Experiments with Face-to-Face Networks

A significant amount of our attention is devoted to the development of new ways to intervene in social networks to promote public health. It's part of our effort to answer the "so what" question: So what if we can understand how networks form and operate, what can we do what this knowledge to make the world better?

Our first effort in this regard, which took place during the H1N1 flu outbreak in 2009, involved the demonstration of a whole new way to forecast epidemics in advance of their striking the general population. In the past few years, we have been showing that it is possible to manipulate not only online, but also face-to-face, networks. Manipulating the structure and function of social network ties can affect diverse outcomes, from the adoption of new health behaviors to the maintenance of cooperation to the cultivation of innovation and creativity.

Our ongoing experimental work on face-to-face networks includes an innovative, large randomized controlled trial of network targeting (with a maternal and child health intervention) being conducted with 30,000 people in 176 villages in rural Honduras. We have used our network mapping software, Trellis, to map the social networks of these villages, and then to test different network targeting algorithms to see how we might be able to create artificial tipping points for behavior change in the population. If we are able to identify the right 10% of "structurally influential" people in these village networks, for example, and persuade them to change their behavior, can we get the whole village to change? Through this work, we hope to develop more effective ways to optimize the spread of desirable behaviors (and our ideas are not limited solely to public health interventions in developing world settings). Relatedly, we are studying how social interactions are themselves re-shaped by public health interventions.

We also use our data-collection efforts in real-world settings (even if not part of a field experiment) to advance our understanding about a range of topics, from antagonistic ties in human populations, to inter-personal violence, to work-group performance in multi-national firms.

Related Papers