Social Network Analysis

Social network analysis examines how relationships among entities (people, firms, countries) affect behavior.

With my co-authors we collected network data in Eastern DRC for 40 villages. We then used this data to optimize the diffusion of a new technology (Paper 1), to examine how social networks affect distribution behavior (Paper 2) and to look for the correlation between social preferences and social networks (Paper 3). See the abstracts for these three papers below. If you are interested in a reading a draft version of these papers shoot me an e-mail.

The Importance of First Adopters for Technology Diffusion: Evidence from a Field Experiment in the Congo (with Jennifer M. Larson, Peter van der Windt and Maarten Voors)

Introducing and distributing new technologies are essential for development. Social networks spread goods and ideas, but little is known about the consequences of selecting initial recipients from different positions within village networks. We collect full social network information in 40 rural communities in the Democratic Republic of Congo. We subsequently randomize whether initial recipients of a key agricultural technology, fertilizer, are central or isolated in their network. Recipients are trained and invited to spread fertilizer and information about fertilizer. We find that the intervention substantially increases adoption and knowledge, and that these effects taper off. Experimentally, there are no effect of recipient centrality on fertilizer use, knowledge, or willingness to pay. However, we do find important differences in the pattern of distribution. Isolate recipients are most likely to gift to central villagers. This result has important implications because low centrality is correlated with political marginalization.
This paper used to be called "Social Networks and Technology Diffusion: Evidence from a Field Experiment in the Congo"

Selection of Targets in Network Diffusion Interventions (with Martha Ross, Peter van der Windt and Maarten Voors)

Interventions aiming to increase adoption rates through information campaigns often rely on network diffusion strategies. A critical design component within these initiatives is selection of the network entry points in order to facilitate faster and/or wider information spread. This study combines a community-wide polling of network entry-points combined with detailed community network and socio-economic data. First, we explore what attributes are prioritized by community members in nominating a resident farmer as an extension contact-point. Second, we use simulations to compare the diffusion spread of top-nominated individuals as network entry-points compared to entry-points that achieve maximal spread within diffusion simulations. We find that community members prioritize network connectedness, pro-social preferences, and socioeconomic indicators of gender, age, formal leadership, and education levels within their nomination decisions. Furthermore, receiving the top three most amount of nominations is found to be significantly correlated with selection as an optimal entry-point within the diffusion simulation.

Social Networks and Social Preferences: A Lab-in-Field Experiment in Eastern DRC (with Jennifer M. Larson, Martha Ross, Peter van der Windt and Maarten Voors)

Considerable research points to social ties, and the social networks underlying these ties, as the underlying driver of pro-social behaviors upon which large-scale societal organization is based. However, little is known about the empirical relationship between social network position and pro-social preferences. Based on original network data and two lab-in-the-field experiments, we explore the relationship between social networks and social preferences of trust, trustworthiness and cooperation. We explore whether an individual’s observed social preferences are correlated with an individual’s centrality within the network structure. Our results indicate that individuals with high centrality are more trusting and more trustworthy than individuals with lower centrality. We also find that measures that explore the type of relationship between players are more predictive of trusting behavior. Within a group context, little evidence is found of more central individuals displaying more cooperative behavior. Instead, for group cooperation, when a single monitor can observe contribution decisions, the presence of a direct link and more mutual network connections with a monitor correlates to more cooperative behavior by that individual. Our results suggest that local network relations are more predictive of social behavior than network position.