The Paradox of Network Intervention: Exploring the Consequences of Policies for Promoting Connection

Abstract

Social networks are everywhere, but until recently, the particularities of the structures were seldom remarked upon outside academia. Today, however, policymakers and managers are becoming more aware both of how networks are structured and how those structures create opportunities and constraints. Consequently, network interventions—deliberate efforts to shape a network configuration—are increasingly common. Nevertheless, researchers have devoted little systematic attention to them. We take a first step towards building systematic understanding of network interventions by examining the relationship between the structure of existing networks in targeted groups and responsiveness to network interventions. Specifically, we theorize that groups with greater fragmentation are less likely to engage with incentive-based network interventions and conditional on engaging, the interventions are less likely to be effective at transforming the network. To evaluate these ideas, we conducted a study of accountable care organizations. Our approach leveraged data on 100 million patient sharing relationships millions of providers in 3,000 communities across the United States. Additionally, we used several instruments to better identify the proposed relationships, including the Medicaid expansion and local political environments. Our results reveal a network intervention paradox—groups that may benefit most from such interventions appear least equipped to respond positively to them.