Learning from Shared News: When Abundant Information Leads to Belief Polarization

Bowen, T. Renee; Dmitriev, D. A.; Galperti, Simone

doi:10.2139/ssrn.3785799

Cited by 4 publications

(5 citation statements)

References 76 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Since projects with investment tend to be those with favorable private information, future observers who do not adjust for this fact will tend to be overoptimistic about the viability of projects. Bowen et al (2021) and Weatherall et al (2020) study social learning models, where some individuals don't share information that goes against their preferences, and receivers do not account for this. These models focus on network structure and when these biases may prevent learning the truth.…”

Section: Related Theoriesmentioning

confidence: 99%

Selection Neglect and Political Beliefs

Little¹,

Brundage²,

You³

2022

Preprint

View full text Add to dashboard Cite

Individuals, like researchers, often have to form beliefs about the political world from non-representative samples; e.g., their friends, what they see on TV, or content on social media. Substantial evidence indicates that many struggle to account for this selection problem, and generally form beliefs as if what they observe is representative. In this review, we provide a formal typology of how this phenomenon of selection neglect affects political beliefs. Homophily leads individuals to believe others’ traits and beliefs are closer to their own. The squeaky wheel effect biases beliefs towards more visible or vocal groups. And the man bites dog effect leads to excessive belief in extreme or unusual events. Selection neglect is a unifying way to understand disparate literatures on perceptions of the economy and demographics, beliefs about others beliefs, partisan media, and social media. Much empirical work is consistent with biased beliefs driven by selection neglect, but rarely directly tests this mechanism outside of lab settings. We discuss how future research can provide more direct evidence.

show abstract

Section: Related Theoriesmentioning

confidence: 99%

Selection Neglect and Political Beliefs

Little¹,

Brundage²,

You³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…The combination of these two model features generates the social version of confirmation bias that we outlined earlier. Bowen, Dmitriev, and Galperti (2022) study a model where signals are selectively shared at most once with network neighbors, but agents are misspecified and partially neglect this selection. This bias leads to mislearning, and it also generates polarization in social networks with echo chambers.…”

Section: Related Literaturementioning

confidence: 99%

“…By contrast, we focus on rational agents who make endogenous sharing decisions in equilibrium. Bowen et al (2022) note that: "[...]the Internet has also brought an abundance of information, which should lead people to learn quickly and beliefs to converge (not diverge) according to standard economic models. "…”

Section: Related Literaturementioning

confidence: 99%

Learning from Viral Content

Dasaratha¹,

He²

2022

Preprint

View full text Add to dashboard Cite

We study learning on social media with an equilibrium model of users interacting with shared news stories. Rational users arrive sequentially and each observes an original story (i.e., a private signal) and a sample of predecessors' stories in a news feed, then decides which stories to share. The observed sample of stories depends on what predecessors share as well as the sampling algorithm, which represents a design choice of the platform. We focus on how much the algorithm relies on virality (how many times a story has been previously shared) when generating news feeds. Showing users more viral stories can increase information aggregation, but it can also generate steady states where most shared stories are wrong. Such misleading steady states selfperpetuate, as users who observe these wrong stories develop wrong beliefs, and thus rationally continue to share them. We find that these bad steady states appear discontinuously, and even a benevolent platform designer either accepts these misleading steady states or induces fragile learning outcomes in the optimal design.

show abstract

“…Papanastasiou (2020) shows that news virality can result from a traditional rational cascade logic that is also facilitated by a costly signal acquisition. Bowen et al (2021) show that selective sharing and selection neglect could result in belief polarization when agents learn from what others share, while Cheng and Hsiaw (2022) consider a signaling model in which uncertainty about a sender being benevolent or malevolent could lead to users' beliefs about an unknown state disagreeing in the long term. Finally, Acemoglu et al (2022) develop a model in which users have heterogeneous priors about an unknown state, and find that homophily introduces a trade-off between virality and the emergence of eco chambers.…”

Section: Introductionmentioning

confidence: 99%