Damon Centola

When it comes to public health, the conventional wisdom about how best to spread information on social networks might need to be turned on its head, according to a recently published paper by Professor Damon Centola of the MIT Sloan School of Management, entitled "Social Media and the Science of Health Behavior."

"If you really want people to change their diet, their health, their prophylactic use, the kinds of networks effective for diffusing behavior are quite the opposite of the networks for diffusing information," Centola told MobiHealthNews.

Social epidemiologists hold up a theory called "the strength of weak ties." If you want to spread information through social networks, you want a lot of connected people who aren't very close friends, because that increases the likelihood that they'll pick up a message from one person and spread it to a new person. Conversely, in networks made up of close friends, the same message gets spread around over and over within the circle, but never gets spread outward.

That theory can't be generalized to health information, which is nearly all behavior-based, Centola argues. In the case of using a social network to encourage smoking cessation, for instance, it's much more valuable for people to hear the same message over and over again from close friends than it is for many people to hear it once. Centola built support for this theory in a small study published in 2007.

This is just an example of the kind of finding Centola said is possible with online social networks like Facebook and Twitter, as well as online forums created specifically for research. They are much easier for sociologists to study than offline social networks, and new technology makes it easier than ever for researchers to study social networks, by facilitating anonymized data and online informed consent forms.

"The idea here would be to identifiy what the effects are of different peer-to-peer structures of environments on population health," Centola said. "You can rarely compare a population to another population and have everything hold the same except for the social network. This is sort of the magic of these online communities. Twenty years ago, 30 years ago you couldn't have imagined doing population studies like these."

For instance, Centola was able to do a fitness study where he compared networks in which people were grouped randomly with networks where participants had a common age, gender or BMI. He found that subjects with more in common with one another had better adoption of fitness regiments.

Centola is particularly excited about the potential for connected activity trackers, many of which interact with social networks, for ongoing research. Device data could be much more reliable than self-reported data, providing direct evidence of the impact of social network participation on actual exercise or heart rate.

"We have a spat of new companies arising in this space that have a record of exercise and heart rate during the day and they can just upload these directly and integrate them into social network sites," he said. "Our social interventions would be measurable in terms of the amount of running and exercising people are really doing."