Why a school beats Facebook: how behaviors spread through networks
[Via Ars Technica]
We all spend much of our days engaged in social networks, whether it’s online, at work, or out with our friends, and we have a tendency to pick up new habits through these connections. A new study in Science set out to determine how behaviors travel through these social networks, and how the topography of the networks affects the diffusion of the behaviors.
The experiment studied two different structures of social networks. In “random” networks, individuals are connected to others scattered throughout the network by connections that are called “long ties.” In more “clustered” networks, social ties exist mostly between individuals that are close together in the network; there are few (if any) long ties connecting individuals from different topographical areas.
[More]
This is an interesting paper because it shows a very important aspect of social networks and human behavior. It also makes the de facto assumption that online behavior directly mimics real world behavior.
Truthfully, it provides hard data for what we already know. But in a really novel way. People adopt change at different rates, putting themselves into several different groups – disrupters, mediators, doers and memory. Interestingly, it is possible to differentiate each group by the shape of their social network. Doers are highly connected to each other, with little connections outside. Disruptors have many connections to other groups outside the local network but few in the doers group connect to them. Mediators have a similar pattern as disputers but have many doers that connect to them as well as connect to doers.
The hallmark of doers is that they will only change behavior when others that are close in the community to them change. Disruptors, on the other hand, change behavior when given information from outside the community.
Essentially he created a network that mimics doers – where everyone connects closely to everyone else – and a network of disruptors – where long connections between groups are present.
He randomly put people into each network and looked at their behavior. Since most people are doers, they are going to change their behavior when others around them do. That is what roughly 67% of a normal population would do. So it is no surprise that the network that has the most changes is the one that most closely mimics a normal network of doers. They would not be comfortable at all existing in a network of disruptors.
It would be interesting to test people beforehand – to see whether they are doers or disrupters, for instance. And then put them into a network designed for their type ot not. I would suspect that disruptors would be very happy to change behavior even when put in a random network.
This is also a wonderful demonstration of how online metrics can reveal things.
