Network Lift from Dual Alters: Extended Opportunity Structures from a Multilevel and Structural Perspective

Abstract

This paper uses multilevel network analysis to identify an extended and latent opportunity structure for actors dually positioned in both intra-organizational and inter-organizational networks. This extended opportunity structure combines actors' direct ties with indirect ties that they can add to their own network by 'borrowing' some of their boss's two-path contacts. We call 'dual alters' contacts that can be reached through this multilevel path with help (or absence of obstruction) from such 'embedded brokers'. We test the specific effect of this extension on actors' performance using a dataset derived from a multilevel study of the elite of French cancer researchers (1996-2005). We find a significant effect of this extension on members' performance when dual alters provide complementary resources, thus providing proof of a 'network lift from dual alters' presence in the focal actor's network. Network lift allows sociologists to measure the extent to which performance measured at the individual level depends in a complex way on the multilevel and combined characteristics of the intraand inter-organizational context in which individuals belong. We believe that this measurement of latent and extended opportunity structures will help meso-level sociologists in their approach to social processes and inequalities in the organizational society.

Social selection models for multilevel networks

“…For our empirical example, we use the same two-level network data as in Lazega et al (2008Lazega et al ( , 2013 and Wang et al (2013) about French cancer researchers and their laboratories. The data comprises a directed researcher advice seeking network at level B, and a directed laboratory collaboration network at level A.…”

“…The cross level four-cycle configurations generally represent a cross-level alignment effect where ties from the two different levels are aligned based on the crosslevel affiliations of the nodes involved (see "tetradic" configurations in Lazega et al, 2013). By adding a tie (either within-level or cross-level) to the degree assortativity configuration, we can derive the four-cycle configurations of various forms.…”

“…Using the same data set as in Lazega et al (2008Lazega et al ( , 2013, Wang et al (2013) demonstrated the interdependent nature of ties among the three networks, and highlighted how network ties at one level may affect the structure of the network at the other level. However, attributes of the researchers and of the laboratories were not taken into account.…”

“…Note that we use the term level here because that is its common usage in empirical network context; the general data structure need not be construed as at different levels, but it does require two separate sets of nodes. Lazega et al (2008Lazega et al ( , 2013 developed theories and mechanisms of multilevel networks in organizational contexts. Using two-level networks of elite 1 French cancer researchers and their affiliated research laboratories, Lazega et al (2008Lazega et al ( , 2013 proposed hypotheses on how multilevel network structure and various nodal attributes may affect or contribute to an individual's performance.…”

“…Lazega et al (2008Lazega et al ( , 2013 developed theories and mechanisms of multilevel networks in organizational contexts. Using two-level networks of elite 1 French cancer researchers and their affiliated research laboratories, Lazega et al (2008Lazega et al ( , 2013 proposed hypotheses on how multilevel network structure and various nodal attributes may affect or contribute to an individual's performance. These analyses used within-level network centralities and various nodal attributes to classify nodes.…”