The “cold start” problem
For years, researchers and intervention designers have struggled to realize the potential benefits of social networks in driving behavior change. Despite well-publicized research showing that real-life social networks can impact behavior (such as Nicholas Christakis’ work showing the spread of health-related behaviors through social networks in Framingham), creating “new” networks that are engineered to assist change has been challenging.
A paper recently published in the Annals of Behavioral Medicine demonstrates these challenges in an explicit way. The researchers randomized participants from three different Pennsylvania communities into one of three groups. The first group received tips on walking. Members of the second group received the same tips, plus an online walking program and pedometer. They also interacted with research staff who provided them with tailored walking goals based on walking logs. The third group received everything in the first two groups and was also given access to an online group for their community, either in Ning or Facebook. A total of 308 people were randomized to the three groups across the three communities, or about 33 people per group per community.
The researchers found no difference between the three groups. The group that got the simple walking tips increased their walking by the same amount as the group with pedometers, interaction with research staff, and access to the social network site. Not surprisingly in retrospect, participants were not terribly engaged with their online networks (and in an effort to promote engagement, the researchers even resorted to offering financial incentives to two of the three community groups). This problem will sound familiar to anyone who has ever tried to implement a real-world wellness program – and as you can see from this trial, what can seem challenging in a 1,000-person employer is magnified when groups are much smaller.
There are two very different models for engineering social networks and social support. The more common model, which is often used in research programs and worksite programs, takes a set of enrollees and assigns them together with resources and networking tools in an attempt to bootstrap a social network. In doing so, they suffer from what has been called the “cold start” problem, or the simple fact that it is difficult to create a true network from scratch, particularly on a timeline.
The MeYou Health solution
An alternative model attempts to build a much larger, persistent social network over time, enrolling new members into that social network after it is stable. This model, which MeYou Health uses and pioneered with QuitNet, has the benefit of disentangling the process of creating a network from its effect on an individual participant. For researchers in particular, however, it presents a challenge, as it takes significant time and resources to build out a large, persistent social network dedicated to a given behavior or medical condition.
In our paper on the QuitNet social network, we suggested that researchers should document the existence of a social network in any social network intervention using formal methods. Put another way, a social network is a set of people with connections to each other actively communicating with each other. Without this, all you have is a “social network site,” which is a very different thing.
This question matters for academic researchers as well as purchasers of wellness services. If someone tells you they have a social network, they should be able to demonstrate it with straightforward network measures, such as the mean number of connections or the network density. Without evidence of a social network, it's safe to assume that the network site is suffering from a cold start problem.