Do Chinese Trust State Media More?

• In Event: Poster Session: Political Communication

Abstract

Introduction

This research studies the level of trust of Chinese people on different media sources. The author wants to find out whether Chinese people tend to trust more if the same article is reported by a state-owned media, rather than a foreign media.

One of the many puzzles in political science is why non-democratic regimes could sustain, and from time to time, people living under the rule of the non-democratic regime believes in the discourse of the ruler and do not revolt and overthrow the regime. Chen and Yang (2017) suggests that censorship in media may have discredited foreign media in China. As a result, even given the access to foreign media, citizens living under the authoritarian regime would not visit the non-censored foreign media. Hence, it makes it possible for an authoritarian regime to eliminate intentions of revolts by shaping the belief of the citizens with state-owned media.

There are two shortcomings of the existing literature mainly featuring observational studies. First, different media write stories from different perspectives with different styles, even on the same topic. These differences are often correlated with the political leanings of these media both in China and the United States (Gentzkow and Shaprio, 2010; Qin, Stromberg and Wu, 2018). Second, people with media biases self-select into the media they prefer, and thus a simple regression of trust on media choice may overestimate the level of trust.

This project is one of the first few ones trying to overcome these two shortcomings with newly developed experimental design framework by Knox, Yamamoto, Baum and Berinsky (2018), which enables researchers to estimate the effect of self-selection and holding the contents and thus the style of the article constant.

Experimental Design

The author has conducted an online experiment in a southern city in China with subjects recruited offline with the help of local kindergarten heads. The reason that the author is specifically interested in Chinese parents is that political scientists have found evidence of inter-generational transmission of political ideologies (Jennings 1984). Parents with political ideologies influenced by the state media are likely to transmit these political ideas to their children. Further, the author is also interested in replicating Hetherington and Weiler (2009) questions in China, exploring the relationship between a citizen’s authoritarian personality, measured by parental style, and her political ideology.

To test whether the source rather than the content or style of a piece of information will determine the respondent's level of trust of the information, the author asks respondents to read the same piece of article labelled with different media sources, and then answer 5 questions measuring their level of trust on the piece of article they have just read, and 3 questions on their subjective evaluation of China's relative performance compared with other countries in the world, respectively how they believe China is ranking in K-12 education, average years of schooling and economic growth all over the world. The 5 questions measuring media trust are like those asked in Arceneaux et al (2012), while the 3 questions on the subjective evaluation of China’s relative performance are new. Demographic questions and political knowledge tests have also been added in a similar way as Huang (2015). The comparison in means of these questions will provide a good estimate for the level of trust of these respondents across different media. Methods proposed in Knox et al (2018) are be applied to estimate the effect of self-selection on level of trust.

Implementation

The author collects 7 news reports from 7 different media sources, namely New York Times, BBC, China Central Television Network, Netease.com, Phoenix Television Network, People's Daily, Global Times on a same issue – UK translates a Chinese math textbook and makes it available for use in British primary and junior schools. The author then composes a new article by synthesizing words and paragraphs from these 7 news reports.

To capture the effect of self-selection into preferred media sources, following Knox et al (2018), the author randomly draws 20% respondents from the pool, telling them that they will read a piece of news article and poll them on which source of information they would prefer to read among the 7 different sources. The article they read will be the one that the author composes but they will be told that the article is from the media source they choose. For the rest 80%, the author directly gives them the article he composes, and tell them the article is from media A, while the label A is randomly drawn from the 7 media sources.

The respondents are debriefed about the deception at the end of the survey that the news article that they have read is not a true and full replication of the article published on the media, links to the true article will be provided.

Author

• Shiyao Liu, Massachusetts Institute of Technology