The Missing Links: Formation and Decay of Economic Networks

6. On Firmer Ground: The Collaborative Team as Strategic Research Site for Verifying Network-Based Social-Capital Hypotheses

Chapter 6 On Firmer Ground: The Collaborative Team as Strategic Research Site for Verifying Network-Based Social-Capital Hypotheses Ray E. Reagans, Ezra Zuckerman, and Bill McEvily Social networks command the interest of scholars and others because these relational patterns are assumed to have causal force. In particular, network theories typically adopt the premise that such patterns often lead to individual or collective outcomes that cannot be fully ascribed to the exogenous forces that determined such configurations . When this premise of network exogeneity is undermined , network analysis may be a useful tool for viewing the operation of other forces, but it cannot fulfill the network analyst’s ambition of demonstrating how social networks shape important outcomes. Moreover, network analysts must concede that network structures are always subject to manipulation by actors and can thus never be considered exogenous to the same degree as are, say, a person ’s natural endowments and fixed characteristics, such as age, gender, or innate ability. Thus, to place network theories on firmer ground, there must be strong theoretical and empirical reasons for adopting the premise of network exogeneity with confidence. In this chapter we focus on a key class of network hypotheses that is particularly vulnerable to the criticism that networks do not have causal force: claims that differential success or performance may be attributed to an actor’s position in a social network.1 Such hypotheses have gained considerable popularity in recent years and may be summarized as hypotheses regarding social capital.2 Examples of such hypotheses include the claim that individuals whose networks are better (or more poorly) constructed are more (or less) likely to achieve their goals—say, obtaining a job (Granovetter 1974/1995) or advancing through a corporate hierarchy (Burt 1992). The general difficulty that such analyses face is that, to the extent that certain network positions are more advantageous than others, all actors should be expected to strive for them. And to the extent that such efforts are made by all actors but that only some succeed, this suggests that occupancy of differentially valuable network positions reflects prior differences among actors that are responsible both for observed differences in performance and for network position. In short, the premise of network exogeneity can be undermined by “unobserved heterogeneity,” or the possibility that the variation in social capital reflects differences in intrinsic attributes ; and by “reverse causality,” or the possibility that prior performance differences (or even the prior anticipation of future performance differences) are responsible for the social capital attained. It is crucial to note that the fact that hypothesized network effects might be endogenous in a particular case does not mean that they always are. Indeed, we shall argue that there are ample theoretical reasons to expect network exogeneity in many cases. The challenge is to find nonexperimental research contexts that may serve as “strategic research sites” (Merton 1987) for testing social-capital hypotheses . The primary objective of this chapter is to make progress in meeting this challenge. In particular, we argue that contexts that include sets of overlapping collaborative teams represent such a strategic research site under certain conditions. Such teams, which may be defined as collective actors that comprise multiple individuals who are responsible for a joint product, include film projects, academic or scientific collaborations , and work groups in organizations—even organizations themselves. The hypothesis that a given team—like an individual— outperforms others because of how it is positioned in a wider social network is weakened by the possibility that the team’s network is 148 The Missing Links not causally responsible for the outcome. For instance, the observed performance differences could be due to differences in composition —for example, a team with more highly skilled members both succeeds at its task and attracts many useful ties, but the latter does not actually cause the former. Alternatively, it could be that teams that either have a history of past success or who experience success at the beginning of a project develop effective networks as a result. How can these quite reasonable objections be countered? We show below that teams can have two features that make them useful for countering such objections and thereby serving as strategic research sites for testing hypotheses about social capital. Such teams have relatively short lives and contain members with multiple and overlapping team memberships. The team’s having a short life means that it will not have a prior performance record...