Organizational learning across multi-level networks
Introduction
Organizational learning, the process by which organizations generate, disseminate and exploit knowledge, translating it into innovation (March and Simon, 1958, Cyert and March, 1963, March, 1991) is a key topic in organizational studies.
An extensive literature has demonstrated that learning is a never-ending process, which contributes significantly to organizational growth, performance and survival (March, 1996). Because of these benefits, significant attention has been devoted to understand how learning occurs (Argote et al., 2003). A crucial mechanism consists in learning from the experience of others, either within or across the organizational boundaries. Learning from others requires some form of knowledge transfer, which is made possible mainly by interpersonal interaction among organizational members (Krackhardt and Hanson, 1993, Tushman, 1977). Interpersonal interaction, in fact, allows people to search for (i.e., ‘look and identify’) knowledge available in some parts of the organization and to transfer (‘move and incorporate’) it to other parts (Hansen, 1999, p. 83).
In examining knowledge transfer within the organization – which is the focus of this paper – the capability of interpersonal relations to connect different units, divisions and departments has been particularly investigated. Because organizational units, divisions and departments are pools of homogenous knowledge (Reagans and McEvily, 2003, Tortoriello and Krackhardt, 2010), searching in different units across organizational boundaries increases the heterogeneity of knowledge available and promotes learning (Beckman and Haunschild, 2002).
Since people are generally reluctant to cross-cut the boundaries defined around their units, several papers have attempted to identify how boundary spanning can be facilitated (Argote and Epple, 1990, Tortoriello et al., 2012). These studies have focused mostly on characteristics of the people as well as on types and structures of the relations among them (Burt, 2004, Dokko et al., 2014), disregarding the context in which relations occur.
This paper assesses whether and how the characteristics of the organization, and of the organizational structure in particular, can affect the presence of boundary-spanning ties. We examine the effect of the existence of work-flow ties connecting the units among which people are expected to search and transfer knowledge. The purpose of such an investigation is to provide a better understanding of the extent to which boundary-spanning relations of informal knowledge search-transfer at interpersonal level can be sustained by the formal work-flow ties between organizational units (Reagans and McEvily, 2003).
We emphasize the benefits of addressing these purposes by conceiving organizations as hierarchical systems of nested relations – i.e., multilevel network systems. We show how such an approach would allow a better representation of the interdependences between formal and informal relations and a clear assessment of the role of both.
We provide empirical evidence on this claim specifying and estimating newly developed Multilevel Exponential Random Graph Models (MERGMs – Wang et al., 2013, Wang et al., 2015). They represent a significant improvement on previous multilevel methods that are unable to analyze hierarchical systems of nested relations in depth. Standard Hierarchical Linear Models (HLMs) account for individual membership to a unit only. Assuming that either lower- (i.e., individual) or higher- (i.e., unit) level actors are independent from one another, conditional on the nesting structure of individuals within units, HLMs do not fit the structure of network then, do HLMs allow modeling interdependences among actors either within levels (i.e., interpersonal and interunit network separately) or between levels (i.e., associations and overlaps between the two network structures). The structural linked design (Lazega et al., 2008), popular in network studies, relaxes some assumptions of HLMs and, therefore, addresses some of their limitations. This approach respects the multi-level nature of the data, but models only some kinds of interdependences between the ties at the two levels.
In the empirical part of the paper, we specify and estimate MERGMs on original relational data that we have collected in a multiunit division of a regional government institution based in Northern Italy. Comparing the results of MERGM estimations with those of simpler ERGMs (Robins et al., 2007, Snijders et al., 2006), we show the advantages of addressing organizational learning through the multilevel network lens.
Section snippets
Organizational learning as interpersonal knowledge sharing
Informal interpersonal networks are one of the main conduits through which knowledge flows within the organization (Krackhardt and Hanson, 1993). Informal ties of advice seeking and knowledge searching, in particular, allow individuals to have access to knowledge accumulated by close contacts (Reagans and Zuckerman, 2001, Zaheer and McEvily, 1999), either inside or outside the organization (Argote et al., 2003).
Indeed, informal interpersonal networks can provide access to heterogeneous others (
Representing multilevel network data
Multilevel Exponential Random Graph Models (Wang et al., 2013, Wang et al., 2015) are the only existing method which directly assesses network interdependences across levels. MERGMs are a new class of ERGMs for multilevel network data.
Let M = [A,X,B] denote the network variable for a (u,v) two-level network, and m = [a,x,b] the corresponding realizations. M consists of a network A = [Ahu] representing a relation among a set U of higher-level actors with h, u nodes in U; a network B = [Biv] representing
Results
We specified the effects in an increasing order of complexity (Table 4). Model 1 is the baseline model, Model 2 is the intermediate model, and Model 3 is the multilevel complete model. We comment on the latter, drawing attention to its ability to either improve or simplify the representation of knowledge transfer.
First, we look at the interpersonal network. The positive Reciprocity estimate indicates that knowledge seeking is likely to be a mutual relation, based on knowledge sharing (McFadyen
Discussion and conclusions
Studies on organizational learning have long emphasized the role of interpersonal ties spanning unit boundaries as the main conduit through which knowledge flows within the organization. Although recognizing the hierarchical nature of organizations, these studies have treated organizational units as independent, and investigated relational and individual characteristics that allow organizational members to cross-cut boundaries. In this paper we have taken a different perspective, focusing on
References (75)
- et al.
Knowledge transfer: a basis for competitive advantage in firms
Organ. Behav. Hum. Decis. Process.
(2000) - et al.
The network paradigm in organizational research: a review and typology
J. Manage.
(2003) - et al.
Knowing what we know: supporting knowledge creation and sharing in social networks
Organ. Dyn.
(2001) Advances in exponential random graph (p*) models applied to a large social network
Soc. Netw.
(2007)Effects of missing data in social networks
Soc. Netw.
(2006)- et al.
Catching up with big fish in the big pond? Multi-level network analysis through linked design
Soc. Netw.
(2008) - et al.
An introduction to exponential random graph (p*) models for social networks
Soc. Netw.
(2007) - et al.
Closure, connectivity and degree distributions: exponential random graph (p*) models for directed social networks
Soc. Netw.
(2009) - et al.
Exponential random graph models for multilevel networks
Soc. Netw.
(2013) - et al.
Learning curves in manufacturing
Science
(1990)
Managing knowledge in organizations: an integrative framework and review of emerging themes
Manage. Sci.
Emergence of scaling in random networks
Science
An evolutionary model of organizational performance
Strateg. Manage. J.
Interorganizational learning and network organization: toward a behavioral theory of the interfirm. The economics of choice, change, and organization: essays in memory of Richard M.
Cyert
Generalized exchange 1
Am. J. Sociol.
Network learning: the effects of partners’ heterogeneity of experience on corporate acquisitions
Adm. Sci. Q.
Friends or strangers? Firm-specific uncertainty, market uncertainty, and network partner selection
Organ. Sci.
A relational view of information seeking and learning in social networks
Manage. Sci.
Taking stock of networks and organizations: a multilevel perspective
Acad. Manage. J.
The social structure of competition
Structural holes and good ideas
Am. J. Sociol.
Concentration and specialization: dynamics of niche width in populations of organizations
Am. J. Sociol.
Absorptive capacity: a new perspective on learning and innovation
Adm. Sci. Q.
Testing multitheoretical, multilevel hypotheses about organizational networks: an analytic framework and empirical example
Acad. Manage. Rev.
Organizational innovation: a meta-analysis of effects of determinants and moderators
Acad. Manage. J.
One of us or one of my friends: how social identity and tie strength shape the creative generativity of boundary-spanning ties
Organ. Stud.
Organizational learning curves: a method for investigating intra-plant transfer of knowledge acquired through learning by doing
Organ. Sci.
The focused organization of social ties
Am. J. Sociol.
Social Cognition
The impact of social structure on economic outcomes
J. Econ. Perspect.
The strength of weak ties
Am. J. Sociol.
Modeling social networks from sampled data
Ann. Appl. Stat.
The search-transfer problem: the role of weak ties in sharing knowledge across organization subunits
Adm. Sci. Q.
Building theoretical and empirical bridges across levels: multilevel research in management
Acad. Manage.
Inference in curved exponential family models for networks
J. Comput. Graph. Stat.
Goodness of fit of social network models
J. Am. Stat. Assoc.
Cited by (51)
Moderating influence of product diversification on the international diversification-performance relationship: A meta-analysis
2022, Journal of Business ResearchHow the organizational design of R&D units affects individual search intensity – A network study
2021, Research PolicyCitation Excerpt :Both boundary spanning – unit as recipient and boundary spanning – unit as source are positive and significant. In line with previous research (Sosa et al., 2004; Zappa and Robins, 2016), R&D unit task collaboration designating a unit as the recipient or as the source of knowledge for another unit positively influences individual boundary-spanning search from the unit's members to colleagues in the other unit. The remaining multilevel patterns accounting for the cross-level influence of unit attributes on individual search are insignificant.
Multilevel determinants of collaboration between organised criminal groups
2020, Social NetworksThe embeddedness of social relations in inter-firm competitive structures
2020, Social NetworksCitation Excerpt :This means the fixed density sampler adjusts the basic sampler by fixing the number of ties of the simulated networks equal to the number of ties in the observed network (Byshkin et al., 2016). Consequently, we excluded the baseline arc parameter for those models in which we fixed the density since the density is already defined (e.g., Zappa and Robins, 2016). This procedure assists us with reaching model fit and has been applied in several practical network studies (Brennecke and Rank, 2016; McAllister et al., 2014; Pina-Stranger and Lazega, 2011; Zappa and Robins, 2016).
Can Leadership Improve Interorganizational Collaboration? Field-Experimental Evidence From a Team-Based Leadership Training Intervention
2024, American Review of Public Administration