Stopping Smoking with Social Networks
http://www.nejm.org/doi/full/10.1056/NEJMsa0706154#t=article
In taking a look at the underlying social network of a population, we can see dramatic roles that it plays in the transmission of not only technology, but also health-related behaviors related to non-communicable diseases. A networks study on smoking by the New England Journal of Medicine revealed that smoking cessation is highly correlated between friends and family members. A smoker was 36% more likely to stop smoking if their perceived friend stopped smoking. Between mutual friends who were smokers, if one friend stopped, the other friend was 43% more likely to also stop smoking. If one spouse stopped smoking, the other spouse was 67% more likely to stop smoking as well.
The study demonstrates that smoking cessation acts a form of social trend acting very strongly amongst people with close ties. In essence, smoking is a lifestyle choice that becomes less socially acceptable as more closely linked people are no longer smoking. This can be viewed in the context of our recent discussions on diffusion through networks. For any person, there is a higher payoff to not smoking from health benefits, but there is the barrier of withdrawal to choosing to not smoke. The increased social pressure to not smoke from close ties helps to offset the cost of withdrawal and motivate the person to quit smoking.
This can be viewed in the following example: between two friends A and B who have an edge between them in a network, each has the choice to smoke or stop smoking. If both are smoke, they each get a payoff of 1. If one smokes and the other does not, they have social tension about the differences in lifestyle choices they each have a payoff of 0. If both do not smoke, they each have a payoff of 2 as there is no social pressure to smoke and there is an increase in health benefits.
Thus people who have more close links that quit smoking are more likely to copy the attitude that smoking is less socially acceptable, and will have a higher chance of quitting. The study discovered that in the final analysis, smoking behavior became clustered over time and became at the periphery of the social network. This is very analogous to the idea brought up in class of the threshold rule in network diffusion. If the payoff of both A and B smoking divided by the sum of the payoff of both A and B smoking and the payoff of both A and B not smoking is less than or equal to the fraction of close links that are not smoking, then that person will quit smoking as well. In the example above, the threshold would be 1/3.
It is not hard to imagine then, that if a person at the center of the social network decided quit smoking and their close links surpassed the threshold, the behavior of quitting smoking can diffuse outward through the network, only stopping at key people in the network that smoking does not cross the threshold. They then can serve as a barrier to the diffusion of not smoking and prevent further close links from adopting non-smoking behavior, resulting in the peripheral clustering of smokers.
Thus, if the right people in the network structure chose to stop smoking there can be dramatic impacts for public health.
That is interesting game theory, and makes a lot of sense. I have found most people go back to smoking if they live in the house with someone who smokes, or work with smokers. Smoking is a social thing and so must quitting. I like the corporate wellness programs because it encourages a whole workplace to quit at once. Quit Tea is used in these programs http://www.quittea.com and drinking tea could be a social thing.