The Power of Being Influenced
By Julie J. Rehmeyer
Sometimes an idea spreads through society like a newly-mutated cold virus zooming through a class of first-graders. Other times, a good idea never seems to take hold. What makes the difference? Scientists want to know, and marketers want to know even more, since they make their living spreading ideas about their products.
A key reason some ideas are so successful, conventional wisdom has held, is that a few highly influential people espouse them. In his book The Tipping Point, Malcolm Gladwell wrote that what he calls "social epidemics" are "driven by the efforts of a handful of exceptional people." Those exceptional people tend to be experts on a subject who love to talk. Such people can convince dozens of others of their opinions. An excellent sales strategy, then, would be to find those few critical people, persuade them of the value of your product, and leave it to them to convince others.
It's a compelling idea, but does it really work? Social network theorists Duncan J. Watts of Columbia University and Peter Sheridan Dodds of the University of Vermont in Burlington decided to put the notion to a test. What they found is a disappointment for "viral marketers" who specialize in selling products by influencing influential people.
The researchers started with a basic premise: When many of your friends believe something, you're more likely to believe it, too. How many friends it takes to persuade you depends on your personality. Watts and Dodds developed models in which each individual in a network would adopt a particular idea if some percentage of the individuals connected to it had already adopted it. In the models, that percentage varied from one individual to another, so some individuals were easily influenced while others were more resistant.
Next, Watts and Dodds had to decide on the shape of the network: how many connections each individual would have and to whom. Unfortunately, it's hard to determine in the real world who influences whom, so no one knows much about the properties of the networks that people form as they influence one another. Therefore, the researchers couldn't create a perfect network that closely mimics the real world. Instead, they decided to create a variety of different types of networks to see how ideas spread through each of them.
The researchers compared how far an idea would spread depending on whether it started with a random individual or with an influential individual who was connected to a lot of other individuals. They found that highly influential individuals usually spread ideas more widely, but not very much more widely. For example, if an individual had three times as many connections as the average person, ideas espoused by that individual almost always spread substantially less than three times as far as the ideas of an average individual. Sometimes, the researchers found, the difference wasn't even measurable.
In a few networks, ideas espoused by influential individuals spread much further than those of average folk, but those types of networks were not common and not likely to be similar to the real world, the team reported in the December Journal of Consumer Research.
More important than the influencers, the researchers found, were the influenced. Once an idea spread to a critical mass of easily influenced individuals, it took hold and continued to spread to other easily influenced individuals. In some networks, it was far easier to get an idea established this way than in others. The entire structure of the network mattered, not just the few influential people.
Dodds compares the spread of ideas to the spread of a forest fire. When a fire turns into a conflagration, no one says that it was because the spark that began it was so potent. "If it had been raining," Dodds says, "that same match wouldn't have had an effect." Instead, a fire takes off because of the properties of the larger forest environment: the dryness, the density, the wind, the temperature.
The upshot of the study, Dodds says, is that "in the end, you don't have control over how people spread your message." The best way to increase the odds of person-to-person transmission of an idea is to make it a good idea and to give it "social worth," he says. "Some things are just fun to talk about."
References:
Watts, D.J. and P.S. Dodds. 2007. Influentials, networks, and public opinion formation. Journal of Consumer Research 34(December):441-458. Abstract available at http://www.journals.uchicago.edu/doi/abs/10.1086/518527. Preprint available at http://cdg.columbia.edu/uploads/papers/watts2007_influentials.pdf.
Peterson, I. 2006. College friendships and social networks. Science News Online (Jan. 21). Available at www.sciencenews.org/20060121/mathtrek.asp.
Gladwell, M. 2000. The Tipping Point: How little things can make a big difference
Comments
You can find the full final version of the article at http://cdg.columbia.edu/uploads/papers/w_d_JCR.pdf
Posted by: Josh | January 5, 2008 12:08 AM
It seems that information via mass media creats an instant critital mass of opinion. In conjunction with a citizenry unschooled in critical thinking, the acceptance of dangerous and/or unethical practices (such as war or torture) could be easily and quickly created.
Posted by: j.cunningham | January 6, 2008 12:40 PM
one of the stupidest things I've read all year. How do you model an influential person's power to inspire? Influential people have greater resources to package their message in compelling ways (with powerful footage, photography and music) and to reach larger audiences (in magazine interviews, as guests on TED, in radio interviews.) Try modelling all those factors and get back with us.
Posted by: max hodges | January 6, 2008 01:01 PM
I wonder how much effect is gained by the *belief* that others have been or will be similarly influenced, and whether that might be a factor that gives "influential people" large parts of their power. In other words, perhaps some of the influence that accrues to certain people is based on how much attention one can predict will be paid to their opinion (they have influence because...they have influence?).
Posted by: David | January 8, 2008 11:40 AM
At least from this description, it seems like the researchers got out of their experiment only what they put in: the initial assumption appears to be that a person is influenced only by the number of associated people who believe something. The "result" that influencing a bunch of easily influenced people has the strongest effect on belief change is merely a restatement of that assumption. The supposed negative result that influential people are not much better at triggering a belief than anyone else is flawed because the definition of an influential person is incorrect. Someone with more connections to other people is not more influential than anyone else: their effect on a particular connection is exactly the same as any of the other people connected. Sure, you could claim that they are "more influential" because they bump up the count of neighbors for more people, but that is an indirect effect heavily dependent on graph topology.
The initial assumption defined the (direct) influence of every person as being identical, so it is odd to categorize some people as being more influential than others using a definition related to the hypothesis being tested.
A more interesting study would be to allow the direct influence of individuals to vary, and then compare the effects of that variance with the effects of the network topology.
Posted by: Steve Fink | January 8, 2008 09:19 PM
This is of profound interest and application to public health and the understanding of politics. The key idea that needs to be expanded is the idea of "easily influenced", and what this exactly means at a network level needs to be explored.e.g. an idea which fits into pre existing ideas may be more readily accepted.
Posted by: Bill O'Slatter | January 9, 2008 09:23 PM