Are Online News Consumers Sophisticated Enough to Detect Media Bias?

How to cite

Huang, Yihong; Meng, Juanjuan; Weng, Xi. 2021. "Are Online News Consumers Sophisticated Enough to Detect Media Bias?." AEA RCT Registry. AEARCTR-0007680. https://doi.org/10.1257/rct.7680-1.1

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Yihong Huang* , Juanjuan Meng , Xi Weng

* Primary investigator

Registry ID
AEARCTR-0007680
DOI
https://doi.org/10.1257/rct.7680-1.1
First registered
Last updated
Status
Completed
Trial dates
2021-04-20 – 2021-06-30
Countries
China ()
Registry record
http://www.socialscienceregistry.org/trials/7680

Abstract

This project examines whether news consumers are sophisticated enough to detect media bias using both observational and experimental data. The observational data from a Chinese micro-blogging platform suggests that users are more likely to repost "inconsistent" news, defined as news with attitudes different from the media outlet's usual attitudes. To explore whether this reposting pattern reflects strategic thinking, we run an online experiment that randomly varies whether news sources are revealed to news consumers and tests its effect on their inference about the importance of news, tendency of reposting etc.

Experimental design

We run an online experiment to test if news consumers take media bias into account when inferring the importance of news and considering whether to repost it. We randomly assign participants into two groups and show them different information related to the four pieces of news (two consistent with the media outlet's general attitudes, and two inconsistent ones).

- For those in the control group, they don't know which media outlet posted the news.
- For those in the Media names group, we reveal the media outlet that posted the news.
- For those in the Media Attitudes group, we include a short introduction about the general attitudes of the media outlets.
- For those in the Debias group, we additionally remind participants about the existence of media bias before they read news.

After reading each piece of news, participants answer the following questions:
- To what extent do you agree with the opinion expressed in the news above?
- Please guess the Baidu index of the news above. (Proxy for the importance of the news)
- What's the probability that you repost the news above?
- If you decide to repost the news above, will you add comments to show support/rejection?

Design details

Randomization method
Randomization done in office by a computer
Randomization unit
Individual
Clusters
The treatment is not clustered.
Observations
600 individuals in each round, 2 rounds in total.
Arms
In each round, we will have 600 individuals, with 150 participants in each treatment arm. In total, we will have 300 participants in each treatment arm.
Primary outcomes
Probability of reposting news, inference about the importance of the news, level of agreement with the news

Cited by

No citation data available for this registration yet.

Keywords