|
Field
Abstract
|
Before
Fake news is a global phenomenon that continues to shape our political reality in significant ways. We design an experiment accounting for three prominent features of fake news: (i) Senders have superior knowledge compared to receivers, (ii) they have incentives to push receivers in a specific direction known by receivers, and (iii) receivers’ decisions whether to follow the sender’s messages depends on their prior beliefs and their ideological worldview.
In the role of receivers, participants are first asked whether they believe that unemployment and crime rates in a given U.S. state were higher under the Trump (T) or Obama (O) administration. They are then presented with a message from a sender—taken from a previous study—who suggests the correct answer. Participants are informed that senders knew the correct answer, were allowed to send either T or O, and received a bonus if the receiver they were matched with in the previous study answered T (in treatment TRUMP) or O (in treatment OBAMA). Participants themselves receive a bonus for answering correctly.
This setup allows us to examine how receivers respond to messages, depending on the senders’ incentives, their own prior beliefs, and their political attitudes. In additional treatments, we either elicit participants’ estimates of how frequently senders lie or provide them with explicit information on this frequency. This enables us to investigate how the (perceived or actual) lying frequency influences the likelihood that participants accept or reject the senders’ messages.
|
After
In many economically important environments, people receive information from sources whose interests are transparent. Such communication naturally invites skepticism because receivers understand that senders have incentives to influence their beliefs. However, skepticism should not be indiscriminate. When a sender communicates a message that contradicts the direction in which the sender benefits from influencing the receiver, the message becomes particularly credible because it is difficult to reconcile with purely opportunistic behavior.
In this paper, we ask whether receivers exploit this credibility of such incentive-incongruent messages or whether they continue to rely primarily on their own judgment, even when sender incentives imply that a message is unusually likely to be true?
Participants in our experiment act as receivers. They first answer factual questions comparing outcomes under the Obama and Trump administrations. They then observe a message from a sender in a previous study. Senders knew the correct answer, could send either possible message, and had a monetary incentive to induce either an "Obama" or a "Trump" answer. Receivers know the sender's incentive and earn a bonus for choosing the correct answer. This setup allows us to examine how receivers respond to messages, depending on the senders’ incentives and their own prior beliefs.
In additional treatments, we either elicit participants’ estimates of how frequently senders lie or provide them with explicit information on this frequency. This allows us to examine how underreaction to credible messages can be explained and mitigated.
|
|
Field
Additional Keyword(s)
|
Before
Decision making, lying, fake news, partisanship, type-I and type-II errors
|
After
Decision making, lying, fake news, type-I and type-II errors
|