Students

In the Twitter world, the weak ties are strong

Wielding influence on Twitter is all about getting thousands or even millions of followers, right? Not exactly. It turns out there are other, potentially better ways to measure someone’s impact on the burgeoning social-media platform. New research by Dr. Michael (Zhan) Shi, assistant professor of information systems in the W. P. Carey School of Business finds that the strength or weakness of relationships between Twitter users and their followers are important factors in determining how widely users’ messages will spread.

2014-01-30

Wielding influence on Twitter is all about getting thousands or even millions of followers, right? Not exactly. It turns out there are other, potentially better ways to measure someone’s impact on the burgeoning social media platform. New research by Dr. Michael (Zhan) Shi assistant professor of Information Systems in the W. P. Carey School of Business and colleagues, finds that the strength or weakness of relationships between Twitter users and their followers are important factors in determining how widely users’ messages will spread.

It might seem counterintuitive, but Shi’s team found that weak, one-way relationships between Twitter users made it more likely that a follower would re-tweet or re-post, an author’s tweet or message. On the other hand, strong, mutual relationships made it less likely that users would re-tweet each other’s messages. The difference in re-tweeting behavior has notable implications for anyone wanting to tap the power of social media. Followers in weak relationships, the research found, were 50 percent more likely to re-tweet messages than were users in strong relationships.

The findings surprised Shi’s colleague who thought strong-tie followers would re-tweet often as favors to their friends. But Shi, a longtime Twitter user, said he re-tweets when he considers content valuable to his followers. “I think Twitter is more about information, so it is less social,” Shi said. “Twitter users, if they want to build a reputation as a connected person, as people who know what is going on in this world, would be willing to share content that has high informational value to their followers.”

Twitter began in 2006 and is known as a tool for transmitting messages of up to 140 characters. It has grown to more than 230 million active users who send some 500 million tweets a day. Twitter often is lumped in with Facebook and LinkedIn as another social network, but Shi and his colleagues argue that Twitter, Tumblr and similar networks belong in a new category they call “social broadcasting networks.”

Twitter’s success, they note, depends on users creating content and broadcasting that content to audiences, or followers, who then re-broadcast the content to their own followers. Unlike Facebook and LinkedIn, whose users consent to be friends or connections, Twitter is more like radio or television broadcasts that need no mutual consent and are open to anyone with the technology to receive them. Shi collaborated on his research paper, “Content Sharing in a Social Broadcasting Network Environment: Evidence from Twitter,” with Huaxia Rui of the University of Rochester in New York and Andrew B. Whinston of the University of Texas at Austin.

Exploring how, why we share information

The field of information systems continually tries to understand how people disseminate information through technology. Shi and his co-authors became curious about why people choose or don’t choose to re-tweet messages they receive. “Your followers and your followings on Twitter typically don’t know each other … and they’re still willing to do this,” Shi said. “People’s willingness to share a piece of information actually makes a platform like Twitter very powerful in disseminating information, especially when big events happen.”

Their curiosity led the team to the work of American sociologist Mark Granovetter and his 1973 landmark paper, “The Strength of Weak Ties.” Granovetter theorized that weak ties connect diverse groups of close friends and allow new information about potential jobs to flow between mere acquaintances. In groups with strong ties, such as people who spend time together and confide often in each other, there are fewer connections with diverse groups and thus less access to new information.

Granovetter’s research indicated that people who found jobs were more likely to have heard about them from people with whom they had weak ties than from people with whom they had strong ties. Shi’s team found other research exploring the social rewards that motivate people to share information. High on the rewards list is reputation: Although it takes time and effort to share new or high-quality information, people perceive that doing so can benefit their reputations. Shi and his colleagues asked how the strength of ties affects people’s behavior in sharing content on Twitter. They recognized that Twitter provides two kinds of ties:

• Bidirectional, in which two people “follow” each other. Bidirectional ties are relatively strong, on average, and these circles of followers would be more likely to overlap. For example, coworkers who follow each other on Twitter would have bidirectional relationships, and their circles of followers would closely overlap.
• Unidirectional, in which one person “follows” another, but is not followed in return. Unidirectional ties are weak, on average, and these users would be less likely to have circles of followers that overlap. For example, coworkers who follow people in other companies, industries or organizations, but who are not followed by those users, would have unidirectional relationships, and the various circles of followers would overlap only slightly.

To study re-tweeting behavior, Shi’s team collected data on 65 tweets, all sent by different users with fewer than 1,500 followers; on those users’ followers; and on the users those followers were following, for a total of 24,403 observations. They collected the tweets from Twitter’s database of Top Tweets over 140 days in 2010, and tracked those that were shared through Twitter’s Retweet function for five days. For comparison, they also took a random sample of 200 tweets from all of Twitter and tracked the re-tweets for two weeks. The researchers examined IDs of the tweets’ authors, followers and users whom the followers were following.

That step confirmed that users in unidirectional relationships had fewer overlaps in their circles of followers than did users in bidirectional ones. The team also developed a two-stage model of how Twitter users share information. In the first stage, authors send tweets to their followers, and in the second stage, followers decide whether to re-tweet the message. Two stages are important, Shi said, because the team found that followers missed or ignored 54 percent of tweets they received, so only the tweets they read could reach the decision stage.

Crunching the data proved Shi’s early instincts right. Unidirectional followers, or those with weak ties to a tweet’s author, re-tweeted 9.1 percent of the messages received. Bidirectional followers, or those with strong ties to an author, re-tweeted 6 percent of the messages. The difference was just 3.1 percentage points but translated into a 50 percent greater chance that an unidirectional follower would pass along a tweet than a bidirectional follower would. In workplace terms, the research suggests that coworkers who follow Twitter users in other companies, industries or organizations are 50 percent more likely to re-tweet messages than coworkers who mainly follow each other.

Applying ‘70s theories to Twitter

Why do weak ties result in more re-tweets? Shi thinks two different mechanisms — new audiences and new information — are at work. First, he said, users with strong ties figure that their circle of followers overlaps with the tweet author’s circle, so they assume many followers already have received the information. They see little benefit in sharing old news. Second, Shi said, users with weak ties are more likely to re-tweet because they assume information will be new to their own followers.

They expect their followers to find the new information valuable, a scenario that enhances the re-tweeter’s reputation. The research, Shi emphasizes, does not focus on the number of a Twitter user’s followers but on followers’ willingness to disseminate information. It’s a first step in developing new indexes to quantify a Twitter user’s social influence. “We’re not saying that the number is not important,” he said. “But our study shows that it might not be the best measure … of social influence, because you may have the same number of followers, but their willingness to disseminate your information can be very different.”

Though the “Strength of Weak Ties” theory has been around for decades, Shi and his colleagues showed that it applies to technology unimagined back then. And though the theory originally applied to real-world connections of a few hundred people, they showed it also can apply to online connections of potentially hundreds of thousands of people. Going forward, they say, there is room for much more research on how people share content on platforms like Twitter. Because of resource limits, they couldn’t track re-tweets to infinity, but they expect their experiment could be replicated step by step and the results would hold.

They also couldn’t track celebrities and other users with huge numbers of followers, so the median number of a user’s followers in their sample was 190. Work also could be done, they say, to measure the strength of users’ ties based on the amount of conversation between them, and to see if demographics or tweet-specific language affect re-tweeting behavior.

Bottom line

In applying the notion of weak ties to Twitter users’ willingness to share information, Shi sees these takeaways:

• For users seeking social influence: The number of your followers might not be the best measure of your influence on Twitter. Weak ties with diverse followers might be a bigger factor in people’s willingness to disseminate your words.
• For followers on Twitter: If you use the network as a personal news feed, consider whether you want to hear information you already believe or a more diverse set of voices. Following only people who say things you know might make information from Twitter less valuable to you over time.
• For information systems professionals: Twitter’s broadcast platform can help disseminate information quickly to large groups, because people are motivated to share new information with new audiences. However, information overload is a severe and commonplace problem on Twitter.

KnowIT