A smaller world, but not more intimate

Our increasingly interconnected world has made it easier for information and disease to spread. However, a new study from Harvard University and Cornell University shows that fewer “degrees of separation” can make social networks too weak to disseminate behavioral change. The finding that “small world” networks are limited in their power to shape individual behavior could have implications for health care policy and the treatment of epidemics.

Published in the November American Journal of Sociology, the study was led by Damon Centola, a Robert Wood Johnson Fellow with the Institute for Quantitative Social Science in Harvard’s Faculty of Arts and Sciences, with Michael Macy of Cornell. The work was done while Centola was a doctoral candidate at Cornell.

“Our research shows that the difference between contagions that spread through simple contact, such as diseases and information, and the spread of behavior, which requires multiple contacts, has important consequences,” says Centola. “We find that while making the world smaller increases the speed at which diseases and information can spread, it can actually slow down, and even prevent, the spread of health behaviors.”

Social networks consist of connections between individuals, whether they are “short ties” among family or good friends or “long ties,” as among acquaintances. Long ties are good for hearing about a job opening or housing opportunity because they provide new information that is not already circulating in your closely connected social network. However, when it comes to getting a cholesterol screening or going on a diet, you are more likely to get reinforcing signals from co-workers or neighbors who know each other and can collectively encourage you to change your behavior.

Previous studies have shown that long ties are highly effective at rapidly diffusing information among disparate social groups, and relatively few long ties are necessary to create the effect. Called the “small world theory,” it is exemplified in the idea that everyone on Earth is connected by “six degrees of separation.” Extensions of the small world theory have argued that anything to be diffused can travel faster as the world becomes smaller. But this study apparently disproves that generalization.

Most information and diseases have a threshold of one, meaning that contact with one network neighbor is sufficient to pass the contagion to another. Information or diseases can be spread along long ties because reinforcement from multiple sources is not required. However, many changes in behavior have a threshold of greater than one, such as the adoption of a new technology, diet, or exercise routine. In these cases, social affirmation or reinforcement from multiple sources is frequently required, making an isolated contact insufficient to persuade an individual to change her habits.

Using the “small world” model of social networks, the researchers showed that as the connectedness of the world increases, the level of local social reinforcement decreases. While people can get more information from more sources (and are susceptible to contact with greater numbers of germs), they simultaneously have fewer contacts connected to each other who can coordinate to provide social reinforcement for desired behavior change.

Closely knit ties are less effective for spreading information because of the increased likelihood that these individuals will know the same people, causing redundancy and repetition of information. However, the same redundancy that inhibits the spread of information provides vital channels of reinforcement for fostering changes in behavior. If social clustering is reduced too much, it may altogether prevent coordination on behaviors that require social reinforcement.

“As the world becomes smaller, these changes in social networks can reduce the ability to spread behaviors that are hard to adopt,” says Centola. “For health agencies, this suggests that the spanning networks that allow disease to spread so rapidly may not be as good for changing behavior. Rather, it may be better to focus on building and extending local, community networks, and encouraging grassroots support for desired behaviors.”

Current research extends these findings into the realm of “virtual” communities. At Harvard, Centola is investigating how building online networks affects the diffusion of health behaviors, while Macy at Cornell is using online data to study the spread of innovations.

The research was funded by the National Science Foundation and the Robert Wood Johnson Foundation.