“Never doubt that a small group of thoughtful, committed citizens can change the world,” wrote anthropologist Margaret Mead. “Indeed, it's the only thing that ever has.” A new study from the University of Pennsylvania puts a number to that theory: If you want to begin changing the culture, you want to get at least 25 percent of the people in your community on your side.

If at least 25 percent of a community’s population is committed to changing what is considered the social norm, the group will see a shift. The majority will adopt the new behavioral norm introduced by the minority. The study’s lead author, Damon Centola, an associate professor in the Annenberg School for Communication at the University of Pennsylvania, says this theory might come in handy when behaviors in the workplace, online, or in the political scene have crossed a line, and a small group of people decide to stand up to a majority that is turning a blind eye.

Centola has spent the last 10 years developing a method for studying change in large-scale populations. “The problem, historically,” he said, “has been that there is no way of testing or evaluating these theories.” So Centola and his fellow researchers began recruiting hundreds of people through ads posted on forums like Reddit and Quora.

In the experiments that led to this study, they formed 10 similar communities from the pool of recruits. "Using internet-based groups is a very exciting sort of scientific innovation,” Centola said. “It means that a lot of our theories about how social change happens can now be studied experimentally.”

After the 10 communities were formed, members were shown pictures and asked to name the object in the picture. The participants were quickly able to agree on one name. Once they all agreed and began, for example, calling the man in one picture “James,” Centola introduced a small activist group into each community that tried to change the social norm (give “James” a different name) and advance their alternative behavior.

“History, or the real world, only happens once. What we did here was to run the tape of history multiple times,” said Centola of the 10 communities. In order for the experiment to play out as close to reality as possible, each of the communities were identical; people were randomized to each community to ensure the 10 were statistically identical in terms of age, race, and gender, and that the size of the groups were the same. The only thing that was different was once they all reached the social norm, the researchers varied the size of the activist group—consisting of new participants—that was introduced into the population.

Five groups had committed minority groups less than 25 percent of the population. The other five had greater than that percentage. “Right up to 21 percent, there was no difference compared to 10 percent,” said Centola.

“But everything tipped when we reached 25 percent and beyond.”

Centola sees the Chinese social media giant Weibo as an example of this principle in action. The Chinese government sends many of its employees on Weibo to track civilians’ conversations and message threads. If complaints about the government are seen, these “actors” pose as regular Weibo users and begin a new conversation about things the government has done right, like building more parks or improving public transportation.

“If you get enough of these actors who coordinate and steer the conversation,” said Centola, “then the people who were airing their grievances feel less comfortable talking about them, and no one talks about grievances anymore.”