Date Published: May 15, 2019
Publisher: Public Library of Science
Author(s): Soojong Kim, Luis M. Miller.
How do echo chambers operate? Why does social propagation of information become trapped within the boundaries of social groups? Previous studies of these questions have identified informational and structural factors which hinder information exchange across group boundaries; these factors constitute “chambers” in which information flows are confined and transformed into “echoes.” However, empirical evidence has indicated that these factors may not sufficiently explain the mechanism of echo chambers. Hence, the present study investigated whether the insular flow of information emerges and endures without the chambers. A randomized controlled experiment was conducted in which participants, who were classified into two political groups, exchanged randomly selected articles with the same number of ingroup and outgroup neighbors. The experiment manipulated the directionality of incoming information flow by varying the number of articles sent from ingroup neighbors across two conditions. Analyses revealed that the ingroup-slanted inflow induced ingroup-slanted outflow, suppressing transmission toward neighbors in a different social group. The biased inflow also promoted positive reactions to information exchanges and reduced negative evaluations on the exchanged information. Furthermore, the ingroup-slanted inflow increased false perceptions of ingroup majority, which is known to encourage information dissemination by a social group. The present study suggests two self-reinforcing mechanisms of ingroup-biased flows that generate echoes even without the chambers. These mechanisms may enable a small group of strategic actors to exacerbate polarization within a large population by manipulating directions of information flow.
Information spreads through pathways of social networks and reaches a large population within a short period of time [1–3]. The Internet and social media have reduced temporal and spatial constraints on communication, further facilitating information dissemination on social networks. However, social propagation of information is still often constrained by people’s social identities, such as race, gender, and political affiliation. The spread of information tends to be bounded within a social group, leading to “echo chambers” that could cause social segregation and political polarization [4–7].
Participants selected 91.8% of the shared articles on average (M = 11.01, SD = 2.04), and 94.2% of them selected more than half of the shared articles. The participants read an article for 11.2 seconds (SD = 4.64) in the balanced condition and for 10.8 seconds (SD = 4.36) in the ingroup-biased condition on average. Those in the balanced and biased conditions selected 7.6 (SD = 1.62) and 7.3 (SD = 1.69) unique neighbors for transmission on average. For these observations, differences between the conditions were not statistically significant (see Methods for details). The results demonstrated that participants were able to distinguish each article based on their social identity and the message content, and a social group’s evaluation of an article was associated with the likelihood that the social group transmits the article to ingroup neighbors (see Methods for details).
These findings reveal that ingroup-slanted inflow of information biases the direction of transmission, the social reaction to information exchange, the evaluation of information, and the perception of social networks. First, it was found that ingroup-biased inflow promotes ingroup-biased outflow, which in turn increases the ingroup directionality of inflows of their neighbors. Thus, the finding provides evidence of positive and mutual reinforcement between biased flow of information and biased information behavior. This “spiral of segregation” insulates people from information exchange across group boundaries, generating and maintaining barriers that hinder intergroup communication. Along with informational and structural segregations, this dynamics may also contribute to segregated information flows on social networks [4,19,41], which lead to echo chambers and political polarization.