Every day the morning starts either around the watercooler, grabbing coffee, or checking Facebook, sharing the latest "did you hear" stories. These casual exchanges of information are referred to as the Watercooler Effect. It's through these exchanges, where unverified information is spread through social networks, that rumors emerge.
"A rumor is what you do when you try to make sense of things with other people," notes Dr. Nicholas DiFonzo, professor of psychology at RIT and leader of the research team. "It's collective sense-making. The classic example is, 'I heard that the department is downsizing; what did you hear?'"
The social networks in which a rumor spreads can be as small as a married couple or as large as a million-person chat room on the Internet, and the rumor can be either positive, "Did you know Hank in accounting got drafted by the Cowboys?" or negative, "I heard Sam got promoted because his dad is friends with the boss." While rumors make up a great deal of the information we hear on a daily basis and are a central component of how we socialize and relate to each other, very little is known about how rumors spread in society and how different types of social networks impact transmission.
"Numerous rumors, such as those surrounding the death of Michael Jackson or the assassination of John F. Kennedy, linger for months—even years—and sometimes are considered more truthful than well-established facts," says DiFonzo, who is also the author of the scholarly book Rumor Psychology and the mass market book The Watercooler Effect. "Many rumors engender mistrust, suspicion, and conflict between people and groups, often causing societal chaos. Through better understanding of how rumors grow and spread, we hope to decrease their negative impact and often damaging consequences."
One of the most interesting and little-understood facets of rumors is how they propagate or spread from one person or group to another, and how different social networks affect that propagation. Through a grant from the National Science Foundation, DiFonzo's team has sought to mathematically model rumors in spatial networks and create a predictive equation to simulate how a rumor spreads.
"Social interactions are much more difficult to model than biological or chemical reactions, because many of the data points are incredibly fluid and many variables may not be known," adds Dr. Bernard Brooks, associate professor of mathematics at RIT who is collaborating with DiFonzo on the project.
To create a proper model for rumor propagation, the team used several years' worth of empirical data, measuring how different rumors had spread, and applied Dynamic Social Impact Theory (DSIT) in developing the model parameters. DSIT states that beliefs and attitudes are based on:
Strength of influential sources—You are more likely to believe a rumor told by a friend or family member. Immediacy of influence—Rumors often take hold in close-knit, homogeneous neighborhoods and communities. Number of sources—The more people in your network there are who believe a rumor, the more likely you will believe it.
Based on this, Brooks and DiFonzo developed the MBN-Dialogue Model of Rumor Transmission, where rumor spread is based on the motivations (M) for spreading the rumor, the strength of belief (B) in the rumor, and the novelty (N) or newness of the rumor. The model is one of the first to simulate how a rumor moves through a group based on empirical rumor research and can be used to model belief in derogatory rumors about an opposing goup, explore why some persist and some die out, and how different social networks impact propagation.
The team, which also includes David Ross and Deana Olles of the department of mathematics and Christopher Homan of the department of computer science at RIT, as well as faculty and students from the University of Wyoming and the University of South Australia, used the model to conduct a number of experiments in which groups of networked participants selected and discussed rumors via e-mail. Three 16-person groups each discussed eight ambiguous situations, such as "you hear that a professor you had was found dead from a gunshot wound." They then chose one of four alternate rumors provided to explain each situation, such as "he was robbed," and then discussed the different scenarios among themselves. The analysis investigated how the type of network that students were put into affected the degree to which homogenous pockets or clusters of rumors emerged, the extent to which rumors persisted or died out, and the level of confidence participants placed in rumors over time.
"Our data indicates that whether a rumor will be believed is less dependent on the rumor itself and more on the specific dynamics of the network it is introduced in," says DiFonzo. "For example, rumors introduced where groups of individuals are clustered together into clumps or cliques and have few connections to other groups in the network are more likely to take hold than in configurations that are less segmented."
Among the key findings: spatially proximate pockets of rumors emerged based on like-minded individuals adopting belief; over time rumor diversity decreased as certain rumors gained majority or plurality support; but minority rumors believed by a small number in a network did not typically die out even when a majority of individuals disbelieved the information.
"The results verify the use of the Dynamic Social Impact Theory to explain rumor spread and confidence and suggest that rumor propagation can be measured and predicted based on an analysis of the network in which it is introduced," says DiFonzo. "This has clear implications not only for psychologists and social theorists but also for policy makers looking to analyze how information, particularly false information, spreads through communities and affects citizen reaction, especially during disasters and crisis events such as September 11, 2001."
The team is now exploring how rumor propagation is related to social categorization and in-group/out-group dynamics through an analysis of praising or derogatory rumors that circulate in situations of rivalry, conflict, or different group memberships. The research will better define how specific social networks influence rumor propagation and provide a deeper understanding of how rumors impact relationships between often contentious groups such as Democrats and Republicans.
"There is a tendency to categorize people in us-versus-them terms, which can often lead to hostility toward the out-group," says Brooks. "Positive and negative rumors play an important role in this dynamic by enforcing in-group worthiness and out-group inferiority. Opposing political groups provide an excellent vehicle to test how rumors can enforce self-categorization."
An initial experiment used nine 16-person groups, each containing eight moderate to strongly identified Democrats and Republicans. Groups were presented with a series of nine controversial rumors, some positive and some negative, about the respective political parties, and asked to discuss the rumors.
The data shows that Democrats were more likely to believe negative rumors about Republicans and positive rumors about Democrats, while the Republican participants indicated the reverse. In addition, as the rumors propagated, confidence in out-group negative rumors increased more than in-group positive ones, illustrating the power of negative rumors, particularly when social categorization is a factor. Interestingly, when the networks were more integrated, with less homogeneity, rumor confidence decreased, especially with negative rumors.
Next, the team will conduct similar experiments using groups of men and women and hearing and deaf individuals to examine how other social groups with different in-group/out-group dynamics are impacted by positive and negative rumors. They will also look to modify the MBN-Dialogue Model to better assess out-group negative rumor propagation and examine how it differs from traditional rumor propagation. DiFonzo also hopes the work can ultimately assist in reducing rumor polarization to improve conflict reduction and cooperation efforts, particularly among conflicting social groups.
"Our findings suggest that integrating groups with individuals that have opposite viewpoints reduces the strength of belief and overall spread of negative information," notes DiFonzo. "This enhances the notion of increased social interaction as an antidote to out-group prejudice."
Rumors play an important role in everything from office politics to popular culture to whom we elect as president. By creating a broader scientific framework with which to analyze the impact of rumors and better educate individuals on how to deal with it, researchers at RIT hope to decrease the negative effects of this social phenomena. In addition, they want to enhance the general study of rumor psychology as a mechanism for creating better communication between citizens, communities, and governments.
"By better understanding rumors' impact on how we think and what we believe, individuals and groups can better communicate goals and reduce the possibility of misunderstanding and distortion at all levels of society," concludes DiFonzo.