Researchers explore truth-telling on social networks
by Stephanie Koons, writer/editor for the College of IST
Online social networks such as Facebook and Google+ create opportunities for users to share personal information, but also to bend or withhold the truth to suit their own purposes. While privacy may be a commonly voiced concern, according to two Penn State researchers, a social network user’s decisions to lie or withhold information are more closely tied to peer pressure and a desire for popularity.
Anna Squicciarini, an assistant professor at the College of Information Sciences and Technology and Christopher Griffin, a research associate at the Applied Research Laboratory, recently won a best paper award for “An Informed Model of Personal Information Release in Social Networking Sites,” which Squicciarini presented to the 2012 Privacy, Security, Risk and Trust conference in Amsterdam. In the paper, Squicciarini and Griffin describe a game theoretic approach to characterizing a user’s willingness to release, withhold or lie about information depending on the behavior of individuals within the user’s circle of friends.
“If everyone in a user community thinks that bending the truth is OK, there is less incentive to be truthful,” Griffin said.
To ground their model, Squicciarini and Griffin conducted an extensive empirical study, collecting data about users’ common behavior and their attitude towards personal information disclosure. The study, which employed a Web-based survey, involved almost 300 subjects, all active social network users.
The study revealed important insights on users’ attitudes and practices, Squicciarini and Griffin said. The survey was constructed to study three specific aspects of users’ behavior: privacy awareness, attitude toward information withholding and practices, and attitude toward lies and misrepresentation.
The results led to some interesting findings, the researchers said. First, Squicciarini said, the participants expressed concern with their privacy on social networking sites as well as with the potential loss of privacy. However, the results also confirmed the existence of a “privacy paradox,” in which individuals state that they have privacy concerns but nonetheless provide detailed personal information.
While 94 percent of the users surveyed declared that they had lied on social networks, Squicciarini said, their level of truthfulness was dependent on certain factors. A regression analysis showed that participants feeling peer pressured are more likely to display a detailed profile and that the more they feel pressure from their friends, the less likely they are to withhold information. In addition, the study confirms that the more users perceive certain social interactions to boost their social status, the more likely they are to disclose personal information.
Users are also selective in what details they lie about and the degree to which they misrepresent themselves, Griffin said. Completely fabricated profiles are not common, while “white lies” are more widespread. People rarely lie about their name, age, gender or other basic information on highly connected networks such as Facebook unless they are trying to be playful or ironic, the researchers said. Information that is generally regarded as private, such as telephone number or grade-point average, is mostly withheld or misrepresented. Users display more variability with respect to how truthful they are with respect to their location, current occupation and relationship status.
“Lying is convenient only for certain types of information that is not going to be verified by the social network,” Squicciarini said.
By gathering data that yields insight into users’ motivations in sharing, withholding or misrepresenting certain types of information on online social networks, Squicciarini and Griffin said, they hope to deepen their understanding of the dynamics that drive successful online communities. They may run a larger study that would examine how a social network system evolves over time so that users learn what the right levels of truth, withholding and disclosure are. Eventually, they may transfer their deception model into a new social network monitoring system.