances is actually accomplished) directly been studied.
as the unit of analysis. Second, recent work has started to recognize
First, studies in the social networks literature the tensions and complementarities between dif-
have usually examined two of the three sets of vari- ferent forms of social capital as well as differ-
ablestie characteristics, knowledge processes, ent types of ties, but such tensions and comple-
and performanceand treated the third, unmea- mentarities have not directly been studied. For
sured set as implicit (see Figure 1 for an illustra- example, Regans, Zukerman, and McEvily (2004)
tion). The entire nomology from social network emphasized the tradeoffs between network cohe-
characteristics to knowledge processes to perfor- sion and range in promoting creativity on the
mance has received scarce direct attention. For one hand, and cooperation and coordination on
example, prior work has examined how network the other. The tension between strong and bridg-
ties influence performance (Ingram and Roberts, ing ties is also implicit in the two different con-
2000; Regans and McEvily, 2003; Tsai, 2001), but ceptualizations of social capital theory (Obstfeld,
the intervening knowledge processes are treated 2005). One view subscribes to the benefits of
as a black box. Other work has provided insights dense social networks based on the premise that
into how knowledge processes, such as knowledge they facilitate intensive coordination and realiza-
transfer and acquisition, influence performance tion of collective ideas (Coleman, 1988). The other

Levin and Cross (2004)

Hansen (1999)
Yli-Renko et al. (2001)

Structure/ Tie
Ingram and Roberts (2000) Simonin (1999)
Regans and McEvily (2003) Darr et al. (1995)
Tsai (2001) Performance

Figure 1. Representative studies and gaps at the intersection of network structure configuration, knowledge processes,
and performance
Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
Do Bridging Ties Complement Strong Ties? 253

view emphasizes the benefits of structural holes, through knowledge integration. We also theoret-
based on the premise that they expose members ically develop the idea that bridging ties comple-
to novel domains, unique ideas, diverse resources, ment strong ties in such portfolios. We conceptual-
and multiple thought worlds that presumably facil- ize alliance performance as the capacity to simul-
itate innovative outcomes (Burt, 1997). Yet, the taneously exhibit alignment with alliance objec-
observed benefits of such sparse ties are mixed tives and adaptiveness to changes in the environ-
and their observed influence on innovativeness ment; or what Gibson and Birkinshaw (2004) label
marginal (Rodan and Galunic, 2004). as ambidexterity. Our focal dependent variable is
A third theme is the greater emphasis in the therefore alliance ambidexterity. The context of
social networks literature on dyadic ties between the field study is 42 innovation-seeking e-business
individuals or firms, and the relative scarcity of (Internet) project alliances between Epsilon (a
network ties perspectives in the interfirm alliances pseudonym), a 400,000 person American services
literature (with the exception of Yli-Renko, Autio, conglomerate with operations in 200 countries, and
and Sapienza (2001)). This has constrained prior its myriad partners in 20002002. This was an
work predominantly to the examination of spe- appropriate context to examine these ideas because
cific dyadic ties rather than portfolios of inter- these projects were innovation-seeking, with little
firm ties. Nevertheless, the significance of studying precedent, and required a diverse array of spe-
portfolios of ties is well acknowledged (Baker, cialized knowledge, skills, and capabilities span-
1990; Uzzi, 1997). In a similar vein, most alliance ning a variety of industries, firms, and expertise
research has not made an explicit distinction domains. The results support many of the proposed
between alliance characteristics at the broad ideas.
alliance level and at the individual project level The study makes three noteworthy theoreti-
(Gerwin and Ferris, 2004). Such inattention to
cal contributions. First, it highlights a paradoxi-
portfolios of project-specific ties masks impor-
cal tension between strong ties and bridging ties
tant subtleties, the notable one being that a given
in empirically showing how the former improve
portfolio can simultaneously have two character-
alliance ambidexterity by enhancing project-level
istics (e.g., both strong ties and structural holes)
knowledge integration, but the latter hinder it
that would be less likelyeven mutually exclu-
siveat the dyadic level or broader alliance while providing access to a broader repertoire
level. Refocusing the unit of analysis allows delin- of skills, expertise, and capabilities. Second, it
eation of compatibilities between network struc- demonstrates complementarities between strong
ture configurations that were previously assumed ties and bridging ties. In other words, strong ties
incompatible.1 Finally, while the intervening pro- provide mechanisms to integrate a diverse reper-
cess of knowledge integration is implied in both toire of skills and expertise that are made accessi-
the ties and alliances literature, we know of no ble by bridging ties, which span structural holes.
prior work that has explicitly used it to develop Finally, the paper introduces knowledge integra-
and test an explanation for how bridging and tion into the nomology and shows that the influ-
strong ties in an alliance tie portfolio translate ence of strong ties, bridging ties, and their inter-
into alliance performance or outcomes. There- actions on alliance ambidexterity is fully medi-
fore, the complementarities between strong ties ated by knowledge integration. These findings
and bridging ties and the role of the knowledge raise some provocative theoretical questions about
integration process through which they influence managing the tradeoffs between bridging ties and
alliance performance remains imperfectly under- strong ties in designing innovation-seeking inter-
stood. The objective of this study is to address firm alliances.
these gaps. The remainder of the paper is organized as
We theoretically develop a nomological network follows. The next section theoretically develops
in which two tie portfolio characteristicsthe the hypotheses. The data collection, analyses, and
presence of structural hole-spanning bridging ties results are discussed in the subsequent sections. We
and strong tiesinfluence alliance performance conclude the paper with a discussion of its contri-
butions and its implications for the literature on
We are grateful to an anonymous reviewer for suggesting this social network configuration, interfirm alliances,
point. and firm strategy.
Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
254 A. Tiwana

THEORY AND HYPOTHESES The motivations for forming such multifirm

project alliances are therefore two-fold. First, to
Research context: Internet-centric e-business gain access to a broader array of complementary
project alliances expertise, skills, and capabilities. Second, to share
risk given the associated technological uncertainty,
Building on Amit and Zott (2001), we define an systemic complexity, costs, and paucity of prece-
e-business project alliance as a formalized collab- dent solutions. Such projects also require a variety
orative arrangement among two or more firms to of tacit technical, business, and domain knowl-
jointly develop a previously nonexistent software- edge, as is typical of innovative software projects
based system used to create business value through (Faraj and Sproull, 2000; Robillard, 1999; Rus
the Internet. As firms attempt to competitively and Lindvall, 2002). These projects therefore are
differentiate themselves by creatively exploiting well representative of project alliances that are
Internet technologies, understanding how they col-
knowledge-intensive and require the melding of
laborate to develop the enabling Internet applica-
specialized knowledge spanning diverse domains.
tions is critical because their successful realiza-
Figure 2 summarizes the proposed research model,
tion is what breathes life into innovative Internet-
which is theoretically developed in the forthcom-
centric business models. E-business systems create
ing discussion.
value by reconfiguring coordination mechanisms
and integrating transactional structures, resources,
capabilities, and relationships spanning suppliers, Knowledge integration and alliance
partners, and customers (Amit and Zott, 2001). ambidexterity
However, many e-business processes that attempt
to redefine existing value chains frequently cross Solutions to complex innovation problems often
traditional industry boundaries. Interorganizational require integration and synthesis of diverse, com-
development of an e-business system thus encom- plementary knowledge (Henderson and
passes know-how that resides at the interstices Clark, 1990; Nickerson and Zenger, 2004; Obst-
of specialized firms, their suppliers, customers, feld, 2005). The successful realization of an inno-
and other partners (Amit and Zott, 2001; Pow- vative project solution represents what Obstfeld
ell, Koput, and Smith-Doerr, 1996). This neces- (2005) describes as combinatorial innovation i.e.,
sitates bringing together an assemblage of project it requires novel recombinations of ideas, re-
participants with diverse industry, functional, and sources, and knowledge. Others have described
technical knowledge (Srinivasan, Lilien, and RaN- this process of integrating disparate knowledge and
gaswamy, 2002), which must be integrated and skills as knowledge transformation (Carlile and
then successfully embodied in the design of the Rebentisch, 2003), combinative capacity (Kogut
software. and Zander, 1992), transformative capacity (Garud

Tie Portfolio

Strong ties


Strong ties
H4(+) Knowledge H1(+) Alliance
Integration Ambidexterity
Bridging ties


Bridging ties

Figure 2. Research model

Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
Do Bridging Ties Complement Strong Ties? 255

and Nayyar, 1994), and realized absorptive capac- knowledge is integrated. For example, an Internet-
ity (Zahra and George, 2002). In complex tech- based logistics system project in our study drew on
nology development projects, a variety of comple- merge-in-transit logistics heuristics developed by
mentary, specialized knowledge must be applied a shipping company, the software objects exper-
to solve project-specific problems (Carlile, 2002, tise of a software firm, the satellite positioning
2004; Kogut and Zander, 1992). Therefore, the expertise of a global positioning system (GPS)
challenge in our context is one of integrating mul- manufacturer, and the street-level mapping exper-
tiple specialized inputs to successfully accomplish tise of a cartographer. This variety of expertise
a project. was collectively brought to bear on the conceptual-
Prior research has conceptualized knowledge ization, design, and subsequent implementation of
integration in two different ways: one as across an innovative Internet-based logistics management
individuals in different, dispersed groups (Adler, system.
1989; Nickerson and Zenger, 2004; Okhuysen and Although alliances provide a mechanism for
Eisenhardt, 2002), and the second as integration gaining access to complementary know-how (tacit,
of different streams of knowledge (Carlile, 2004; sticky, and noncodifiable knowledge) and tech-
Larsson et al., 1998; Nahapeit and Ghoshal, 1998). nological capabilities (Kale, Singh, and Perlmut-
Recent work has incorporated both these facets by ter, 2000; Mowery, Oxley, and Silverman, 1996),
proposing a richer conceptualization of knowledge their coordinated utilization at the project level is
integration across groups and across specialized necessary to translate them into alliance perfor-
streams of knowledge (Sabherwal and Becerra- mance. In innovation-seeking projects, it is often
Fernandez, 2005). Following recent project-level initially difficult to clearly envision the intended
extensions (Carlile and Rebentisch, 2003; Okhuy- project outcome. For example, different project
sen and Eisenhardt, 2002; Sabherwal and Becerra- stakeholders might have different perspectives on
Fernandez, 2005; Tiwana and McLean, 2005) what they view as the ideal solution (Dougherty,
of Grants (1996a) conceptualization, we define 1992). These perspectives must be reconciled for
knowledge integration as the process of jointly the project participants to arrive at a shared con-
applying specialized knowledge held by various ceptualization of the envisioned solution.
alliance partners at the project level. In this per- Furthermore, as the development process pro-
spective, knowledge integration creates value gresses, the project team might also encounter
through the application of alliance partners spe- unexpected problems, recognize new opportuni-
cialized knowledge to project specific activities. ties, or face changing market needs. For exam-
This emphasis on accessing and utilizingrather ple, project objectives can evolve during develop-
than acquiring (Argote, McEvily, and Reagans, ment in synchrony with new information, unantic-
2003)alliance partners complementary knowl- ipated shifts in the underlying technologies, and
edge has also been noted in the strategic alliances emerging market requirements that did not exist
literature (Dyer and Singh, 1998; Grant and Baden- or were not identified at the outset of a project
Fuller, 2004; Nickerson and Zenger, 2004; Oxley (Bhattacharya, Krishnan, and Mahajan, 1998). Fur-
and Sampson, 2004) and in the new product thermore, the absence of guiding precedents or
development literature (Carlile, 2002; Carlile and solutions in innovation-seeking projects and the
Rebentisch, 2003). Our emphasis on knowledge inherent need for improvisation make it inappropri-
integration is therefore a notable departure from ate to assess performance solely using traditional
prior work on social networks, where knowledge yardsticks such as efficiency and effectiveness
transfer or acquisition have often been used even (Yourdon, 2002). It is therefore not only impor-
though the underlying argument implicitly remains tant for such project teams to produce outcomes
one about knowledge integration. that are well aligned with the project alliance
In software development projects, integrated objectives, but also to successfully adapt to new
knowledge is embodied in the design of the soft- information that emerges after development work
ware. The process of knowledge integration allows has begun (e.g., by quickly producing alternative
the alliance partners to develop a shared concep- designs (Dougherty, 1992)). Gibson and Birkin-
tualization of what the software ought to do and shaw (2004) describe this capacity to be simul-
how it should do it. To illustrate this point, con- taneously aligned and adaptive as ambidexter-
sider an example of how, where, and to what end ity. Drawing on this thinking, we use alliance
Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
256 A. Tiwana

ambidexterity to assess alliance performance at the ties have a greater potential for generating novel
project level. solutions but lack the characteristics to realize
The greater the extent of knowledge integra- that potential. However, in project alliances where
tion, the greater the prospect that the perspectives individual representatives bring diverse skills and
of diverse alliance partners will be cross-fertilized knowledge, reducing the issue of integrating their
and reflected in the project solution. Thus, knowl- knowledge to one of information flow is problem-
edge integration facilitates alignment with alliance atic because the relevant knowledge also includes
objectives through syntheses of the unique insights tacit, difficult to communicate know-how in addi-
from the thought worlds of various project par- tion to codifiable information. Such tacit knowl-
ticipants. Greater knowledge integration is also edge is easier for the holder to apply than it is
likely to simultaneously facilitate the recognition to express or transfer (Grant and Baden-Fuller,
and integration of new information about new 2004),2 a subtlety that the theory of weak ties fails
needs and constraints that arises while develop- to take into account (McEvily and Marcus, 2005).
ment work is in progress. Knowledge integration A different type of tie, distinct from the weak-
therefore facilitates correction of misalignments strong characterization, is identified in
with changing exogenous environments and stake- Burts (1992) structural holes theory: A bridging
holder needs during the development process, thus tie, defined as a tie that spans a structural hole
enhancing alliance ambidexterity. Empirical stud- (Regans et al., 2004). The notion of bridging ties
ies in a variety of contexts broadly concur with this is grounded in structural holes theory that sug-
perspective. For example, studies in pharmaceuti- gests that a tie that connects actors separated by
cal (Henderson and Cockburn, 1994), biotechnol- a structural hole enables access to new and novel
ogy (Pisano, 1994), scientific tools (Hoopes and information, thus serves as a bridge to new oppor-
Postrel, 1999), and software development (Faraj tunities. Thus bridging ties link a focal firm or
and Sproull, 2000; Patnayakuni, Rai, and Tiwana, actor to contacts in economic, professional, and
2007; Tiwana and McLean, 2005) contexts have social circles that are otherwise not accessible to
shown a positive association between various mea- it (McEvily and Zaheer, 1999). A defining charac-
sures of performance and the effective integration teristic of bridging ties is nonredundancy (McEvily
of the expertise of different actors in the develop- and Zaheer, 1999; Regans and Zuckerman, 2001;
ment process. This leads to our first hypothesis. Uzzi, 1996), i.e., the connected actors are hetero-
geneous in their backgrounds, experiences, knowl-
Hypothesis 1: Knowledge integration at the edge, and skills. In contrast, the perspectives, capa-
project level enhances alliance ambidexterity in bilities, and knowledge of individuals in homoge-
innovation-seeking project alliances. nous groups are likely to be relatively redundant
(Regans and Zuckerman, 2001). This characteri-
Features of the tie portfolio that binds project zation of bridging ties is also recognized by Burt
participants in an alliance, and by association their (1992), who emphasizes that the potential of a tie
organizations, influence knowledge integration at to provide novel and diverse information depends
the project level, as discussed next. on whether the tie spans a structural hole. Burt
(2004) also found evidence in a recent field study
Strength of ties and structural holes that structural holes are indeed sources of novel
knowledge and perspectives, or what he described
Much of the early research on tie strengths draws as good ideas. Therefore, individuals connected
on Granovetters (1973) conceptualization of ties by bridging ties are likely to have heterogeneous
with a focus on information flows among individu- backgrounds, capabilities, skills, and expertise. In
als. The crux of his theory is that strong ties among summary, strong ties bind and bridging ties span
individuals facilitate information flows but such
individuals are likely to possess highly redundant
information. In contrast, individuals with weak 2
Technological capabilities are based on tacit knowledge (Mow-
ties will likely possess more diverse information ery et al., 1996), which is carried through firms human capital
(increasing the likelihood of finding novel ideas), (Amit and Schoemaker, 1993; Verona, 1999) and know-how that
resides in individuals minds (Nonaka, 1994). It is therefore dif-
but the weakness of their ties increases the diffi- ficult to disassociate technological capabilities and know-how
culty of information flow among them. Thus weak from individuals by treating it only as information.

Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
Do Bridging Ties Complement Strong Ties? 257

structural holes. We next consider these two char- The influence of strong ties in project alliances
acteristics in the context of project alliances tie
portfolios. Strong ties refer to the level of trust, reciprocity,
and proximity of interaction that characterize the
portfolio of ties among the participants in a project
Alliance tie portfolios alliance (Kale et al., 2000). Such strength of ties
therefore reflect the level of relational embedded-
Since the locus of project activities is an assem- ness and cohesiveness among alliance partners at
blage of individuals drawn from alliance firms the project level (Kale et al., 2000; Uzzi, 1996).
into a project team, we focus on projects rather Stronger ties, which are more conducive to infor-
than firms or the network as our unit of analy- mation flows (Granovetter, 1973), are also more
sis. This is an important distinction because nei- likely to facilitate knowledge integration at the
ther prior network structure studies nor strategic project level for three reasons. First, because the
alliances studies have usually made the distinc- source and recipient are more likely to share a
tion between ties at the project level (instead common language, they are better able to absorb
focusing on the dyadic level or broader alliance new ideas from each others domain of specializa-
level), although much innovation work in strate- tion (Regans and McEvily, 2003). Second, knowl-
gic alliances is accomplished at the project level edge also includes tacit elements that cannot be as
and by project teams (Gerwin and Ferris, 2004). readily communicated as information. Innovation-
This choice also complies with other scholars sug- seeking alliances are especially vulnerable to such
gestions to use a production system rather than knowledge transfer problems because it is often
a firm or network as the unit of analysis (Child difficult to enforce, measure, or monitor the tacit
and McGrath, 2001), and to examine in the small knowledge contributions of various participants
group setting where tacit knowledge in collabora- (Gulati and Singh, 1998). Shared values, coop-
tive networks of partners is actually used (Dyer erative norms, and a sense of reciprocity that
and Nobeoka, 2000). Focusing on the project as characterize such cohesive ties can collectively
the unit of analysis allows us to conceptualize enhance knowledge transfer (Regans and McEvily,
characteristics in a portfolio of ties at the project 2003), and by extension knowledge integration.
level, some of which have been assumed to be Third, the trust and reciprocity facets of strong
in conflicteven mutually exclusivein prior ties provide a context conducive for knowledge
research. For example, a tie portfolio that char- integration because the source and recipient are
acterizes an alliance at the project level can simul- less likely to discredit each others perspectives in
taneously possess characteristics of strong ties and attempting to solve project problems. Such trust
bridging ties. While the importance of a portfo- is a critical antecedent to joint problem solving in
lio conceptualization of ties is recognized in the alliances (McEvily and Marcus, 2005). By virtue
literature (Baker, 1990; Uzzi, 1997), the majority of their reciprocal relationships, alliance members
of prior of work has focused on the dyadic level are also less likely to engage in cost-benefit cal-
that was more appropriate to its research questions culus before contributing proprietary or valuable
(e.g., Levin and Cross, 2004; Regans and McEvily, knowledge to a collaborative project (Molm, Peter-
2003). Examining how tie portfolio characteristics son, and Takashaki, 1999). The degree of such
at the project level influence alliance ambidex- cohesion among individuals increases the will-
terity in conjunction with knowledge integration ingness and motivation to invest the time and
allows the conceptualization of a hitherto unex- effort to share knowledge with others (Regans and
plored mechanism vis-`a-vis knowledge integra- McEvily, 2003).
tionthe process of collectively applying alliance Several recent studies provide implicit support
partners specialized knowledge to collaborative for this perspective. For example, the level of
project activitiesthrough which they influence relational embeddedness or strong ties among col-
alliance ambidexterity. We therefore focus on laborators positively affects reciprocal assistance
these two project-level tie portfolio characteris- (Hansen, 1999), increases cohesiveness (Gulati and
ticsstrong ties and bridging tiesto unmask the Singh, 1998), lowers opportunistic withholding
subtleties than can potentially be veiled by a dyadic of knowledge (Yli-Renko et al., 2001), enhances
focus. communication effectiveness (Dyer and Singh,
Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
258 A. Tiwana

1998), and enhances cooperation (Uzzi, 1996). A manifests itself remains theoretically under-
positive relationship between the proximate con- explored. For example, a recent study showed that
struct of tie cohesiveness and knowledge trans- Canadian mutual funds with bridging ties tend to
fer at the dyadic level has also been empirically perform better than those without (Zaheer and Bell,
demonstrated (Regans and McEvily, 2003). The 2005). Regans et al. (2004) similarly found that
social capital literature also supports this assertion, relationships that span multiple knowledge pools
as evidenced in the positive association between (network range) increase individuals capacity for
the relational dimension of social capital and novel problem solving. In contrast, Burt (2004)
ease of knowledge transfer (Tsai, 2001). There- found that structural holes lead to good ideas
fore, strong ties in a project alliance tie portfo- but found no evidence that they also lead to the
lio enhance knowledge integration. It is primarily successful implementation of those ideas. Burts
because of such synergistic recombination of spe- finding is consistent with Obstfelds observation
cialized knowledge that unique new solutions can that bridging ties pose difficulties in integrating
be generated and relational rents (Dyer and Singh, ideas, or face what Obstfeld (2005) describes as
1998) realized. Without knowledge integration, the the action problem. A theoretical explanation for
latent potential of strong ties for enhancing alliance how bridging ties influence alliance ambidexterity
outcomes cannot be realized. These ideas are sum- however remains underdeveloped. We develop the
marized in the following hypotheses. idea that bridging ties create a potential for novel
knowledge recombinations, which, only if realized
Hypothesis 2a: Strong ties are positively related through knowledge integration, enhance alliance
to knowledge integration in innovation-seeking ambidexterity.
project alliances. Although bridging ties facilitate access to nonre-
dundant expertise, perspectives, and capabilities,
Hypothesis 2b: The effect of strong ties on which increases the potential for innovative recom-
alliance ambidexterity is fully mediated by bination, their dissimilarity lowers the likelihood
knowledge integration. that such recombinations can be realized. Bridg-
ing ties pose an action problem because individ-
uals around structural holes have different exper-
The influence of bridging ties in project alliances
tise, unique perspectives, and often employ differ-
Recall that the defining characteristic of bridg- ent professional language. As Spender and Grant
ing ties is that they connect individuals with (1996) caution, the benefits of knowledge depend
diverse and heterogeneous backgrounds, experi- not on how much knowledge is available but how
ences, knowledge, and skills. As the heterogene- effectively it is recombined and exploited. The
ity of members in a project alliance increases, so more dissimilar these expertise bases and perspec-
does the range and diversity of ideas, perspectives, tives, the more arduous it is to realize their recom-
and information (Regans and Zuckerman, 2001). bination at the project level (Hollenbeck et al.,
This broadens the repertoire of available solu- 1995). The ability to exploit such knowledge at
tions and the likelihood of novelty emerging from the project level is a function of their absorp-
the recombination of previously isolated perspec- tive capacity or the prior knowledge of the project
tives (Ahuja, 2000; Lapre and Wassenhove, 2001; participants (Cohen and Levinthal, 1990; Szulan-
Pelled, 1996). Project participants with heteroge- ski, 1996; Zahra and George, 2002). Since bridging
neous expertise then serve as ports of access to ties by definition lack such redundancy in exper-
potentially valuable, nonredundant bodies of spe- tise, the potential to integrate at the project level
cialized expertise (McEvily and Zaheer, 1999). A the diverse array of knowledge made accessible
team with an abundance of structural holes (i.e., through such bridging ties is constrained by the
high on bridging ties) therefore creates potential alliance partners dissimilarity. In a related vein,
opportunities for novel syntheses of diverse ideas Doughertys (1992) study of 17 innovation-seeking
from multiple specialized domains of expertise. projects showed that the differences in team mem-
However, the relationship between bridging ties- bers thought worlds prevented them from syn-
like features and innovation outcomes has pro- thesizing their perspectives and knowledge. Fur-
duced somewhat ambivalent results, in part thermore, project participants with heterogeneous
because mechanisms through which the effect knowledge and skills are likely to be embedded in
Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
Do Bridging Ties Complement Strong Ties? 259

different social and professional networks, and are are integrated and jointly applied to a projects
therefore less likely to share the norms, vocabu- conceptualization and implementation. The action
lary, and culture that facilitate knowledge sharing problem can therefore overwhelm the informa-
(Lovelace, Shapiro, and Weingart, 2001; Powell tional advantages that structural holes might offer
et al., 1996). When the project team members can- (Obstfeld, 2005), especially when it is difficult
not frame their knowledge in terms that others can to successfully integrate alliance partners diverse
understand, comprehending and subsequently inte- knowledge, capabilities, and resources. This is
grating that knowledge can be difficult. This makes where strong ties can complement bridging ties.
it hard for the project participants to develop a Complementarities are said to exist when having
shared understanding about the project, therefore more of one thing increases the returns of hav-
making it difficult to coordinate the application ing more of another (Milgrom and Roberts, 1995).
of their skills to their joint tasks (Ancona and Statistically, this represents a positive interaction
Caldwell, 1992). We therefore hypothesize that the effect. Viewed at the project tie portfolio level, it
knowledge integration difficulties posed by bridg- is plausible that a project alliance can simultane-
ing ties will overshadow the potential for realizing ously be high on strong ties and on bridging ties. A
novel recombinations of the available expertise, project team that simultaneously possesses strong
skills, and capabilities. Further, since access to het- ties and bridging ties will have access to a diverse
erogeneous expertise influences alliance ambidex- array of specialized knowledge, perspectives, and
terity only if it is effectively integrated at the skills and have the mechanisms to integrate that
project level, we expect their influence on alliance knowledge at the project level. This combination
ambidexterity to be fully mediated by knowledge of tie characteristics is what Burt (1992) would
integration. These ideas are summarized in the fol- describe as an ideal configuration. According to
lowing hypotheses. Regans and McEvily (2003), as the diversity of
the knowledge pools that such members are con-
Hypothesis 3a: Bridging ties are negatively nected with increases (i.e., simultaneous presence
related to knowledge integration in innovation- of bridging ties in a project), so does their ability
seeking project alliances.
to convey complex ideas to heterogeneous audi-
ences that span diverse knowledge pools. Individ-
Hypothesis 3b: The effect of bridging ties on uals connected by strong ties can therefore serve
alliance ambidexterity is fully mediated by
as brokers, translators, and interpreters of the per-
knowledge integration.
spectives of other team members with whom they
possess stronger ties. Therefore, strong ties help
Complementarities between strong ties and integrate knowledge, skills, and capabilities that
bridging ties are made accessible by bridging ties. Therefore,
The foregoing arguments suggest that the success- we expect strong ties and bridging ties to exhibit
ful accomplishment of innovative projects requires complementarities, i.e., a positive interaction effect
the diversity of expertise and capabilities provided on knowledge integration. (Although later authors
by bridging ties on the one hand and the mech- such as Regans, Zuckerman, and McEvily (2004)
anisms to integrate them at the project level on have also implicitly recognized potential comple-
the other. Absence of the former presents what mentarities between network structures previously
Obstfeld (2005) characterizes as the idea prob- assumed to be in conflict, we label these Burt com-
lem and the latter as the action problem. There- plements for expository purposes.)
fore, while bridging ties provide a greater potential However, such complementarities influence
for innovation, the necessary integration of var- project activities only because they facilitate
ious specialized knowledge bases requires bridg- knowledge integration at the project level. There
ing across terminology, nomenclature, lexicon, and is no theoretical reason to expect that they would
rules of thumb associated with specialized knowl- benefit a project other than by enhancing knowl-
edge domains spanned by a project. The potential edge integration at the project level. We therefore
for realizing an innovative project solution is lost expect that their effect on alliance ambidexterity
unless the diverse, specialized knowledge, capabil- will be fully mediated by knowledge integration.
ities, and perspectives of project alliance partners This leads to our final hypotheses.
Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
260 A. Tiwana

Hypothesis 4a: Strong ties complement bridg- METHODS

ing ties in enhancing knowledge integration in
innovation-seeking project alliances. Research and data collection context
Since the objective of the study was to examine
Hypothesis 4b: The influence of the comple- innovation-seeking project alliances, we focused
mentarity between strong ties and bridging ties on studying novel projects with little precedent
on alliance ambidexterity is fully mediated by instead of incremental projects that refined existing
knowledge integration. technologies. In such collaborations, both the need
for knowledge integration as well as the variance
To summarize the key ideas: (1) There exists in the degree to which it is achieved is likely to
a tension between strong ties and bridging ties, be high. This is especially noteworthy because the
(2) they exhibit complementarities, and (3) their majority of prior research on network structure and
influence on alliance ambidexterity is fully medi- ties, including Obstfelds (2005) recent study, have
ated by knowledge integration, a central explana- studied incremental innovation work. A second
tory mechanism in our nomological network. The notable facet of the research setting is our emphasis
hypothesized relationships and their underlying on tie portfolios at the project level in multifirm
theoretical logic are summarized in Table 1. alliances, as opposed to dyadic ties.

Table 1. Summary of the core theoretical ideas and hypotheses

Core theoretical Hypothesis Specific theoretical idea Testable hypothesis Supported?

idea #

Knowledge integration 1 Higher levels of knowledge Knowledge integration Yes

enhances alliance integration among (+) Alliance
ambidexterity collaborators in an alliance ambidexterity
enhances alliance
A tension exists between 2a Strong ties among collaborators Strong ties (+) Yes
strong ties and structural in an alliance facilitates Knowledge integration
hole spanning (bridging) knowledge integration.
ties in interfirm alliances.
2b Strong ties influence alliance Statistical mediation of the Yes
ambidexterity because they relationship between
enhance knowledge strong ties and alliance
integration. ambidexterity by
knowledge integration.
3a Bridging ties among Bridging ties () No
collaborators in an alliance Knowledge integration
impedes knowledge
3b Bridging ties influence alliance Statistical mediation of the No
ambidexterity primarily relationship between
because they influence bridging ties and alliance
knowledge integration. ambidexterity by
knowledge integration.
Complementarity between 4a Strong ties complement (Strong ties x Bridging ties) Yes
strong and bridging ties. bridging ties in facilitating (+) Knowledge
knowledge integration i.e., integration
there exists a positive
interaction effect.
Such complementarities 4b The effect of the interaction Statistical mediation of the Yes
enhance alliance between strong ties and relationship between the
ambidexterity primarily bridging ties on alliance (Strong ties Bridging
because they enhance ambidexterity is fully ties) interaction and
knowledge integration. mediated by knowledge alliance ambidexterity by
integration. knowledge integration.

Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
Do Bridging Ties Complement Strong Ties? 261

In the dot com era (19982002) associated assess within-project interrater agreement using the
with the widespread adoption of the internet in Intraclass Correlation Coefficient (James, Dema-
business, it was common for diverse firms with ree, and Wolf, 1993). This approach is com-
different specialties to partner in developing inno- parable to a nested ANOVA test to determine
vative e-business applications. An example of one whether membership in a given project team also
such partnership involving a logistics firm, a soft- led to similar patterns of responses. The Intra-
ware object technology firm, a satellite global posi- class Correlation Coefficient values summarized
tioning services (GPS) provider, and a cartogra- in Table 2 (0.64 to 0.87) suggest sufficient within-
phy firm was described earlier in the paper. Firms project agreement to justify aggregation of individ-
that have long-standing histories of partnering can ual responses as simple averages. This approach
potentially confound the focal relationships being enhances the reliability of project level assess-
examined here; we therefore chose to focus on ments, mitigates the threat of common methods
interfirm project alliances that had no precedent bias, and has precedent in project-level studies
in the members industries and projects involv- (e.g., Faraj and Sproull, 2000). The construct mea-
ing participants with no prior collaborative history. sures exhibited sufficiently high reliability (as sug-
While this narrow focus restricts generalization to gested by the high scale alphas ranging from
other types of projects, it allows careful delineation 0.89 to 0.93 at the project alliance level and
of the focal theoretical relationships without con- 0.87 to 0.96 at the unaggregated level), conver-
founding them with alliance history. gent validity, and discriminant validity (assessed
using exploratory factor analyses among the mea-
Data collection surement items both before and after aggregation)
(Nunally, 1978). Since the scales were adapted,
The sample of project alliances included in the we also confirmed discriminant validity and con-
study was identified based on information provided vergent validity using confirmatory factor analysis
by the top management team at the Internet busi- in a structural equation model.
ness (e-business) applications incubator of Epsilon The face validity of the survey items was
(a pseudonym), a 400,000 person American ser- assessed by iteratively refining the item wording
vices conglomerate with $43 billion in annual rev- and terminology with a panel of nine experts (three
enues and operations in 200 countries. Data were senior managers and six academic domain experts)
collected from multiple informants in each project and then pretesting the instrument with a conve-
alliance using a survey instrument. The objective nience sample of 79 software project participants
of this multi-informant data collection strategy was (none of whom were included in the final dataset).
to mitigate threats of bias that might arise if only Table 2 summarizes the key psychometric proper-
one or two respondents were used to assess each ties, within-alliance interrater agreement statistics,
project. A questionnaire was sent to 173 individual and construct correlations. Since the aggregation of
team participants in 46 project alliances spanning individual responses into project alliance-level data
various organizations (as many as five, on average points reduces the sample size, all measurement
2.5 organizations), using contact information pro- properties for the aggregated data were also veri-
vided by the top management at the incubator. We fied using confirmatory factor analyses in a partial
received 82 percent (142/173) individual- and 91 least squares-based structural equation model (Hul-
percent (42/46) project-level response rates. Data land, 1999).
for all exogenous constructs were collected from
multiple participants in each project alliance, and
data for alliance ambidexterity were collected from Measures
the senior managers responsible for each project.
On average, we had four respondents for each Whenever possible, preexisting scales were
project alliance. Individual responses for the items adapted for measuring the constructs in the study.
pertaining to strong ties, knowledge integration, All our measures used five-point Likert scales.
bridging ties, and turbulence were aggregated to The items and their validity statistics at both the
project-level construct scores, consistent with our individual and aggregated levels are summarized
use of projects as the unit of analysis. Before in the Appendix. Following the conceptualization
such aggregation can be done, it is necessary to of tie strength in a project alliance ties portfolio,
Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
A. Tiwana

Copyright 2007 John Wiley & Sons, Ltd.

Table 2. Descriptive statistics, psychometric properties, and construct correlations

Mean SD ICCa 1 2 3 4 5 6 7 8 9 10 11

1. Technological turbulence 2.79 0.83 0.79

2. Project length 6.2 4.2 0.03
3. Project novelty 3.27 0.71 0.09 0.40
4. Alliance organization count 2.47 0.24 0.03 0.02
5. Project stageb 0.02 0.02 0.10 0.12
6. Project team size 8.54 4.41 0.29 0.05 0.08 0.14 0.12
7. E-business experience 2.25 1.13 0.41 0.10 0.17 0.17 0.16 0.01
8. Information technology experience 7.87 8.46 0.17 0.11 0.20 0.37 0.12 0.02 0.14
9. Strong ties 3.71 0.74 0.73 0.19 0.01 0.11 0.20 0.16 0.01 0.43 0.19
10. Bridging ties 3.88 0.49 0.87 0.18 0.22 0.15 0.15 0.22 0.15 0.23 0.11 0.03
11. Knowledge integration 3.74 0.59 0.64 0.20 0.10 0.04 0.30 0.16 0.17 0.24 0.08 0.71 0.08
12. Alliance ambidexterityc 16.1 4.57 0.68 0.17 0.09 0.16 0.09 0.64 0.02 0.31 0.07 0.18 0.40 0.17

p < 0.05; p < 0.01
ICCs indicate within-alliance interrater agreement; all data aggregated to project alliance level; b median value 80100% complete
alignment dimension mean = 3.87 (sd 0.65) and adaptation dimension mean = 4.09 (sd 1.01)

DOI: 10.1002/smj
Strat. Mgmt. J., 29: 251272 (2008)
Do Bridging Ties Complement Strong Ties? 263

the degree of presence of strong ties was mea- course of development, the project successfully
sured using Kale et al.s (2000) five-item measure met its time, budget, features and functionality,
that assessed the extent of close personal inter- project objectives, and business needs. Three of
action, reciprocity, mutual trust, mutual respect, these five items were retained after the exploratory
and personal friendship at multiple levels among factor analysis, as shown in the Appendix. The
the members of the project team. The scale was adaptation dimension used three items to assess
labeled relational capital in the original study and how well the project alliance was able to success-
maps to the conceptualization of strong ties at fully manage scope changes, resolve unexpected
the project level. This approach overcomes the problems, and deliver a relatively stable system to
coarseness of using infrequency of interaction as the projects emergent requirements.
a proxy for tie weakness type concepts, of which
McEvily and Zaheer (1999) caution. We measured
Control variables
the extent of presence of bridging ties in a tie port-
folio by tapping into the degree to which it spans We included controls for technological turbulence
structural holes, i.e., the diversity of experiences, (Poppo and Zenger, 1998), project duration (Nidu-
backgrounds, skills, and expertise among mem- molu, 1995), perceived project novelty, team size
bers of each project alliance (Burt, 1992, 2004; (Regans et al., 2004), average information tech-
Regans and Zuckerman, 2001; Uzzi, 1996). This nology and e-business experience in each project
was measured using an adaptation of a three-item team, and project stage. Except for Poppo and
scale from Campion, Medsker, and Higgs (1993) Zengers (1998) two-item measure for turbulence
that assessed the project participants diversity of and a one-item team-rated measure for perceived
expertise, skills, and backgrounds. A new scale novelty, all control variables were measured using
was developed to measure knowledge integration single item measures. We also controlled for the
using an iterative, multistep procedure, beginning number of different organizations from which par-
with a review of the literature. We drew from exist- ticipants were drawn for each project (alliance
ing tacit knowledge inventories (Sternberg, 2000), organization count), recognizing that this affects
conceptual descriptions of knowledge integration the extent to which a project alliance draws on
(Grant, 1996a, 1996b; Pisano, 1994), and Hoopes shared knowledge and on informal knowledge net-
and Postrels (1999) qualitative case descriptions. works (Powell et al., 1996). The sample included
We retained three items that measure individual teams with no prior history, thereby mitigating the
perceptions of knowledge integration at the project confounding effect of prior collaborative experi-
level by assessing the extent to which the partici- ence (Regans et al., 2004).
pants in each project alliance had synthesized their
knowledge and skills in a coherent project-level
design. RESULTS
Following the logic presented in the theory
section and Gibson and Birkinshaws (2004) oper- The hypothesized relationships were tested fol-
ationalization, we measured alliance ambidexterity lowing the mediated regression guidelines out-
as the product of alignment with project alliance lined by MacKinnon et al. (2002). This required
objectives and adaptation to new information that establishing a relationship between the indepen-
emerged over the course of the project. Gibson and dent variables with the mediator (knowledge inte-
Birkinshaw (2004 : 211) describe these as the two gration), and the mediator with the dependent vari-
interrelated and nonsubstitutable dimensions of able (alliance ambidexterity). The superannuated
ambidexterity. Their conceptualization mirrors a approach outlined by Baron and Kenny (1986)
second order formative construct (Jarvis, Macken- also required assessing the direct effect between
zie, and Podsakoff, 2003); its reliability was also the independent variables and the dependent vari-
confirmed using partial least squares-based con- able, and showing that the relationship signifi-
firmatory factor analysis (Hulland, 1999). The cantly weakens after the introduction of the medi-
alignment dimension of alliance ambidexterity was ator. Both Kenny, Kashy, and Bolger (1998) and
measured by five items that assessed the extent to MacKinnon et al. (2002) have more recently rec-
which, given the marketplace-mandated changes ognized that this approach of also requiring direct
and new business requirements that arose over the effects is overly restrictive. Since no direct effect
Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
264 A. Tiwana

was observed from the independent variables to and positive relationship with alliance ambidex-
the dependent variable in the absence of the medi- terity ( = 0.328; t-value = 1.89; p < 0.05), sup-
ator (Model 1, Step 2 in Table 3), we followed porting Hypothesis 1. The R2 increase attributable
the aforementioned guidelines provided by MacK- to adding knowledge integration to the model was
innon et al. (2002). statistically significant at the 1 percent level (F-
Two stepwise regression models were used to change = 3.57, p < 0.001), thereby suggesting the
test the hypotheses, one with alliance ambidexter- predictive relevance of knowledge integration to
ity as the dependent variable (Model 1) and the the model. All control variables with the excep-
other with knowledge integration as the dependent tion of project novelty, size, alliance organization
variable (Model 2). The results of the analyses are count, and information technology were signifi-
presented in Table 3 (with the regression coeffi- cant. An interpretation for the nonsignificance of
cients pertaining to the hypothesis tests italicized). novelty might be due to low variance in per-
A summary of the results in the context of the ceived novelty across the projects in the sample
research model is presented in Figure 3. Since all because of our sampling strategy of examining
of the hypotheses are unidirectional, one-tailed T- only innovation-seeking projects. Average infor-
tests are appropriate. However, as a robustness mation technology experience might have been
check, we also evaluated the results using two- nonsignificant in explaining alliance ambidexter-
tailed tests. All of the relationships, with the excep- ity because the projects examined also drew from
tion of the knowledge integration to ambidexterity a variety of other specialized domains of exper-
path, that were significant with one-tailed tests tise. Variance from alliance organization count,
remained statistically significant with two-tailed which controlled for the number of organizations
tests as well. from which project participants were drawn, might
To test Hypothesis 1, which proposed a pos- have already been accounted for in the tie portfo-
itive relationship between knowledge integration lio variables. Two other patterns are noteworthy in
and alliance ambidexterity, we used a stepwise Model 1: (a) the direct effects of the predictors and
regression model in which the control variables, interaction term on the dependent variable remain
the predictors, the interaction term (bridging ties nonsignificant across all model steps and (b) the
x strong ties), and finally the mediator (knowl- statistical significance of control variables remain
edge integration) were sequentially introduced in unchanged as the predictors and interaction term
the model. The results are summarized as Model 1 are introduced in the model. We will return to these
in Table 3. Knowledge integration had a significant results in light of Model 2.

Control Variables
Technological turbulence (-)
Project length in months(-)
Alliance Project novelty
Tie Portfolio Alliance organization count
Configuration Project stage (+)
Project size
Strong ties E-business experience (-)
Information technology experience
H2 -0.23(-1.34)

Strong ties H4 Knowledge H1 Alliance

x 0.23*(2.16) Integration 0.33*(1.89) Ambidexterity
Bridging ties

-0.10(-0.91) 0.15(1.18)
Legend (T-statistic)
* p <0.05; ** p <0.01; *** p <0.001
Bridging ties Significant results are in bold
Dotted lines represent nonsignificant paths

Figure 3. Summary of results

Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
Table 3. Results

Model 1 Model 2
Alliance ambidexterity Knowledge integration

Copyright 2007 John Wiley & Sons, Ltd.

Step 1 Step 2 Independent Step 3 Step 4 Step 1 Independent Step 1
Controls variables (Main effects) Interaction term Mediator variables (Main effects) Interaction term

(T-value) (T-value) (T-value) (T-value) (T-value) (T-value)

(Constant) (4.99) (1.84) (1.90) (1.59) (3.47) (3.63)
Technological turbulence 0.40 .3.11/ 0.33 .2.37/ 0.33 .2.31/ 0.40 .2.84/
Project length (months) 0.19(1.56) 0.21 .1.68/ 0.21 .1.72/ 0.22 .1.83/
Project novelty 0.17(1.41) 0.15(1.21) 0.14(1.04) 0.17(1.35)
Alliance organization count 0.16(1.33) 0.15(1.12) 0.13(0.94) 0.15(1.16)
Project stage 0.60 (5.35) 0.58 (4.95) 0.56 (4.74) 0.53 (4.54)
Project size 0.02(0.18) 0.02(0.12) 0.01(0.12) 0.07(0.53)
E-business experience 0.33 .2.65/ 0.27 .1.91/ 0.29 .1.97/ 0.33 .2.37/
Information technology experience 0.11(0.92) 0.08(0.64) 0.10(0.78) 0.14(1.09)
Strong ties 0.02(0.15) 0.01(0.07) 0.23(1.27) 0.71 (6.32) 0.71 (6.61)
Bridging ties 0.15(1.15) 0.16(1.19) 0.15(1.18) 0 .10 (0 .91 ) 0.10(0.95)
Bridging ties x Strong ties 0.08(0.70) 0.17(1.34) 0.23 (2.16)
Knowledge integration 0.33 (1.89)
R2 Adj (Model F) 62.1% (6.75) 63.6% (5.43) 64.2% (4.89) 68.1 (5.17) 48.4% (20.25) 52.9 (16.33)
R2 (F-change) 1.5%(0.67) 0.6%(0.49) 3.9% (3.57) 4.5% (4.68)

p < 0.05; p < 0.01; p < 0.001, one-tailed test; significant in bold.
Most relationships also remain significant using two-tailed tests.
Coefficients associated with the hypotheses are italicized.
Do Bridging Ties Complement Strong Ties?

Strat. Mgmt. J., 29: 251272 (2008)


DOI: 10.1002/smj
266 A. Tiwana

Model 2, which predicts knowledge integration from this interaction term (Step 3, Model 1) was
(the mediator) using the independent variables, is nonsignificant. This suggests that the influence of
used to test Hypotheses 2, 3, and 4. Hypothesis the interaction between strong and bridging ties
2 predicted a positive relationship between strong and the dependent variable is fully mediated by
ties and alliance ambidexterity that is fully medi- knowledge integration, supporting Hypothesis 4b.
ated by knowledge integration. As Step 1 in Model
2 shows, the relationship between strong ties and
knowledge integration was positive and significant Limitations
( = 0.709; t-value = 6.32; p < 0.001) (support-
ing Hypothesis 2a) and the relationship between Before discussing the results, five limitations of
knowledge integration and alliance ambidexterity this study merit further discussion. First, the cross
was also positive and significant (Model 1, Step sectional nature of our data, while apt for test-
2). Furthermore, no direct effect was observed ing a variance model, cannot yield insights that
from strong ties to alliance ambidexterity, suggest- longitudinal data can into the microprocesses of
ing that the relationship is fully mediated. This knowledge integration. Second, the projects in our
supports Hypothesis 2b. Hypothesis 3a predicted sample were innovation-seeking and represent a
a negative relationship between bridging ties and finite universe of business partners of one large
knowledge integration, which was not supported multibusiness conglomerate. An important bound-
( = 0.10; t-value = 0.91; ns). The results also ary condition on the results is that they represent
failed to support the mediating relationship through knowledge-intensive projects in which a variety
knowledge integration proposed in Hypothesis 3b. of specialized tacit knowledge spanning the areas
(Note that the main effect terms cannot be inter- of technology, business, and application domain
preted in the subsequent step in the presence of is required. They also represent a finite universe
the interaction term because their coefficients then of project alliances whose primary intent was to
represent conditional simple effects.) generate a combinatorial (Obstfeld, 2005) inno-
Hypothesis 4a proposed that strong ties and vation by melding the unique perspectives, skills,
bridging ties in a project alliance complement experiences, and knowledge of alliance partners.
each other in enhancing knowledge integration, In contrast with prior research that has focused on
i.e., strong ties and bridging ties exhibit positive longer-term alliances, the project alliances studied
interaction effects. We therefore first created an here: (1) were designed without an assumption of
interaction term (strong ties x bridging ties). We longevity, (2) were often formed with no explicit
used Lances (1988) residual-centering technique intention of learning, and (3) fast-paced projects
to overcome distortion of the main effects due to using relatively small teams with members drawn
the tendency of main effects and interaction terms from a large network of partners (averaging six
to be highly correlated. This interaction term was months with nine project participants). Caution is
added to Model 2 in Step 2 to test Hypothesis 4a. therefore warranted in generalizing the findings to
The interaction term had a positive and signifi- alliances that focus on incremental projects or to
cant relationship with knowledge integration ( = other types of knowledge-intensive alliances such
0.23; t-value = 2.16; p < 0.05), and its addition as research and development or outsourcing (see
to the model led to a statistically significant (p < Tiwana and Keil, 2007). Third, although some of
0.001) increase in explained variance, supporting the participants in our sample were software com-
Hypothesis 4a. We also proposed in Hypothe- ponent providers, we did not control for whether
sis 4b that this complementarity between strong each alliance partner had a longer-term stake in
and bridging ties influences alliance ambidexterity the outcome of a given project or merely provided
primarily because it enhances knowledge integra- component technology. There is also considerable
tion, i.e., the relationship is fully mediated. The potential for extending this research by explicitly
relationship between the interaction term and the considering the type of innovation sought by a
mediator was positive and significant as just dis- project alliance (such as modular or architectural).
cussed; the relationship between the mediator and Fourth, although it is plausible that the control
alliance ambidexterity was also positive and sig- variables could also be used to explain variance
nificant ( = 0.709; t-value = 6.32; p < 0.001) as in the mediator, there is little prior empirical basis
previously described. Furthermore, the direct effect for including them in Model 2. Furthermore, the
Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
Do Bridging Ties Complement Strong Ties? 267

addition of the predictors and knowledge integra- were intended to generate innovative technologi-
tion to Model 1 does not change the significance of cal solutions. An average score of 3.3 on a five-
any control variables, suggesting a negligible risk point single-item perceived novelty scale lends
of model underspecification. Caution is also war- some support for this plausibility. So it is possible
ranted in interpreting the knowledge integration that this construct did not have sufficient vari-
to alliance ambidexterity relationship, which was ance purely as an artifact of our sampling strategy.
the only relationship that was significant using the Another plausible explanation is that bridging ties
more appropriate one-tailed test but not with the might influence knowledge integration in a pat-
two-tailed test. Finally, following recent theoreti- tern wherein the difficulty of integrating diverse
cal conceptualizations of bridging ties, we assessed expertise overwhelms its benefits until a threshold,
this characteristic of the project alliance tie port- past which the benefits outweigh the costs. How-
folio by tapping into the diversity of knowledge, ever, our post-hoc tests failed to show a U-shaped
backgrounds, perspectives, and capabilities of the relationship and therefore do not support this inter-
members of each project team. This is an adequate pretation. Third, evidence of a positive interac-
but imperfect proxy for bridging ties. Future work tion effect strongly supports the idea that strong
should attempt to more precisely measure this tie ties complement bridging ties and that their influ-
portfolio characteristic. ence on alliance ambidexterity is fully mediated by
knowledge integration. These findings have con-
siderable theoretical implications for research on
social network configuration, strategic alliances,
DISCUSSION AND IMPLICATIONS and for the broader strategy literature, as discussed
The objective of this study was to examine the
tensions and complementarities between structural Implications for the social network
hole-bridging ties and strong ties in influenc- configuration literature
ing ambidexterity in innovation-seeking project The primary contributions of this study are to
alliances. We theorized knowledge integration as an emerging line of research that recognizes that
a central explanatory mechanism through which the optimal network configurations combine seem-
such tie portfolio characteristics directly and inter- ingly conflicting elements of both cohesion and
actively influence alliance ambidexterity at the range, proximity and diversity, strong and weak
project level. Data from 42 such alliances were ties, and cohesion and structural holes (Levin and
used to conduct one of the first systematic empir- Cross, 2004; Obstfeld, 2005; Regans and McEvily,
ical tests of these ideas. 2003; Regans and Zuckerman, 2001). The results
There are three major sets of results, with complement recent works such as Levin and Cross
considerable theoretical implications for network (2004) and Obstfeld (2005) who recognize the
structure configuration and strategic alliances need for network structures to promote creativity
research. The results are summarized in Figure 3. and innovation on the one hand and cooperation
First, we found a positive and significant relation- and coordination on the other. A shift in focus
ship between strong ties and knowledge integra- from dyadic ties to portfolios of ties at the project
tion, as hypothesized. Their influence on alliance level allowed us to examine configurations where
ambidexterity was fully mediated by knowledge tie strength and the extent to which a tie portfo-
integration. Second, although we expected a nega- lio bridged structural holes could both be more
tive relationship between bridging ties and knowl- readily observed. The results show that strong ties
edge integration, the results do not support that and bridging ties are not necessarily at odds, as
assertion. One explanation given the novel nature previously assumed in the literature on social net-
of the projects in our sample is that the value of works (Coleman, 1988) and structural holes (Burt,
access to a broader repertoire of capabilities off- 1992). An optimal configuration meshes elements
set the difficulty of integrating such variety of of both bridging ties and strong ties in a tie port-
knowledge and skills. Unlike prior studies that folio. On the one hand, heterogeneous expertise
have focused on incremental innovation tasks (e.g., made accessible by bridging ties creates the poten-
Obstfeld, 2005), the project alliances in our study tial for novel recombinations of knowledge and
Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
268 A. Tiwana

skills, while on the other, trust, reciprocity, and the alliance as the unit of analysis (Gerwin and Fer-
close interaction must be nurtured to translate them ris, 2004). Although that focus is entirely appropri-
into alliance ambidexterity. In other words, strong ate for many routine interfirm partnerships, ongo-
ties help integrate the diverse knowledge and capa- ing sourcing relationships, and incremental work,
bilities made accessible by bridging ties. This mix a focus on projects can help unmask hitherto-
of strong and bridging ties in a tie portfolio can overlooked subtleties in knowledge-intensive mul-
be viewed as the ideal configuration of ties in tifirm alliances. Examples of the latter are few
Burts (1992) perspective, which is why we label and far between (e.g., Ethiraj et al., 2005; Pisano,
them Burt complements. Such complementarities 1994), and this work complements them. A distinc-
and tensions among different types of ties, while tive contribution of this study is that it potentially
widely acknowledged, have rarely been directly informs the selection of an appropriate mix of
examined. This finding is also consistent with and team members to participate in multifirm project
refines Levin and Crosss (2004) observation that alliances. Such team member selection contin-
trusted weak ties are often a source of novel knowl- ues to be a challenge facing contemporary man-
edge that is also perceived as being useful. It agers (Regans et al., 2004). A second contribu-
also provides new insights into the proximate con- tion to the alliances literature is the full medi-
cept of network range, which has received mixed ation of the effect of strong ties by knowledge
support in empirical studies because it provides integration. This finding lends empirical support
informational benefits but impedes coordination to Kale et al.s (2000) speculation that the qual-
(Regans et al., 2004). While Regans and McEvily ity of the relationships among alliance partners
(2003) found that network range is weakly but pos- enables realization of the opportunities for quasi-
itively associated with ease of knowledge transfer, internalization of know-how and technological
our results point to the subtlety that the benefits of capabilities.
bridging ties that span structural holes (which is
comparable to their notion of network range) are Broader implications for strategy theory
more readily realized when they are complemented
by strong ties in an alliance tie portfolio. These results also have three broader implications
A second, more distinctive contribution is the for the strategy literature. First, the complemen-
theoretical development of the role of knowl- tarities between strong ties and bridging ties and
edge integration as a mechanism that mediates their joint effect on knowledge integration fur-
the effects of alliance tie portfolio characteris- thers recent work that has emphasized the need
tics on alliance ambidexterity. The results showed for knowledge integration in new product devel-
that knowledge integration fully mediates the influ- opment teams (Carlile, 2004). Since new product
ence of the complementarities between strong ties development is increasingly done through inter-
and bridging ties on alliance ambidexterity in firm, outsourcing alliances (Ethiraj et al., 2005;
our nomological network. A significant mediat- Oxley and Sampson, 2004; Tiwana and Keil,
ing role of knowledge integration illustrates that 2007), these results provide new insights into
strong ties facilitate integration of a broad reper- how such projects can be better organized to
toire of specialized knowledge from collaborat- achieve knowledge integration. Second, the con-
ing alliance partners at the project level, which ceptualization of tie portfolios as an antecedent
in turn influences the successful realization of an to knowledge integration complements the bur-
innovation. The knowledge integration perspec- geoning knowledge-based view of the firm (Grant,
tive therefore offers a theoretical explanation for 1996b; Tsoukas, 1996). While this theoretical per-
how alliance tie characteristics influence alliance spective has focused largely on firms as the unit
ambidexterity. of analysis, this study adds to a growing body of
work that recognizes the importance of knowl-
edge integration across firm boundaries (Grant
Implications for research on strategic alliances and Baden-Fuller, 2004; Oxley and Sampson,
2004). Finally, this study contributes to a grow-
Although much innovation generation in multifirm ing body of work that emphasizes the need to
strategic alliances occurs at the project level, the manage centrifugal tensions and presumed para-
majority of prior alliances studies have focused on doxes (Gibson and Birkinshaw, 2004; Goold and
Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
Do Bridging Ties Complement Strong Ties? 269

Quinn, 1990; Sundaramurthy and Lewis, 2003). Baron R, Kenny D. 1986. The moderator-mediator
It illustrates how notions such as strong ties variable distinction in social psychological research:
conceptual, strategic, and statistical considerations.
and structural holes that appear to be in tension Journal of Personality and Social Psychology 51(6):
when viewed independently can indeed benefi- 11731182.
cially coexist. Bhattacharya S, Krishnan V, Mahajan V. 1998. Manag-
For managers, the results demonstrate that the ing new product definition in highly dynamic envi-
thoughtful formation of an innovation-seeking ronments. Management Science 44(11): 5064.
Burt R. 1992. Structural Holes: The Social Structure of
project alliance can at best assure them of its Competition. Harvard University Press: Boston, MA.
potential. The capacity of an interfirm project Burt R. 1997. The contingent value of social capital.
alliance to successfully realize that potential, how- Administrative Science Quarterly 42: 339365.
ever, is influenced by the combinatory characteris- Burt R. 2004. Structural holes and good ideas. American
tics of the portfolio of ties among project partici- Journal of Sociology 110(2): 349399.
Campion C, Medsker G, Higgs A. 1993. Relations
pants. As product and service innovation becomes between work group characteristics and effectiveness:
an increasingly important firm capability (Eisen- implications for designing effective work groups.
hardt and Martin, 2000)and one that increas- Personnel Psychology 46(3): 823855.
ingly spans historic firm, industry, and geographi- Carlile P. 2002. A pragmatic view of knowledge
cal boundariesmanagers cannot afford to neglect and boundaries: boundary objects in new product
development. Organization Science 13(4): 442455.
these tensions and complementarities.
Carlile P. 2004. Transferring, translating, and transform-
ing: an integrative framework for managing knowl-
edge across boundaries. Organization Science 15(5):
Carlile P, Rebentisch E. 2003. Into the black box: the
knowledge transformation cycle. Management Science
This paper has benefited from inputs from Ash- 49(9): 11801195.
ley Bush, David Kenny, Ikujiro Nonaka, Gabriel Child J, McGrath R. 2001. Organizations unfettered:
Szulanski, Rich Makadok, Sridhar Ramaswami, organizational form in an information-intensive
Anandhi Bharadwaj, and Karl Weick. Develop- economy. Academy of Management Journal 44(6):
mental feedback from the editor, Ed Zajac and Cohen W, Levinthal D. 1990. Absorptive capacity: a new
the anonymous SMJ reviewers is also gratefully perspective on learning and innovation. Administrative
acknowledged. Science Quarterly 35(1): 128152.
Coleman J. 1988. Social capital in the creation of human
capital. American Journal of Sociology 94: 95120.
Darr E, Argote L, Epple D. 1995. The acquisition,
REFERENCES transfer, and depreciation of knowledge in service
organizations: productivity in franchises. Management
Adler PS. 1989. When knowledge is the critical resource, Science 41(11): 17501762.
knowledge management is the critical task. IEEE Dougherty D. 1992. Interpretive barriers to successful
Transactions on Engineering Management 36(2): product innovation in large firms. Organization
8794. Science 3(2): 179202.
Ahuja G. 2000. Collaboration networks, structural holes, Dyer JH, Nobeoka K. 2000. Creating and managing
and innovation: a longitudinal study. Administrative a high-performance knowledge-sharing network: the
Science Quarterly 45: 425455. Toyota case. Strategic Management Journal , March
Amit R, Schoemaker PJH. 1993. Strategic assets and Special Issue 21: 345367.
organizational rent. Strategic Management Journal Dyer JH, Singh H. 1998. The relational view: cooperative
14(1): 3346. strategy and sources of interorganizational competitive
Amit R, Zott C. 2001. Value creation in e-business. advantage. Academy of Management Review 23(4):
Strategic Management Journal , June-July Special 660679.
Issue 22: 493520. Eisenhardt KM, Martin JA. 2000. Dynamic capabilities:
Ancona DG, Caldwell DF. 1992. Demography and what are they? Strategic Management Journal ,
design: predictors of new product team performance. October-November Special Issue 21: 11051121.
Organization Science 3(3): 321341. Ethiraj SK, Kale P, Krishnan MS, Singh JV. 2005.
Argote L, McEvily B, Reagans R. 2003. Managing Where do capabilities come from and how do they
knowledge in organizations: an integrative framework matter? A study in the software services industry.
and review of emerging themes. Management Science Strategic Management Journal 26(1): 2545.
49(4): 571582. Faraj S, Sproull L. 2000. Coordinating expertise in
Baker W. 1990. Market networks and corporate behavior. software development teams. Management Science
American Journal of Sociology 96: 589625. 46(12): 15541568.
Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
270 A. Tiwana

Garud R, Nayyar PR. 1994. Transformative capac- Kale P, Singh H, Perlmutter H. 2000. Learning and
ity: continual structuring by intertemporal technol- protection of proprietary assets in strategic alliances:
ogy transfer. Strategic Management Journal 15(5): building relational capital. Strategic Management
365385. Journal , March Special Issue 21: 217237.
Gerwin D, Ferris J. 2004. Organizing new product devel- Kenny D, Kashy D, Bolger N. 1998. Data analysis
opment projects in strategic alliances. Organization in social psychology. In The Handbook of Social
Science 15(1): 2237. Psychology, Gilbert D, Fiske S, Lindzey G (eds).
Gibson C, Birkinshaw J. 2004. The antecedents, con- McGraw-Hill: Boston, MA; 233265.
sequences, and mediating role of organizational Kogut B, Zander U. 1992. Knowledge of the firm,
ambidexterity. Academy of Management Journal combinative capabilities, and the replication of
47(2): 209226. technology. Organization Science 3: 383397.
Goold M, Quinn JJ. 1990. The paradox of strategic Lance C. 1988. Residual centering, exploratory and
controls. Strategic Management Journal 11(1): confirmatory moderator analysis, and decomposition
4357. of effects in path models containing interactions.
Granovetter M. 1973. The strength of weak ties. Ameri- Applied Psychological Measurement 12(2): 163175.
can Journal of Sociology 78: (May): 13601380. Lapre M, Wassenhove L. 2001. Creating and transferring
Grant R. 1996a. Prospering in dynamically-competitive knowledge for productivity improvement in factories.
environments: organizational capability as knowledge Management Science 47(10): 13111325.
integration. Organization Science 7(4): 375387. Larsson R, Bengtsson L, Henriksson K, Sparks J. 1998.
Grant R. 1996b. Toward a knowledge-based theory of the The interorganizational learning dilemma: collective
firm. Strategic Management Journal , Winter Special knowledge development in strategic alliances. Orga-
Issue 17: 109122. nization Science 9(3): 285305.
Grant R, Baden-Fuller C. 2004. A knowledge accessing Levin D, Cross R. 2004. The strength of weak ties you
theory of strategic alliances. Journal of Management can trust: the mediating role of trust in effective
Studies 41(1): 6184. knowledge transfer. Management Science 50(11):
Gulati R, Singh H. 1998. The architecture of cooperation: 14771490.
managing coordination costs and appropriation Lovelace K, Shapiro D, Weingart L. 2001. Maximizing
concerns in strategic alliances. Administrative Science cross-functional new product teams innovativeness
Quarterly 43: 781814. and constraint adherence: a conflict communications
Hansen M. 1999. The search-transfer problem: the role perspective. Academy of Management Journal 44(4):
of weak ties in sharing knowledge across organiza- 779793.
tional subunits. Administrative Science Quarterly 44: MacKinnon D, Lockwood C, Hoffman J, West S,
83111. Sheets V. 2002. A comparison of methods to test
Henderson R, Clark K. 1990. Architectural innovation: mediation and other intervening variable effects.
the reconfiguration of existing product technologies Psychological Methods 7(1): 83104.
and the failure of established firms. Administrative McEvily B, Marcus A. 2005. Embedded ties and the
Science Quarterly 35: 930. acquisition of competitive capabilities. Strategic
Henderson R, Cockburn I. 1994. Measuring competence? Management Journal 26(11): 10331055.
Exploring firm effects in pharmaceutical research. McEvily B, Zaheer A. 1999. Bridging ties: a source
Strategic Management Journal , Winter Special Issue of firm heterogeneity in competitive capabilities.
15: 6384. Strategic Management Journal 20(12): 11331156.
Hollenbeck J, Ilgen D, Sego D, Hedlund J. 1995. Multi- Milgrom P, Roberts J. 1995. Complementarities and
level theory of team decision making: decision per- fit: strategy, structure, and organizational change in
formance in teams incorporating distributed expertise. manufacturing. Journal of Accounting and Economics
Journal of Applied Psychology 80(2): 292317. 19(2/3): 179208.
Hoopes DG, Postrel S. 1999. Shared knowledge, Molm L, Peterson G, Takashaki N. 1999. Power in nego-
glitches, and product development performance. tiated and reciprocal exchange. American Sociological
Strategic Management Journal 20(9): 837865. Review 64(6): 876890.
Hulland J. 1999. Use of partial least squares (pls) in Mowery D, Oxley JE, Silverman BS. 1996. Strategic
strategic management research: a review of four alliances and interfirm knowledge transfer. Strategic
recent studies. Strategic Management Journal 20(2): Management Journal , Winter Special Issue 17:
195204. 7791.
Ingram P, Roberts P. 2000. Friendships among competi- Nahapeit J, Ghoshal S. 1998. Social capital, intellectual
tors in the Sydney hotel industry. American Journal capital, and the organizational advantage. Academy of
of Sociology 106: 387423. Management Review 23(2): 242266.
James L, Demaree R, Wolf G. 1993. Rwg : an assessment Nickerson JA, Zenger TR. 2004. A knowledge-based
of within-group interrater agreement. Journal of theory of the firmthe problem-solving perspective.
Applied Psychology 78(2): 306309. Organization Science 15(6): 617632.
Jarvis C, Mackenzie S, Podsakoff P. 2003. A critical Nidumolu S. 1995. The effect of coordination and uncer-
review of construct indicators and measurement model tainty on software project performance: residual per-
misspecification in marketing and consumer research. formance risk as an intervening variable. Information
Journal of Consumer Research 30(2): 199218. Systems Research 6(3): 191219.
Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
Do Bridging Ties Complement Strong Ties? 271
Nonaka I. 1994. A dynamic theory of organizational Spender J-C, Grant RM. 1996. Knowledge and the firm:
knowledge creation. Organization Science 5: 1437. overview. Strategic Management Journal , Winter
Nunally J. 1978. Psychometric Theory. McGraw-Hill: Special Issue 17: 59.
New York. Srinivasan R, Lilien G, Rangaswamy A. 2002. Techno-
Obstfeld D. 2005. Social networks, the tertius lungens logical opportunism and radical technology adoption:
orientation, and involvement in innovation. Adminis- an application to e-business. Journal of Marketing
trative Science Quarterly 50(1): 100130. 66(3): 4760.
Okhuysen G, Eisenhardt K. 2002. Integrating knowledge Sternberg R. 2000. Understanding practical intelligence:
in groups: how formal interventions enable flexibility. the role of tacit knowledge. In Practical Intelligence
Organization Science 13(4): 370386. on Everyday Life, Sternberg R (ed). Cambridge
Oxley JE, Sampson RC. 2004. The scope and gover- University Press: New York; 104143.
nance of international R&D alliances. Strategic Man- Stuart T. 1998. Network positions and propensities to
agement Journal , August-September Special Issue 25: collaborate: an investigation of strategic alliance for-
723749. mation in a high-technology industry. Administrative
Patnayakuni R, Rai A, Tiwana A. 2007. Systems devel- Science Quarterly 43(3): 668698.
opment process improvement: a knowledge integration Sundaramurthy C, Lewis M. 2003. Control and collabo-
perspective. IEEE Transactions on Engineering Man- ration: paradoxes of governance. Academy of Manage-
agement 53(2): 286300. ment Review 28(3): 397415.
Pelled LH. 1996. Demographic diversity, conflict, and Szulanski G. 1996. Exploring internal stickiness: imped-
work group outcomes: an intervening process theory. iments to the transfer of best practice within the firm.
Organization Science 7(6): 615631. Strategic Management Journal , Winter Special Issue
Pisano GP. 1994. Knowledge, integration, and the 17: 2743.
locus of learning: an empirical analysis of process Tiwana A, Keil M. 2007. Does peripheral knowledge
development. Strategic Management Journal , Winter complement control? An empirical test in technology
Special Issue 15: 85100. outsourcing alliances. Strategic Management Journal
Poppo L, Zenger T. 1998. Testing alternative theories of 28(6): 623634.
the firm: transaction cost, knowledge-based, and mea- Tiwana A, McLean ER. 2005. Expertise integration
surement explanations for make-or-buy decisions in and creativity in information systems development.
information services. Strategic Management Journal Journal of Management Information Systems 22(1):
19(9): 853877. 1343.
Powell W, Koput K, Smith-Doerr L. 1996. Interorgani- Tsai W. 2001. Knowledge transfer in intraorganizational
zational collaboration and the locus of innovation: networks: effects of network position and absorptive
networks of learning in biotechnology. Administrative capacity on business unit innovation and performance.
Science Quarterly 41: 116145. Academy of Management Journal 44(5): 9961004.
Regans R, McEvily B. 2003. Network structure and Tsoukas H. 1996. The firm as a distributed knowledge
knowledge transfer: the effects of cohesion and range. system: a constructionist approach. Strategic Manage-
Administrative Science Quarterly 48(2): 240267. ment Journal , Winter Special Issue 17: 1125.
Regans R, Zuckerman E. 2001. Networks, diversity, and Uzzi B. 1996. The sources and consequences of embed-
productivity: the social capital of corporate R&D dedness for the economic performance of organi-
groups. Organization Science 12(4): 502517. zations: the network effect. American Sociological
Regans R, Zuckerman E, McEvily B. 2004. How to make Review 61: 674698.
the team: social networks vs. demography as criteria Uzzi B. 1997. Social structure and competition in
for designing effective teams. Administrative Science interfirm networks. Administrative Science Quarterly
Quarterly 49(1): 101133. 42: 3567.
Robillard P. 1999. The role of knowledge in software Verona G. 1999. A resource-based view of product
development. Communications of the ACM 42(1): development. Academy of Management Review 24(1):
8792. 132142.
Rodan S, Galunic C. 2004. More than network structure: Yli-Renko H, Autio E, Sapienza HJ. 2001. Social capital,
how knowledge heterogeneity influences managerial knowledge acquisition, and knowledge exploitation in
performance and innovativeness. Strategic Manage- young technology-based firms. Strategic Management
ment Journal 25(6): 541562. Journal , June-July Special Issue 22: 587613.
Rus I, Lindvall M. 2002. Knowledge management in Yourdon E. 2002. Managing High-Intensity Internet
software engineering. IEEE Software 19(3): 2638. Projects. Prentice Hall: Upper Saddle River, NJ.
Sabherwal R, Becerra-Fernandez I. 2005. Integrating Zaheer A, Bell G. 2005. Benefiting from network
specific knowledge: insights from the Kennedy position: firm capabilities, structural holes, and
space center. IEEE Transactions on Engineering performance. Strategic Management Journal 26(9):
Management 52(3): 301314. 809825.
Simonin BL. 1999. Ambiguity and the process of Zahra S, George G. 2002. Absorptive capacity: a
knowledge transfer in strategic alliances. Strategic review, reconceptualization, and extension. Academy
Management Journal 20(7): 595623. of Management Review 27(2): 185203.
Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj
272 A. Tiwana

APPENDIX: SCALES ITEMS FOR THE Knowledge integration ( = 0.84 aggregated;

PRINCIPAL VARIABLES IN THE STUDY 0.91 nonaggregated; interrater agreement 0.64):
Members of this team: (1) competently blend new
Alliance Ambidexterity was computed as the prod- project-related knowledge with what they already
uct of alignment and adaptation, following Gib- know, (2) span several areas of expertise to
son and Birkinshaws (2004) conceptualization of develop shared project concepts, (3) synthesize
organizational subunit ambidexterity. Alignment and integrate their individual expertise at the
dimension ( = 0.85 aggregated; 0.87 nonaggre- project level.
gated): In light of marketplace-mandated changes Bridging ties ( = 0.90 aggregated; 0.96 nonag-
and new business requirements that arose dur- gregated; interrater agreement 0.87): (1) Members
ing project execution, at the present time, this of this team: (1) vary widely in their areas of
project: (1) is within budget , (2) is on sched- expertise, (2) have a variety of different back-
ule, (3) delivers ALL [emphasis added] desir- grounds and experiences, (3) have skills and abil-
able features and functionality , (4) meets key ities that complement each others.
project objectives and business needs, 5) over- Strong ties ( = 0.93 aggregated; 0.92 nonag-
all, is very successful. Adaptation dimension ( = gregated; interrater agreement 0.73): (1) There
0.86 aggregated; 0.90 nonaggregated): Overall, is close, personal interaction among team mem-
this team has been able to: (1) manage successfully bers at multiple levels; at multiple levels, this
scope changes, (2) resolve unexpected problems, project team is characterized by: (2) high reci-
(3) deliver a relatively stable system to current procity among members, (3) mutual trust among
requirements. members, (4) mutual respect among members,
(5) personal friendship between members.

Scale anchors: 1 = Strongly disagree; 5 = Strongly agree

Item dropped during scale purification using exploratory
factor analyses.
Nonaggregated refers to individual responses; aggregated
represents reliability after aggregation of responses to the project
alliance level.

Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 251272 (2008)
DOI: 10.1002/smj