Network Control Effects on Trust: Aggregating Evidence from Multiple Studies

In earlier work, we have developed a typology of effects of networks of social relations on trust problems along two dimensions. First, we distinguish the embeddedness of a trust problem in the dyadic trustor-trustee relation from the embeddedness in networked interactions of the trustor and trustee with third parties. Second, we distinguish two mechanisms affecting trust: learning and control. Empirical evidence consistently confirms hypotheses on dyadic learning and control effects as well as on network learning effects. This includes evidence from studies employing complementary designs such as experiments, surveys, and vignette studies. On the other hand, the empirical evidence on network control effects is ambiguous. This is an empirical puzzle for research on trust and social networks. This paper assesses the evidence for network control effects on trust and cooperation by reanalyzing the data from heterogeneous experimental studies. We attempt to include all available experimental studies suitable for studying network control effects in two-person dilemma games. To tackle this problem, we employ a novel method called Bayesian Evidence Synthesis. The aggregated empirical evidence provides strong support for network control hypotheses. This evidence supports these hypotheses, especially in studies where network relations are implemented without repeated dyadic interactions. In the studies with network relations implemented in the presence of repeated dyadic interactions, the support is weaker and less consistent. The paper ends with a discussion of our findings and their implications for research on trust and social networks, addressing further theoretical, empirical, and methodological work.