The Effects of Television News Propaganda: Experimental Evidence From China

The Effects of Television News Propaganda:
Experimental Evidence from China∗
Jennifer Pan† Zijie Shao‡ Yiqing Xu§
Abstract
More than half of the world’s population live under authoritarian rule, where pro-
paganda is deployed to manipulate attitudes and behaviors at scale. Television re-
mains one of the most powerful forms of propaganda because audiences continue to
regard television as an authoritative source of information. We create videos that re-
alistically resemble Chinese state news propaganda and experimentally assess their
effects on the policy preferences of the Chinese public. We find that propaganda
moves respondents to adopt policy positions espoused in the video up to 48 hours
after exposure and appears to work through persuasion. In contrast to prior research,
effects do not vary based on individual predispositions or characteristics, which may
be due to the content and format of propaganda.
Keywords: propaganda, media, policy, China, experiment
Word Count: 8,544
∗
Our thanks to Andrew Chadwick, Lance Holbert, Rasmus Nielsen, and seminar participants at Washington
University in St. Louis for many helpful comments and suggestions.
†
Assistant Professor, Department of Communication, Stanford University. Email: jp1@stanford.edu.
‡
Associate Research Fellow and Postdoctoral Researcher, School of Government, Sun Yat-sen University.
Email: shaozijie@vip.sina.com.
§
Assistant Professor, Department of Political Science, Stanford University. Email: yiqingxu@stanford.edu.
Electronic copy available at: https://ssrn.com/abstract=3579148

Introduction
More than half of the world’s population live under authoritarian rule,1 where propaganda
is deployed to manipulate attitudes and behaviors at scale. Television—in traditional and
online streaming forms—remains one of the most powerful forms of propaganda available
to autocrats despite the increasing complexity of media ecosystems and fragmentation of
audiences (Chadwick 2017; Webster and Ksiazek 2012). Audiences continue to regard
television as an authoritative source of information. In China, a nationally representative
survey from 2015 showed the primary way Chinese people consume news is through tele-
vision (CUGS 2015). In Russia, one of the first things Putin did after assuming the presi-
dency in 2000 was to impose state control on all major television channels by co-opting,
imprisoning, or exiling owners of independent stations (Enikolopov et al. 2011). The Rus-
sian regime uses television to influence the political sphere (Nisbet et al. 2017; Oates and
McCormack 2010; Stoycheff and Nisbet 2017), and as of 2018, most Russians say that
television, which reaches more than 70% of the urban population, is their most important
source of information (Zakem et al. 2018). In Azerbajian, AzTV, the state-controlled na-
tional television channel, disseminates pro-government propaganda reaching more than
99.9% of the population.2 In Iran, more than 80% of the population considers the state-
controlled Islamic Republic of Iran Broadcasting to be the most authoritative source of
information, and Iran recently redoubled investment to upgrade television content and
technology.3
Given its continued prominence, does television propaganda work? If yes, how does
it work and how well does it work? In this paper, we examine these questions by experi-
mentally assessing the effects of Chinese television propaganda. We create short segments
of news videos in collaboration with industry experts that realistically resemble Chinese
state propaganda. We embed these videos in an online survey in China to examine effects
on policy preferences. Our results show that exposure to television propaganda moves
1
Regime classification based on Geddes et al. (2014) and population data from the World Bank
(https://data.worldbank.org/indicator/SP.POP.TOTL).
2
AzTV penetration rate from Azerbaijan’s Ministry of Communications and Information Technologies; see
https://bit.ly/2VNqB51 (Accessed April 16, 2019).
3
See https://bit.ly/2ICcjk3 (Accessed December 8, 2019).

respondents to adopt policy positions closer to the policy position espoused in the propa-
ganda, regardless of what that position is. Our results also show that propaganda spills
over to changed attitudes on related policy issues and suggest that effects persist up to 48
hours after exposure. The main result falls in line with a great deal of research showing the
effects of media and propaganda (Adena et al. 2015; Bleck and Michelitch 2017; Geddes
and Zaller 1989; Lasswell 1927; Nisbet and Myers 2011; Stockmann and Gallagher 2011;
Szostek 2017a; Yanagizawa-Drott 2014). However, it runs counter to recent research on
Chinese propaganda showing that state propaganda serves to signal state strength rather
than to shift attitudes (Huang 2015). These results suggest that while Chinese state pro-
paganda may signal state strength, it can also lead to attitude change.
More surprisingly, we find television propaganda to be effective regardless of indi-
vidual predispositions and consistently effective across individuals with differing socio-
economic and demographic characteristics. This finding runs counter to prevailing under-
standings that propaganda effects should vary by individual characteristics such as level
of education and by predisposition toward the topic of propaganda (Bleck and Miche-
litch 2017; Geddes and Zaller 1989; Gunther and Mughan 2000; Stockmann and Gal-
lagher 2011; Szostek 2017b; Zaller 1991). This finding may be explained by the type of
propaganda—narrow and issue-specific—and the video medium of television propaganda
that activates multiple sensory channels and elicits strong emotional and physiological re-
sponses (Houts et al. 2006; Kensinger and Schacter 2006; Lang et al. 1999, 2015; Okdie
et al. 2014). In other words, television propaganda focused on specific policy issues may
generate consistent changes in attitudes on those policies across individuals in ways that
text-only propaganda and/or propaganda that aims to increase general regime support does
not. In this experiment, we find no evidence that propaganda affects broad outcomes such
as regime trust and patriotism or self-reported behaviors such as complaint-making and
discussing government policies.
In terms of how propaganda works, our results provide support for the mechanism
of persuasion, that respondents are persuaded to change their policy preferences. The
experiment shows that very few respondents lack prior knowledge and opinions on the

policy issues contained in the experiment, which decreases the likelihood that their atti-
tudes are changing because they are learning new information. The questions asked in
the experiment are not politically sensitive, which decreases the potential for social de-
sirability bias. In addition, the method we use—an experiment embedded in an online
survey—minimizes incentives of respondents to provide the “right” answer (Mummolo
and Peterson 2019). While we cannot completely rule out learning or social desirability
bias as motivators for some respondents, the results are more consistent with persuasion.
This paper joins a growing body of research on political communication outside of
the U.S. and other developed, consolidated democracies while advancing research in this
area methodologically and substantively. We proceed in six sections. The next section
motivates our research question and hypotheses by discussing existing theories of propa-
ganda. We then provide background on television propaganda in China. We detail the
experimental design and survey and provide the results in the following two sections. We
conclude by discussing the implications of the findings.
Theories of Propaganda
Research on media effects and propaganda has generally found that consuming propa-
ganda leads to changes in attitudes, opinions, and behaviors (Adena et al. 2015; Bleck and
Michelitch 2017; Geddes and Zaller 1989; Gerber et al. 2009; Iyengar and Hahn 2009;
Iyengar et al. 1982; King et al. 2017; Lasswell 1927; Nisbet and Myers 2011; Stockmann
and Gallagher 2011; Szostek 2017a; Yanagizawa-Drott 2014). There is a greater diversity
of views, however, about how propaganda changes attitudes and what attitudes it changes.
Propaganda has been found to work through persuasion, where the provision of informa-
tion about an issue changes opinions and attitudes toward that issue (Chaffee 1977; Lass-
well 1927; Jowett and O’Donnell 2018). Propaganda has also been shown to work through
learning (Prior and Lupia 2008), where information provided by propagandists is novel
so that the public learns and forms new opinions as a result. Instead of changing attitudes
about the particular topic contained in propaganda, researchers have also shown that pro-
paganda can generate more generalized feelings of fear and guilt (Huang 2015; Wedeen

1998), resulting in attitudes and behaviors (e.g., conformity, preference falsification) that
suppress dissent and collective mobilization (Edmond 2013; Kuran 1991). Finally, propa-
ganda has also been theorized to affect, though not necessarily change, attitudes through
agenda-setting and priming the salience of certain issues (Iyengar et al. 1987; McCombs
2018).
In terms of how well propaganda works, the general consensus is that the effectiveness
of propaganda is conditional on individuals’ predispositions and characteristics, and the
credibility of the source (Gunther and Mughan 2000; Zaller 1991). Individual character-
istics may influence what information people consume and trust. Wojcieszak et al. (2019)
show that in Iran, individual characteristics such as religiosity are associated with media
choice. Moehler and Singh (2011) find an association between greater trust in govern-
ment media individual characteristics, such as lower levels of political sophistication and
illiberal attitudes. However, even controlling for exposure and source credibility, studies
have shown heterogenous effects of propaganda across individuals. In studying 1970s
Brazil, Geddes and Zaller (1989) find that exposure to propaganda increases mass support
for the values espoused by the regime, but individuals who have higher levels of politi-
cal awareness are less susceptible to propaganda. Studying Chinese government media
in the mid-2000s, Stockmann and Gallagher (2011) find that highly informed citizens are
more resistant to changing their political views after exposure to propaganda. Through
a field experiment conducted in the wake of the 2012 Mali coup, Bleck and Michelitch
(2017) show that propaganda boosted national identity but did not increase support for
the regime, and that propaganda had different effects on women, who are less politically
socialized, than men. Based on a 2016 survey of Ukrainians, Szostek (2017b) finds Rus-
sian propaganda in Ukraine to be more effective among individuals with stronger ties to
Russia, including those who rely more on Russian news sources as well as those with
personal and religious connections to Russia.
However, most research showing heterogeneous effects of propaganda focus on how
propaganda shifts diffuse yet complex preferences such as regime support. It is straight-
forward to ask a survey respondent for his/her level of support in a regime; however, that

simple response may be informed by a complex and multi-dimensional array of consid-
erations. This means that the same piece of propaganda may influence individuals with
one set of characteristics on a dimension related to regime support and influence others
with different characteristics on a separate dimension related to regime support, result-
ing in differing outcomes on regime support. In contrast, propaganda focused on specific
or single-policy issues may have more uniform effects across individuals with differing
characteristics, especially if individuals are not forced to make trade-offs between differ-
ent issues. In other words, existing research has focused on how individuals with different
predispositions and characteristics react to propaganda in different ways. However, propa-
ganda with different goals—e.g., increase regime support versus increase support for the
redistributive policies of the regime—may have more variable or more consistent effects.
Our primary research question is whether television propaganda can successfully shape
policy preferences of the public. Prior research on media and propaganda effect leads us
to believe that television propaganda will have a narrow, short-term direct effect in influ-
encing policy preference. Specifically, we expect that state television news could persuade
the public toward its espoused policy position on the issue under discussion. We do not
have strong priors as to whether state television will have spillover effects on the policy
position of related issues. We propose the following pre-registered4 hypotheses:
H1: People exposed to the regime’s position on a domestic policy issue will have policy
preferences closer to the regime’s positions on that issue than respondents in the
control condition.
H2: People exposed to the regime’s position on a foreign policy issue will have policy
preferences closer to the regime’s positions on that issue than respondents in the
control condition.
H3: People exposed to the regime’s position on a domestic policy issue will not have
policy preferences closer to the regime’s positions on related issues than respon-
dents in the control condition.
4
All hypotheses and analysis presented in this paper are pre-registered.

H4: People exposed to the regime’s position on a foreign policy issue will not have pol-
icy preferences closer to the regime’s positions on related issues than respondents
in the control condition.
We are also interested in the questions of how television propaganda works and how
well it works. To examine these questions, we conduct exploratory analysis of the effects
of propaganda given different pre-treatment measures of pre-disposition as well as for
different socioeconomic and demographic sub-groups.
Television Propaganda in China
The explicitly stated goal of propaganda in China is to shape public opinion in favor
of the Chinese Communist Party (CCP). The latest government document pertaining to
propaganda—the 2019 “Regulations of the Communist Party of China on Propaganda
Work”—describes propaganda as foundational to the strength of the CCP, and an impor-
tant way to shape the opinions of party leaders and party members as well as those of the
masses.5 The Propaganda Department of the CCP is the organization tasked with pro-
ducing and disseminating information to shape public opinions and behaviors. The CCP
Propaganda Department oversees a vast system of control that encompasses media, educa-
tion, culture, sports, and even non-governmental organizations and research (Cantoni et al.
2017; Clayton et al. 2006; King et al. 2013). Television broadcasting has been part of this
system since the late 1970s (Brady 2009; Lull 2013; Zhao 1998). The National Radio and
Television Administration, which operates the country’s predominant public broadcaster,
China Central Television (CCTV), reports directly to the Propaganda Department.
Television news is used by the Communist Party to shape public preferences on policy
issues. Shaping public opinion on policy issues is narrower than the broader goal of pro-
paganda in garnering public support and suppressing opposition. However, shaping public
attitudes on policy issues plays an important role in autocracies. Public opinion is gener-
ally considered less consequential in authoritarian regimes than in democracies because
the public cannot vote or easily sanction politicians for their policy decisions. However,
5
From “Regulations of the Communist Party of China on Propaganda Work” promulgated on August 2019
(中国共产党宣传工作条例). For details, see https://bit.ly/2R4ebVx (Accessed January 6, 2020).

public preferences can nonetheless constrain domestic and foreign policy making and im-
plementation in autocracies, including in China (Weeks 2008; Weiss 2013). This means
public opposition to policies may require authoritarian regimes to give ground. Shaping
public preference on policies preemptively can forestall opposition and the need for the
regime to give way.
China’s most-watched television news program is Xinwen Lianbo, a nightly broadcast
at 7pm that typically lasts for 30 minutes. Xinwen Lianbo covers domestic issues for 20
to 25 minutes and touches on foreign affairs in the remaining time. The Communist Party
aims to use Xinwen Lianbo to shape public opinions on domestic policy and foreign issues
in line with those of the regime by broadcasting the regime’s official stance, setting the
policy agenda, and framing political, economic, and social issues in line with the regime’s
interests (Zhao and Guo 2005). However, Huang (2015) notes that the propaganda goals
of Xinwen Lianbo are common knowledge and the program is often regarded as staid and
ineffective in shaping public opinions.
Xiwen Lianbo captures 55 to 60 million viewers each evening, which represents 4.6%
of the approximately 1.2 billion people in China with access to television,6 and addi-
tional viewers watch Xinwen Lianbo news segments online.7 As in the U.S. and West-
ern Europe, traditional mass media consumption in China is on the decline, as viewers
have much greater breadth of media choice (Graber and Dunaway 2017; McCombs 2018;
Nielsen and Sambrook 2016). However, in China, the regime has protected state television
through regulations that enable state broadcasters to monopolize viewers and dominate an
intensely competitive advertising market. For example, the regime guarantees monopoly
status for programs such as Xinwen Lianbo by requiring virtually all local TV channels
to broadcast it (Zhao and Guo 2005). In March 2019, the Xinwen Lianbo broadcast on
the channel CCTV-1 captured approximately 1.09% of households with access to televi-
sion and 4.98% of households that were watching television on weekdays between 7pm
and midnight. This puts CCTV-1’s Xinwen Lianbo broadcast 15th in terms of ratings for
6
March 2019 ratings and audience data from CSM Media Research; see http://www.csm.com.cn/cpfw/
(Accessed April 19, 2019).
7
Xinwen Lianbo segments are lived streamed at http://xinwenlianbo.tv/, and available online at http://tv.
cctv.com/lm/xwlb/, http://tv.cntv.cn/videoset/C10437, and YouTube (https://bit.ly/2Gu0xGR).

weekday evening television. However, because the Communist Party mandates that Xin-
wen Lianbo is simultaneously broadcast on 58 other channels,8 Xinwen Lianbo captured
4.61% of all households with television and 21.06% of weekday television viewers, taking
the number one rank in weekday evening television.9 As a result of this state-mandated
monopoly in airtime, CCTV leads China’s TV advertising market (PricewaterhouseCoop-
ers 2015), and CCTV’s financial strength allows it to remain relevant in a market with
changing demographics.
Methods
We worked closely with media professionals in China to create six short news video seg-
ments that closely resemble the content and style of Xinwen Lianbo. We embedded these
news segments in an online survey experiment to measure the effects of television propa-
ganda on policy preferences. We first describe the video segments, and then describe the
survey in which the videos were embedded.
Treatment conditions: The six news segments were all one to two minutes long. Three
of the segments were focused on the domestic policy issue of state-owned enterprise
(SOE) reform. Three of the video segments are focused on the foreign policy (FP) issue of
disputes in the South China Sea between China and the Philippines. We wanted to include
a domestic and a foreign policy issue, and we chose SOE reform and China-Philippines
relations because they are areas where the Chinese government has expressed different
policy positions in recent times. This diversity in the positions the government has taken
allowed us to assess the effect of television propaganda rather than the particularities of
8
These channels are (listed in order of Xinwen Lianbo viewership): CCTV-6, CCTV-3, CCTV-8, CCTV-
News, Hunan TV Cartoons, CCTV-14 (Kids), Beijing TV Kids, CCTV-4, CCTV-Music, CCTV-10, Hunan
TV, CCTV-12, CCTV-5, Beijing TV, CCTV-7, CCTV-2, Anhui TV, Jiangsu TV, Jiangxi TV, Shanghai
Oriental TV, Guizhou TV, Shandong TV, Guangdong TV Kids, Hubei TV, CCTV-9, Hubei TV, Shanghai
Toonmax, Guangdong TV, Sichuang TV, Shanxi TV, Jiangsu Youman Cartoon, Heilongjiang TV, Zhejiang
TV, Chongqing TV, Southeast TV, Henan TV, Liaoning TV, Tianjin TV, Qinghai TV, CCTV-10, Guangxi TV,
Shenzhen TV, Xinjiang TV, Liaoning TV, Inner Mongolia TV, Bingtuan TV, Yunnan TV, Jilin TV, Shandong
Education TV, Shaanxi TV, Xiamen TV, China Education, Amdo TV, Tibet TV, Travel TV, Guandong
Southern, China Education 2.
9
For comparison, the highest rated show broadcast on a single channel on weekday evenings in March
2019 is Wolf Warrior II, with 1.83% of households with TV access and 7.17% of TV viewers tuned in
during weekday evenings.

the message being transmitted.
State-owned enterprises in China account for approximately 30% of GDP and slightly
less than 20% of total employment.10 Unlike private firms, SOEs in China must serve the
political objectives of the government in addition to pursuing economic activities. SOE
reform centers around the question of how these commercial and political activities should
be delineated and separated. One policy stance the Chinese government has taken is that
market forces are the key to successful reform. This policy stance argues that market
competition is how state-owned firms can transform into independent and successful en-
terprises (market orientation). A different policy stance on SOE reform that the CCP has
taken argues that state control is the key to successful reform (state orientation).
Since the early 2000s, China and the Philippines have clashed over sovereignty of is-
lands and shoals in the South China Sea. Chinese policy has emphasized the need for co-
operation between the two countries in resolving tensions in the South China Sea (dove).
Chinese policy has also emphasized the need for China to defend its sovereignty and ter-
ritorial integrity (hawk).
The three videos on SOE reform captured the two policy positions the Chinese gov-
ernment has taken—market orientation treatment condition (market) and state leadership
treatment condition (state)—as well as a control condition where background and facts
related to SOE reform were provided, but the position of the Chinese state was not stated
(control). The three videos about China and the Philippines also captured the two pol-
icy stances—cooperation treatment condition (dove) and sovereignty treatment condition
(hawk)—as well as a control condition where facts about China-Philippines relations were
provided but no state position was given (control).
For videos that contained policy positions, the policy positions were drawn from orig-
inal Xinwen Lianbo transcripts, edited together for coherence while ensuring similarity
in length and types of content (see Appendix for English translations of all video tran-
scripts). Each video also contained interviews we filmed and edited with two real Chinese
academics who have expertise on the two policy issues. We included such interviews be-
cause they are frequently used in Xinwen Lianbo broadcasts. We recognize that including
10
See https://bit.ly/2JO37tj (Accessed May 18, 2019).

these interviews makes it more difficult for us to determine what element of the video
shaped outcomes; however, for this experiment, which we see as the first in a series of
related experiments, we leaned toward realism and maximizing effects. One academic
provided the explanations and rationales for both viewpoints related to SOE reform, as
well as background facts on SOE reform for the control condition. The second academic
provided the same for Sino-Philippine relations. By using the same expert in all treatment
conditions of the same topic and by filming and editing the interviews ourselves, we en-
sured that characteristics of the academics did not influence any outcomes we observed
(see Figure A1 for snapshots of the video segments).
Aside from the interviews, the videos included voiceovers by a broadcaster trained at
the Communication University of China (中国传媒大学), which produces CCTV an-
chors and reporters. The videos were fully captioned in Chinese. Video images came
from a free Chinese image library. A link to an example video in each policy area is avail-
able at https://bit.ly/2RrWlNw.11 The Appendix contains stills from the video segments.
English subtitles have been added to the examples, but were not in the videos shown to
respondents.
Survey sample: We recruited an online sample of 830 respondent from 26 provinces in

China. These provinces vary greatly in their level of development: 51.8% of the sample
came from rich provinces (based on 2017 per capita income), and the remainder from
poor provinces.12 This sample is not representative of China’s overall population, but we
used a quota sampling strategy to ensure that the sample was diverse in gender, age, and
educational background. We screened respondents to include only respondents between
the ages of 18 and 50, and for whom the video could load smoothly. Table A3 reports the
summary statistics of the sample.
Survey design: The overall flow of our survey is illustrated in Figure 1. After screening,
we first asked respondents to answer two sets of questions to measure their predisposition
11
We plan to anonymize these videos if the paper is accepted to protect the identities of the interviewees.
12
Rich provinces include Beijing, Tianjin, Inner Mongolia, Liaoning, Shanghai, Jiangsu, Zhejiang, Anhui,
Fujian, Shandong, Hubie, Guangdong, Hainan, and Chongqin.
10

F IGURE 1. S URVEY AND EXPERIMENTAL FLOW
Start Start
Pre-disposition
Randomized order
Randomized order
Market State Control Dove Hawk Control SOE Reform
Preferences
Manipulation Check Manipulation Check
SOE Reform Preferences Foreign Policy Preferences Foreign Policy
Preferences
Demographics
Other Outcomes
End 48 hours later End
on economic and foreign policies. One set contained seven questions related to pref-
erences on market economy—for example, whether private ownership and sale of land
should be allowed—where responses could range from preferences for free market and
private capital to preferences for state control of the economy and limits on the influence
of private capital (see Appendix Table A1 for question details). The other set contained
seven questions related to preference on national sovereignty, such as whether it is ap-
propriate to use military force to unify Taiwan with China. Here, responses ranged from
highly nationalistic to not nationalistic. Based on these two sets of questions, we con-
structed two simple additive indices as measures of respondent predisposition in the two
dimensions.13
After we measured predisposition, respondents watched two videos, one on each pol-
icy issue. The order in which the videos were watched was randomized, and randomiza-
tion for each video was independent. In other words, the assignments of the two treat-
ments were independent of each other, such that if a respondent was assigned the video
espousing market orientation for SOE reform, the respondent could have been assigned to
any of the three treatment conditions for the China-Philippines relations video with equal
probability. The order of the two policy areas was also randomized. Respondents were
13
Questions were drawn from a recent studyf that shows that policy preferences of the Chinese public mea-
sured this way have high inter-temporal stability (Pan and Xu 2020).
11

well balanced in terms of demographics (see Appendix Table A4 for the balance table).
As shown in Figure 1, immediately following each video, we conducted manipula-
tion checks by asking respondents two multiple-choice questions about the content of the
video they had just viewed. The first multiple-choice question following the SOE reform
video asked: “What share of total assets do SOEs account for?” and the second multiple-
choice question asked respondents to identify the main message of the video. After the
China-Philippines video, the first multiple-choice question asked: “What foreign country
was discussed in the video?” and the second multiple-choice again asked for the main
message of the video. We also assessed the construct validity of our treatments by asking
respondents whether they thought the source of the video was CCTV.
Policy preferences were measured after the manipulation checks. We had two types
of policy preference measures. First, we asked respondents whether they strongly agreed,
agreed, disagreed, or strongly disagreed with the policy position espoused in the video
(respondents could also choose that they were not sure).14 Then, we asked respondents to
explicitly select the statement that best reflected their preferences on each issue. For SOE
reform, respondents could select from three options: SOE reform should be led by the
state (state); SOE reform should be market driven (market), or how SOE reform occurs
does not matter as long as economic outcomes are good (neutral). This variable could take
one of three values: 1 (state), 2 (neutral), and 3 (market), meaning larger values denoted a
more pro-market position. For China-Philippines relations, respondents could select from
three options: China should cooperate with the Philippines and set aside territorial dis-
putes for regional peace (dove), China must defend its national sovereignty and territorial
integrity (hawk), or China’s foreign policy should change with changing circumstances
(neutral). This variable could also take one of three values: 1 (hawk), 2 (neutral), and 3
(dove), meaning larger values denoted a more dovish, cooperative position.
The second way we measured policy preferences was to ask respondents several ques-
tions in related policy domains. After the SOE reform video, we asked nine questions
about whether state-owned or private enterprises should receive preferential treatment
from the government and whether the state or market should drive the allocation of
14
See the exact questions and options in Table A2 in the Appendix.
12

resources—for example, whether SOEs should receive more preferential treatment than
private enterprises (see Appendix for sample list of questions). In line with the SOE
reform treatment as well as the predisposition measure of preferences on the market econ-
omy, responses to these questions varied from supporting free market to supporting state
control. However, these related policy questions focused on SOEs while the predisposi-
tion questions focused on a broader set of economic policies. We used these related policy
questions to construct a simple additive index of market-oriented reform. Similarly, after
the China-Philippines relations video, we asked nine questions about China’s foreign pol-
icy in Southeast Asia and China’s diplomatic strategy. Also in line with the foreign policy
treatment and the predisposition questions, these questions varied along the dove versus
hawk dimension (see Panel B of Table A2 for examples), but these related questions fo-
cused on a narrower aspect of national sovereignty than the predisposition questions. We
used these related questions to construct an additive index of dovish foreign policy.15 We
also asked respondents for demographic information, including their age, ethnicity, mari-
tal status, and income, as well as a number of other outcomes such as political knowledge
and political trust.
Finally, because research in media effects suggests that effects can quickly decay (Hill
et al. 2013), we conducted a follow-up survey 48 hours after the first survey to ask the
respondents about their understanding of the policy positions espoused in the video, their
position on SOE reform and China-Philippines relations, and their policy preferences on
the role of SOEs in the economy and foreign policy more generally.16
We embedded several mechanisms in the survey to maximize the quality of responses.
First, in the predisposition questions prior to treatment assignment, we embedded an at-
tention filter—e.g., we asked the respondents to select “strongly disagree” for a particular
question—and those who failed to do so were screened out of the survey. Second, we
15
The respondents were allowed to choose “I do not know” when answering these questions. We used
multiple imputation methods to fill in missing values based only on each set of nine questions (in other
words, no additional information from the rest of survey was used to impute missing data).
16
We conducted a “soft launch” at the beginning of the survey roll-out to make sure that the main survey
worked properly on our platform. We did not follow up with these respondents, who constituted about one-
third of the full sample. Prior to launch, we also conducted three pilots using off-line convenience samples
to test various aspects of the treatments and survey.
13

measured whether the birth year reported by the same respondent was consistent at the
beginning and the end of the survey. Third, we measured the amount of time respondents
spent watching the video and completing the entire survey to identify speeders.17 Re-
ducing the sample to respondents who passed the age-consistency test and who were not
speeders reduced the sample size from 830 to 762.18
Research ethics: The primary ethical concern in this paper, as with any experimental
study of media effects, was that the experiments would alter policy preferences in politi-
cally meaningful ways. Our substantive focus helped us address this concern because we
were interested in focusing on specific policy issues rather than on propaganda that aimed
to promote support for the regime, to strengthen state legitimacy, or to promote a cult of
personality. In addition, because we chose policy issues where the position of the state
had changed and because our experimental treatments contained these divergent policy
positions, our experiment does not promote any hegemonic political position.
A secondary ethical concern was that the respondents who participated in the survey
would experience retaliation if their responses deviated from the preferred policy positions
of the Chinese regime. We guarded against this threat by not collecting any personally
identifiable information, and by securely storing the data. We obtained approval from our
university IRBs for all aspects of the research.
Results
In line with H1 and H2, we find that exposure to video news content moves respondents
toward the espoused policy position of the state, regardless of what that position is and
regardless of the prior disposition of the respondent. Many of these effects persist up to 48
hours after exposure. Contrary to H3 and H4, video exposure also moves the expressed
policy positions of respondents on related issues. We begin by showing the manipulation
check before moving to the main results, followed by analysis of heterogeneous effects
17
Speeders are defined as those who spent less than 10 minutes completing the survey. The median finish
time was 17 minutes.
18
Not surprisingly, the estimated effect became stronger when we focused on the sub-sample of respondents
who were more attentive when answering the survey.
14

and the 48-hour follow-up.
Manipulation checks: The video treatments have high construct validity. Most respon-
dents, 86%, thought our video segments came from CCTV. The manipulation checks show
that respondents were able to correctly answer actual questions about the videos. Columns
(1) and (4) of Table 1 show that most of the respondents (83.4% for SOE reform, 97.5%
for China-Philippines relations) correctly answered the factual questions after watching
the two videos.19
TABLE 1. M ANIPULATION C HECK
Identify Identify Identify Identify

Pass SOE SOE media SOE media Pass FP FP media FP media
manipulation content as content as manipulatio content as content as
check pro-market pro-state n check dovish hawish
Outcomes (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1)
(1) (2) (3) (4) (5) (6)
Treatment1: Market 0.060* 0.510*** -0.049* 0.011 -0.038 0.029

(0.032) (0.037) (0.027) (0.014) (0.034) (0.022)
Treatment1: State 0.020 0.080** 0.477*** 0.011 -0.031 0.015
(0.033) (0.038) (0.036) (0.014) (0.036) (0.021)
Treatment2: Dove -0.077** -0.024 0.023 -0.001 0.501*** -0.021

(0.032) (0.038) (0.032) (0.012) (0.036) (0.014)
Treatment2: Hawk -0.040 -0.033 0.048 -0.011 -0.183*** 0.127***
(0.030) (0.038) (0.033) (0.014) (0.037) (0.025)
Outcome Variable Mean:

∼Full sample 0.834 0.443 0.278 0.975 0.460 0.075
∼Treatment1 = Neutral 0.807 0.245 0.135 0.967 0.478 0.062
Observations 830 830 830 830 830 830
R-squared 0.012 0.203 0.280 0.002 0.334 0.064
Note: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1.
In terms of comprehending the policy position espoused in the video, the vast majority
of respondents were able to correctly identify the policy position of most of the videos.
Columns (2) and (3) of Table 1 show that 75.5% of respondents who were assigned to
the pro-market condition identified the SOE media content as pro-market (as opposed to
24.5% under the neutral condition), and 61.2% of those who were assigned to the pro-
19
Because treatment was assigned at the individual level, we use robust standard errors throughout this paper.
Clustering standard errors at the provincial level gives almost identical substantive results.
15

state condition identified the media content as pro-state (as opposed to 13.5% under the
neutral condition). Column (5) shows that 85.7% of the respondents who were assigned
to the dove condition identified the media content as dovish (as opposed to 35.6% under
the neutral condition). However, as shown in column (6), only 16.6% of those who were
assigned to the hawk condition identified the media content as hawkish (as opposed to
3.9% under the neutral condition).
Main effects: Respondents adopted policy positions closer to the policy position es-
poused in the propaganda video, regardless of what that position was. Figure 2 presents
the estimated treatment effects on preferences toward SOE reform. The outcome in the
F IGURE 2. T REATMENT E FFECTS ON P OLICY P REFERENCES : SOE R EFORM
Pro−market Pro−market
0.6
0.6
0.4
0.4
● ●
0.2
0.2
Coefficients
Coefficients
0.0
0.0
● ● ●
●
−0.2
−0.2
●
−0.4
−0.4
−0.6
−0.6
Pro−state Pro−state
T1: Market T1: State T2: Dove T2: Hawk T1: Market T1: State T2: Dove T2: Hawk
(a) Agree to Market Reform (b) An Index of Market-oriented Reform
left panel is a three-value indicator of direct policy preference (a bigger value means a
more pro-market approach), and the outcome in the right panel is a standardized index
of market-oriented reform (with 0 mean and 1 standard deviation) based on nine related
policy questions. The thick and thin bars in both plots represent 95% and 99% confidence
intervals for the point estimates, respectively (same for subsequent Figures 3 and 6). We
estimated the effects using regressions, controlling for a full set of individual character-
istics and provincial fixed effects (see Table A6 in the Appendix for the full regression
table).20 In line with H1, we see that the pro-market video content on average shifted the
20
The control variables include the respondent’s gender, age, age squared, level of education, ethnic minority,
religiosity, marital status, working experience, Communist Party membership, experience using a virtual
16

respondents’ stated preference on SOE reform toward a more market-oriented approach.
On the direct question (left panel of Figure 2), there is a shift of approximately 0.3 points
on a 1-3 point scale. In contrast to H3, we see that pro-market video content also shifted
respondents approximately 0.3 standard deviations toward a more market-oriented posi-
tion on related policy issues (right panel of Figure 2). Also in line with H1, pro-state
video content reduced the respondents’ preference for market-oriented reform by roughly
0.3 points based on the direct question (left panel of Figure 2), and 0.2 standard devia-
tion based on the policy index (right panel of Figure 2). It is worth noting that both the
dove and hawk treatments shifted respondents’ preference on SOE reform toward a more
state-led approach (see dove and hawk estimates in Figure 2), compared with the neutral
foreign policy condition. This is likely because a video about foreign policy may remind
respondents of the central role of the state in national interests.
Figure 3 shows the estimated treatment effects on policy preferences regarding China-
Philippines relations in the South China Sea. The left panel shows direct policy prefer-
F IGURE 3. T REATMENT E FFECTS ON P OLICY P REFERENCES : S OUTH C HINA S EA
Dovish Dovish
0.6
0.6
0.4
0.4
0.2
0.2
●
●
Coefficients
Coefficients
●
0.0
0.0
● ● ●
−0.2
−0.2
●
●
−0.4
−0.4
−0.6
−0.6
Hawkish Hawkish
T1: Market T1: State T2: Dove T2: Hawk T1: Market T1: State T2: Dove T2: Hawk
(a) Agree to More Cooperation (b) An Index of Dovish Foreign Policy
ences (a bigger value means a more dovish approach), and the right panel shows a stan-
dardized index of dovish foreign policy (with 0 mean and 1 standard deviation) based on
nine related policy questions. The full regression results are presented in Table A7 in the
Appendix. From Figure 3, we see that the dovish message on China-Philippines relations
private network, English proficiency, income level, and self-reported socioeconomic status.
17

on average shifted the respondents’ stated preference on the issue of China-Philippines re-
lations toward a more dovish, cooperative approach by around 0.15 points on a 1-3 point
scale based on the direct question (left panel of Figure 3) and around 0.19 standard devia-
tion based on the policy index (right panel of Figure 3). In contrast, the hawkish message
reduced the respondents’ support for the dovish approach also by roughly 0.3 points based
on the direct question (left), and 0.26 standard deviation based on the policy index (right).
Policy preferences on foreign policy did not appear to be affected by the content of the
video on SOE reform.
Treatment effects and predisposition: Predisposition does not appear to moderate the
effects of television news propaganda. We measured predisposition toward economic
policies and national sovereignty before exposing respondents to the video treatments.
We find strong correlation between predisposition and post-treatment policy preferences
in both the economic and the sovereignty dimensions. This suggests, first, that many re-
spondents have relatively coherent policy preferences. Second, this suggests respondents
are not simply answering questions to satisfy the researchers. Strikingly, regardless of
predisposition, exposure to the video segments changes policy preferences.
Figure 4 shows the predicted outcomes of three sub-groups: the group that viewed
the pro-market video (left panel), the control group (central panel), and the group that
viewed the pro-state video (right panel). In each group, we see a strong, positive corre-
lation between predisposition for a free market economy (x-axis values) and support for
a more market-oriented approach to SOE reform (y-axis values). However, regardless of
predisposition, we see that the pro-market video increases support for a market-oriented
approach to SOE reform. We can see this by comparing the left and central panels of
Figure 4, where average support for market reform is higher at all values of predisposition
in the market treatment (left panel) than in the control group (central panel). The pro-state
video decreases support for a market-oriented approach to SOE reform across the board.
We observe this by comparing the central and right panels of Figure 4, where average
support for market reform is lower at all values of predisposition in the state treatment
18

F IGURE 4. P REDICTED O UTCOME AND P REDISPOSITION TOWARD M ARKET
E CONOMY
G IVEN T REATMENT 1
T1: Market T1: Control T1: State
2.5 2.5 2.5

Agree to Market Reform
2.0 2.0 2.0
1.5 1.5 1.5
0 25 50 75 100 0 25 50 75 100 0 25 50 75 100
State Control <− Predisposition on Economy −> Free Market
Note: The histogram at the bottom of each plot shows the number of respondents under each treatment
condition at different values of predisposition (in percentile).
group (right panel) than in the control group (central panel).21 There are no interactive
effects between predisposition and exposure to video propaganda on economic policy.
Figure 5 shows predicted outcome of the group that viewed the dovish video (left
panel), the control group (central panel), and the group that viewed the hawkish video
(right panel) on China-Philippine relations. Again, we see a positive correlation between
predisposition for more nationalistic/hawkish views toward sovereignty (x-axis values)
and support for hawkish foreign policy in all three groups (y-axis values). Again, regard-
less of predisposition, the hawkish video increases support for hawkish foreign policy
(comparing the right and central panels of Figure 5) and the dovish video increases sup-
port for dovish foreign policy across the board (comparing the left and central panels of
Figure 5). This means that in the foreign policy domain, we also do not observe interactive
effects between predisposition and exposure to video propaganda.
Treatment effects and demographics: The effects of propaganda do not vary in sta-
tistically significant ways for individuals with different socioeconomic and demographic
characteristics. Using the same regression specification controlling for individual char-
acteristics and provincial fixed effects, we conducted subgroup analyses and explored
21
The differences between curves in the left and central panels and between the central and right panels
of Figure 4 can be understood as the marginal effects of the treatments on the outcome conditioning on
predisposition. See Appendix Figure A2 for the marginal effects plot.
19

F IGURE 5. P REDICTED O UTCOME AND P REDISPOSITION TOWARD NATIONAL
S OVEREIGNTY
G IVEN T REATMENT 2
T2: Dove T2: Control T2: Hawk
2.0 2.0 2.0

Agree to More Cooperation
1.5 1.5 1.5
1.0 1.0 1.0
0 25 50 75 100 0 25 50 75 100 0 25 50 75 100
Hawkish <− Predisposition on Sovereignty −> Dovish
Note: The histogram at the bottom of each plot shows the number of respondents under each treatment
condition at different values of predisposition (in percentile).
heterogeneous treatment effects.

In Figure 6, the left panel plots the coefficients of pro-market (black) and pro-state
(gray) conditions on the index of preference for market-oriented reform and the right panel
plots the coefficients of dove (black) and hawk (gray) conditions on the index of prefer-
ence for dovish foreign policy. We see that the two sets of treatments change respondents’
policy preferences across almost all subgroups. Notably, those who were presumably of
higher socioeconomic status (people with higher education and those with higher income)
were equally affected by the videos as others of lower socioeconomic status.
This result contrasts with prior research showing how higher status and knowledge in-
dividuals are more resistant to propaganda. However, this may be because we are examin-
ing effects of policy-specific propaganda on narrow policy preferences.22 Indeed, we also
explored treatment effects on other, more diffuse and complex outcomes—including mea-
sures of patriotism and nationalism, efficacy, trust in the central and local governments,
and self-reported political behavior such as complaint-making, voting in local National
People’s Congress elections, and discussing government policies. We find no evidence
that the video treatments affected any of these outcomes (see Table A9 in the Appendix
for the details).
22
Although we examine spillover on related policy areas, the effects we examine remain in one policy
domain.
20

F IGURE 6. T REATMENT E FFECTS ON P OLICY P REFERENCES : S UBGROUP A NALYSIS
T1: State ● T1: Market T2: Hawk ● T2: Dove
Full sample ● Full sample ●
Female ● Female ●
Male ● Male ●
Age 30 or below ● Age 30 or below ●
Age above 30 ● Age above 30 ●
Below college ● Below college ●
College or above ● College or above ●
Low income ● Low income ●
High income ● High income ●
Private sector ● Private sector ●
Public sector ● Public sector ●
−1.0 −0.5 0.0 0.5 1.0 −1.0 −0.5 0.0 0.5 1.0
Coefficients Coefficients
(a) SOE Treatments (T1) on Market Index (b) FP Treatments (T2) on Dove Index
We do not find differences in the amount of attention respondents of various subgroups

paid to the content of the videos, so this this pattern is not likely driven by differences in
attention.
Robustness checks: We conducted various robustness checks by pruning the data based
on different criteria. Table A8 presents the results. The outcome variables in Panel A and
B are the two indices of policy preferences. Column (1) shows the benchmark result
using the full sample. In columns (2) through (6), we limit the sample to respondents who
satisfied various conditions, including those who identified the source of the videos as
CCTV (86% of all respondents), those who passed the age consistency filter and were not
speeders (92% of all respondents), those who answered both factual questions correctly
(83% of all respondents), those who agreed the policy toward SOE reform is correct (94%
of all respondents), and those who agreed the policy toward the South China Sea is correct
(96% of all respondents). The results remained substantively the same, and in some cases,
the effects became even stronger.
21

Results from the 48-hour follow-up survey: We sent out a follow-up survey to 556
respondents 48 hours after they had completed the initial survey.23 We were able to reach
353 of them, achieving a recontact rate of 63%. Although the profiles of respondents
who completed the follow-up survey appear to be mostly balanced across the treatment
conditions (see Table A5), the potential for endogenous attrition, that the 37% of respon-
dents who did not respond to the follow-up survey may have done so for reasons related
to the first round of the survey, exists and could generate bias in the casual estimate of the
follow-up survey. Therefore, results based on the follow-up sample should be taken with
a grain of salt.
Table 2 presents the results of manipulation checks conducted in the second survey.
Columns (1) and (4) indicate that 48 hours after the videos were shown to the respondents,
most of the respondents could correctly answer the factual question regarding the two
issues. Moreover, columns (2), (3), (5), and (6) of Table 2 suggest that a significant
proportion of them could still identify the main position of the two videos.
TABLE 2. M ANIPULATION C HECK AFTER 48 H OURS
Identify Identify Identify FP Identify FP

Pass SOE SOE media SOE media Pass FP media media
manipulation content as content as manipulation content as content as
check pro-market pro-state check dovish hawish
Wave 2 Outcomes: (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1)
(1) (2) (3) (4) (5) (6)
Treatment1: Market 0.001 0.386*** -0.047 0.027 -0.102* 0.030

(0.030) (0.062) (0.051) (0.043) (0.062) (0.042)
Treatment1: State 0.001 0.028 0.412*** 0.031 -0.071 -0.042
(0.029) (0.065) (0.058) (0.044) (0.065) (0.037)
Treatment2: Dovish -0.010 -0.169*** 0.107* 0.052 0.294*** 0.037

(0.029) (0.064) (0.056) (0.041) (0.064) (0.031)
Treatment2: Hawkish -0.022 -0.089 0.038 0.036 -0.113* 0.215***
(0.032) (0.065) (0.055) (0.045) (0.067) (0.045)
Outcome Variable Mean:

∼Full sample 0.944 0.463 0.302 0.890 0.497 0.116
Observations 353 353 353 353 353 353
R-squared 0.133 0.181 0.252 0.085 0.182 0.187
23
The other respondents were part of our soft launch.
22

Figure 7 is a visual representation of the mean outcome variables for the survey con-
taining the propaganda videos (wave 1) and the 48-hour follow-up survey. The left panel
F IGURE 7. P ERSISTENCE OF THE T REATMENT E FFECTS

3.0
2.0
●
1.8
●
● ●
2.5
Agree to More Cooperation

Agree to Market Reform
●
●
1.6
●
●
2.0
1.4
●
●
●
1.5
1.2
Control Control
Market Dove
State Hawk
1.0
1.0
Wave 1 Wave 2 Wave 1 Wave 2
(a) SOE Reform (b) China-Philippines Relations in South China Sea
Note: The hollow dots represent outcome variable means in respective treatment groups for the full sample
(wave 1 only); outcome variable means in respective treatment groups for the subset of respondents who
participated in both waves are represented by solid dots (wave 1) and arrows (wave 2), respectively.
of Figure 7 shows that the wave 1 views of respondents who also completed the 48-hour
follow-up (solid circle) persist in the 48-hour follow-up (solid triangle). Respondents
exposed to the pro-market SOE reform video (dashed blue line) remain strongly support-
ive of market-oriented reform of SOEs, while those exposed to the pro-state SOE reform
video (dot-dashed red line) remain less supportive of market-oriented SOE reform. The
right panel of Figure 7 shows similar trends of persisting views for the China-Philippines
relations videos, with those exposed to the dovish video remaining more supportive of co-
operation and those exposed to the hawkish video less supportive of cooperation. These
results are generally similar when we use the index of preferences for market-oriented
reform and South China Sea (see Table A10 for full regression tables).
Discussion
In this paper, we conducted an experiment to study the effect of television news propa-
ganda on the stated policy preferences of ordinary Chinese people. We created six short
23

video segments covering two policy issues that closely resembled China’s Xinwen Lianbo.
We found that exposure to the videos moved people to adopt stated policy positions closer
to the espoused policy position of the state, regardless of what that position was. Expo-
sure also led to changes closer to the espoused policy position on related policy issues,
and these effects persisted up to 48 hours after exposure.
Surprisingly, we found that the news propaganda videos affected respondents regard-
less of their predisposition toward economic policies and issues of national sovereignty,
and our results hold across various social-economic and demographic sub-groups. The
lack of heterogeneous effects runs counter to prevailing understandings of propaganda.
However, this result may be driven by the fact that the type of propaganda we mimic is
narrow and issue-specific and the outcome we measure is also narrow and issue-focused.
We do not find evidence that our treatment has broader effects in changing attitudes toward
the regime or changing self-reported political behaviors.
Our results suggest that attitudes are more likely changing through persuasion than
because of learning or preference falsification. First, the fact that respondents’ predispo-
sitions are highly correlated with expressed policy preferences after treatment exposure
suggests that falsified preference are not driving the results. If our findings are purely
driven by falsified preferences, we should expect to see low correlation between the pre-
disposition of the respondents and the policy outcomes. As shown in Figures 4 and 5, this
is not the case. Second, the fact that preferences are changed by exposure to treatment
regardless of predisposition also suggests that for at least some of the respondents, the
propaganda news videos were persuasive. Learning is more likely to take place when an
individual is exposed to an issue for the first time and/or when the individual does not
have formed preferences on that issue (Prior and Lupia 2008). Respondents in our survey
had an option of saying “I don’t know” for all predisposition questions, but very few re-
spondents selected this option. This suggests that most respondents had prior preferences
on issues we covered, such that it is less likely respondents learned about these policies
and then formed preferences after watching the videos.
Finally, it is important to note that from the perspective of an authoritarian regime, the
24

existence of an effect is likely more important than the mechanism generating that effect.
An authoritarian regime wants the public to support, rather than oppose, its positions. It
may not matter whether the regime obtains support through persuasion, learning, or even
if some people are falsifying their preferences. Conformity—with or without true belief—
reduces opposition to authoritarian rule (Arendt 1959; Havel 1985; Kuran 1997).24 This
experiment shows how television propaganda can generate conformist attitudes among
individuals with differing dispositions and characteristics.
24
Preference falsification may also make mobilization and revolution more unpredictable, but in the absence
of revolutionary bandwagons or information cascades (Kuran 1991; Lohmann 2000), mobilization against
the regime is in general more difficult.
25

References
Adena, M., R. Enikolopov, M. Petrova, V. Santarosa, and E. Zhuravskaya (2015). Ra-

dio and the rise of the nazis in prewar germany. The Quarterly Journal of Eco-
nomics 130(4), 1885–1939.
Arendt, H. (1959). The Origins Odf Totalitarianism. Meridian Books.
Bleck, J. and K. Michelitch (2017). Capturing the airwaves, capturing the nation? a
field experiment on state-run media effects in the wake of a coup. The Journal of
Politics 79(3), 873–889.
Brady, A.-M. (2009). Marketing dictatorship: Propaganda and thought work in contem-
porary China. Rowman & Littlefield Publishers.
Cantoni, D., Y. Chen, D. Y. Yang, N. Yuchtman, and Y. J. Zhang (2017). Curriculum and
ideology. Journal of Political Economy 125(2), 338–392.
Chadwick, A. (2017). The hybrid media system: Politics and power. Oxford University
Press.
Chaffee, S. H. (1977). Mass media effects: New research perspectives. Communication
research: A half-century appraisal, 210–241.
Clayton, R., S. J. Murdoch, and R. N. Watson (2006). Ignoring the great firewall of china.
In International Workshop on Privacy Enhancing Technologies, pp. 20–35. Springer.
CUGS (2015). China urban governance survey. Technical report, Tsinghua University
Research Center on Data and Governance.
Edmond, C. (2013). Information manipulation, coordination, and regime change. Review
of Economic Studies 80(4), 1422–1458.
Enikolopov, R., M. Petrova, and E. Zhuravskaya (2011). Media and political persuasion:
Evidence from russia. American Economic Review 101(7), 3253–85.
Geddes, B., J. Wright, and E. Frantz (2014). Autocratic breakdown and regime transitions:
A new data set. Perspectives on Politics 12(2), 313–331.
Geddes, B. and J. Zaller (1989). Sources of popular support for authoritarian regimes.
American Journal of Political Science, 319–347.
Gerber, A. S., D. Karlan, and D. Bergan (2009). Does the media matter? a field experiment
26

measuring the effect of newspapers on voting behavior and political opinions. American
Economic Journal: Applied Economics 1(2), 35–52.
Graber, D. A. and J. Dunaway (2017). Mass media and American politics. Cq Press.
Gunther, R. and A. Mughan (2000). The political impact of the media: A reassessment.
Democracy and the media: A comparative perspective, 402–447.
Havel, V. (1985). The Power of the Powerless: Citizens Against the State in Central
Eastern Europe. Routledge.
Hill, S. J., J. Lo, L. Vavreck, and J. Zaller (2013). How quickly we forget: The duration
of persuasion effects from mass communication. Political Communication 30(4), 521–
547.
Houts, P. S., C. C. Doak, L. G. Doak, and M. J. Loscalzo (2006). The role of pictures
in improving health communication: a review of research on attention, comprehension,
recall, and adherence. Patient education and counseling 61(2), 173–190.
Huang, H. (2015). Propaganda as signaling. Comparative Politics 47(4), 419–444.
Iyengar, S. and K. S. Hahn (2009). Red media, blue media: Evidence of ideological
selectivity in media use. Journal of Communication 59(1), 19–39.
Iyengar, S., D. R. Kinder, et al. (1987). News that matters: Agenda-setting and priming in
a television age. News that Matters: Agenda-Setting and Priming in a Television Age.
Iyengar, S., M. D. Peters, and D. R. Kinder (1982). Experimental demonstrations of
the “not-so-minimal” consequences of television news programs. American Political
Science Review 76(4), 848–858.
Jowett, G. S. and V. O’Donnell (2018). Propaganda & persuasion. Sage Publications.
Kensinger, E. A. and D. L. Schacter (2006). Processing emotional pictures and words:
Effects of valence and arousal. Cognitive, Affective, & Behavioral Neuroscience 6(2),
110–126.
King, G., J. Pan, and M. E. Roberts (2013). How censorship in china allows government
criticism but silences collective expression. American Political Science Review 107(2),
326–343.
King, G., B. Schneer, and A. White (2017). How the news media activate public expres-
27

sion and influence national agendas. Science 358(6364), 776–780.
Kuran, T. (1991). Now out of never: The element of surprise in the east european revolu-
tion of 1989. World politics 44(1), 7–48.
Kuran, T. (1997). Private truths, public lies: The social consequences of preference
falsification. Harvard University Press.
Lang, A., R. L. Bailey, and S. R. Connolly (2015). Encoding systems and evolved message
processing: Pictures enable action, words enable thinking.
Lang, A., R. F. Potter, and P. D. Bolls (1999). Something for nothing: Is visual encoding
automatic? Media Psychology 1(2), 145–163.
Lasswell, H. D. (1927). Propaganda technique in the world war. Ravenio Books.
Lohmann, S. (2000). Collective action cascades: An informational rationale for the power
in numbers. Journal of Economic Surveys 14(5), 655–684.
Lull, J. (2013). China turned on: Television, reform and resistance. Routledge.
McCombs, M. (2018). Setting the agenda: Mass media and public opinion. John Wiley
& Sons.
Moehler, D. C. and N. Singh (2011). Whose news do you trust? explaining trust in private
versus public media in africa. Political Research Quarterly 64(2), 276–292.
Mummolo, J. and E. Peterson (2019). Demand effects in survey experiments: An empiri-
cal assessment. American Political Science Review 113(2), 517–529.
Nielsen, R. K. and R. Sambrook (2016). What is happening to television news? Digital
news project, Reuters Institute.
Nisbet, E. C., O. Kamenchuk, and A. Dal (2017). A psychological firewall? risk percep-
tions and public support for online censorship in russia. Social Science Quarterly 98(3),
958–975.
Nisbet, E. C. and T. A. Myers (2011). Anti-american sentiment as a media effect? arab
media, political identity, and public opinion in the middle east. Communication Re-
search 38(5), 684–709.
Oates, S. and G. McCormack (2010). Developments in Russian Politics 7, Chapter The
Media and Political Communication, pp. 118–134. Duke University Press.
28

Okdie, B. M., D. R. Ewoldsen, N. L. Muscanell, R. E. Guadagno, C. A. Eno, J. A. Velez,
R. A. Dunn, J. O’Mally, and L. R. Smith (2014). Missed programs (you can’t tivo
this one): Why psychologists should study media. Perspectives on Psychological Sci-
ence 9(2), 180–195.
Pan, J. and Y. Xu (2020). The strength of items: Measuring ideology in china using
surveys.
PricewaterhouseCoopers (2015). China’s entertainment and media handbook.
https://pwc.to/2VOWqKZ.
Prior, M. and A. Lupia (2008). Money, time, and political knowledge: Distinguishing
quick recall and political learning skills. American Journal of Political Science 52(1),
169–183.
Stockmann, D. and M. E. Gallagher (2011). Remote control: How the media sustain
authoritarian rule in china. Comparative Political Studies 44(4), 436–467.
Stoycheff, E. and E. C. Nisbet (2017). Priming the costs of conflict? russian public
opinion about the 2014 crimean conflict. International Journal of Public Opinion Re-
search 29(4), 657–675.
Szostek, J. (2017a). News consumption and anti-western narratives in russia: A case study
of university students. Europe-Asia Studies 69(2), 284–302.
Szostek, J. (2017b). The power and limits of russia’s strategic narrative in ukraine: The
role of linkage. Perspectives on Politics 15(2), 379–395.
Webster, J. G. and T. B. Ksiazek (2012). The dynamics of audience fragmentation: Public
attention in an age of digital media. Journal of communication 62(1), 39–56.
Wedeen, L. (1998). Acting “as if”: symbolic politics and social control in syria. Compar-
ative Studies in Society and History 40(3), 503–523.
Weeks, J. L. (2008). Autocratic audience costs: Regime type and signaling resolve. In-
ternational Organization 62(1), 35–64.
Weiss, J. C. (2013). Authoritarian signaling, mass audiences, and nationalist protest in
china. International Organization 67(1), 1–35.
Wojcieszak, M., E. C. Nisbet, L. Kremer, G. Behrouzian, and C. Glynn (2019). What
29

drives media use in authoritarian regimes? extending selective exposure theory to iran.
The International Journal of Press/Politics 24(1), 69–91.
Yanagizawa-Drott, D. (2014). Propaganda and conflict: Evidence from the rwandan geno-
cide. The Quarterly Journal of Economics 129(4), 1947–1994.
Zakem, V., P. Saunders, U. Hashimova, and P. K. Frier (2018). Mapping russian media
network: Media’s role in russian foreign policy and decision-making. Technical report,
CNA Analysis and Solutions.
Zaller, J. (1991). Information, values, and opinion. American Political Science Re-
view 85(4), 1215–1237.
Zhao, Y. (1998). Media, market, and democracy in China: Between the party line and the
bottom line, Volume 139. University of Illinois Press.
Zhao, Y. and Z. Guo (2005). Television in china: History, political economy, and ideology.
A Companion to Television, 521–539.
30

A. Appendix
Table of Contents
1. Snapshots of the video segments
2. Transcripts of video segments
3. Predisposition questions
4. Related policy questions
5. Summary statistics
6. Balance table
7. Main results
8. Robustness checks of the main results
9. Treatment effects on other outcomes
10. Marginal effects of the treatments
11. Visual illustration of the follow-up results
12. Variable definitions
A-1

A.1. Video Snapshots
F IGURE A1. V IDEO S NAPSHOTS
(a) On SOE Reform
(b) On South China Sea
A-2

A.2. Video Transcripts (English Translations)
SOE: Market In 2014 Ministry of Finance data showed that there were 15 million
enterprises nationwide, and state-owned accounted for less than 1% of enterprises but
assets controlled by state-owned enterprises accounted for at least 30% of total assets. As
a crucial component to economic reform and development, the evolution of state-owned
enterprises plays a decisive role in the development of China’s national economy.
[SOE Expert Interview] The biggest problem for most SOEs is that they bear a lot
of policy burdens because they need to bear a lot of responsibilities and you don’t know
how much policy burden they bear. It is uncertain how much subsidies the government
should give them. Their optimal choice may be to ask the government for more subsidies,
but then they have no incentive to improve their overall performance and efficiency. If
SOE managers and employees do not perform well, you will find that SOEs performance
will be worse than that of private enterprises. Government departments can measure the
performance of SOEs using relative measures, and use this as a basis to provide incentives
to SOE managers.
The key to the SOE reform is to ensure it operates in a market-oriented manner, to
increase its competitiveness in the market such that the market can play a decisive role in
resource allocation. We must break barriers in institutional obstacles, improve the regu-
latory system, follow market forces, reduce burdens, strengthen health, improve quality,
and enhance efficiency in order to truly make SOEs competitive and independent market
players.
SOE: State In 2014 Ministry of Finance data showed that there were 15 million enter-
prises nationwide, and state-owned accounted for less than 1% of enterprises but assets
controlled by state-owned enterprises accounted for at least 30% of total assets. As a
crucial component to economic reform and development, the evolution of state-owned
[SOE Expert Interview] Now SOEs are responsible for some of the most important
tasks of the national economy. So they are indispensable to the entire national economy,
A-3

and for supporting the entire national economy. Because we are a country where public
ownership is central, it is feasible for the government to take the lead in reforming SOEs.
Whether it is from the perspective of its profit rate, or from the perspective that the gov-
ernment plays a irreplaceable role in the process of reforming SOEs, you can see more
aims of this SOEs reform. We should improve efficiency rather than simply privatizing it.
The fundamental attribute of SOEs is that they are inherently unified with the people.
The government controls and invests SOEs on behalf of the people. Adhering to the policy
of SOEs reform led by the state is consistent with the realization of the interests of the
whole people. The public welfare and historical mission of SOEs have determined that
the political core and political leading role of the government should be fully exerted in
SOE reform. Top-down design should be strengthened, so that SOEs responsible for the
national economy and people’s livelihoods can be strengthened with confidence.
SOE: Control In 2014 Ministry of Finance data showed that there were 15 million
enterprises nationwide, and state-owned accounted for less than 1% of enterprises but
assets controlled by state-owned enterprises accounted for at least 30% of total assets. As
a crucial component to economic reform and development, the evolution of state-owned
[SOE Expert Interview] At present, there are actually a lot of views on the reform of
SOEs. SOEs are responsible for some of the most important tasks of the national econ-
omy. So they are indispensable to the entire national economy. SOEs are enterprises
invested or controlled by the government, including not only the central government or
federal government, but also the local governments. As an organizational form of produc-
tion and management, SOEs have the characteristics of commercial and public welfare,
and their commercial performance is the pursuit of preserving and increasing the value of
state-owned assets. The establishment of SOEs is usually to achieve the goal of national
economic adjustment, and to play a role in reconciling all aspects of the national economy.
SOEs have a certain administrative nature. Due to historical reasons, the classification of
Chinese SOEs is quite complex. According to international practice, enterprises investing
in state-owned assets or holding more than 50% of their shares are SOEs. Therefore, en-
A-4

terprises with investments in state-owned assets within the scope of China are regulated
by law.
Since the reform and opening up, great achievements have been made in the develop-
ment of SOEs. At present, China has become a world economic power. To realize China’s
economic status as a powerful country in the world, we need a number of large and inter-
nationally competitive enterprises. However, the existing problems of SOEs, such as the
administrative character of the management system, and generalization of the functions,
do not give full play to their vitality. We are ready to fight a protracted and tough war, and
we are committed to achieving new results in the key links of reform.
China-Philippines: Dove The Republic of Philippines, referred to as the Philippines, is

located in the western Pacific Ocean. It is a multi-ethnic archipelago country in Southeast
Asia, covering an area of 299,700 square kilometers. Its population exceeded 100 million
in 2016. Inhabitants of the Philippines were migrants from the Asian continent, who
established the kingdom of Sulu, a maritime power around the fourteenth century. In
1946, the Philippines became independent. On June 9, 1975, China established diplomatic
relations with the Philippines. Disputes around the sovereignty of the South Island Reef
and the demarcation of the sea area are the main obstacle in the development of China-
Philippines relations.
[Foreign Policy Expert Interview] In the past short year, especially since Duterte came
to power, the cooperation between China and the Philippines has been fruitful in various
fields. China has become the Philippines’ largest trading partner, import market and ex-
porter. From January to July this year, the bilateral trade volume also reached 28 billion
US dollars, with year-on-year growth of 7.6%. At the same time, the Philippines is also
aware that China’s Belt and Road initiative will bring short-term and long-term strate-
gic opportunities to the Philippines and ASEAN countries in various ways. Actually, The
efforts made by both countries are consistent with the Philippines’ long-term goal of seek-
ing common prosperity and achieving common growth. As the world changes, we think
that China-Philippines relations have much room for development and cooperation.
On the relationship between China and the Philippines, China‘s position is that the
A-5

common interests of the two countries are far greater than the differences. Both sides
will adhere to friendly cooperation, common development, peace, and stability. To make
China-Philippines relations good and neighborly and cooperation stable and far-ranging,
the focus should be on long-term interests and regional stability, to deeply promote prag-
matic cooperation between China and the Philippines in different fields. Let more co-
operation come to the two countries and two peoples faster and more broadly, to push
China-Philippines relations to a new stage.
China-Philippines: Hawk The Republic of Philippines, referred to as the Philippines,

is located in the western Pacific Ocean. It is a multi-ethnic archipelago country in South-
east Asia, covering an area of 299,700 square kilometers. Its population exceeded 100
million in 2016. Inhabitants of the Philippines were migrants from the Asian continent,
who established the kingdom of Sulu, a maritime power around the fourteenth century.
In 1946, the Philippines became independent. On June 9, 1975, China established diplo-
matic relations with the Philippines. Disputes around the sovereignty of the South Island
Reef and the demarcation of the sea area are the main obstacle in the development of
China-Philippines relations.
[Foreign Policy Expert Interview] In the relationship between China and the Philip-
pines, the South China Sea issue is the key. On the stage of international relations, na-
tional interests are clearly the eternal theme. National sovereignty and territorial integrity
are essentially the bottom line of diplomacy. To further develop the China-Philippines
relations, both parties must ensure that the relationship is in the right direction. The
Philippine side must realize that on the issue of the South China Sea, China has historical
claim (to the islands). China has also been following international law and international
practice throughout. Any diplomatic action that does not respect this bottom line can only
complicate and worsen the relationship between the two countries. National sovereignty
and territorial integrity are sacred and inviolable. If the Philippine side cannot clarify its
position on the issue of the South China Sea in line with China’s position, it will not only
harm the friendship of the two peoples and the interests of both sides, but also pose a big
threat to lasting peace and prosperity in the Asia-Pacific region.
A-6

China’s position on the relationship between China and the Philippines is that the
two sides must ensure that bilateral relations are moving steadily in the right direction.
On the issue of the South China Sea, China has historical rights. China has also been
following international law and international practice. National sovereignty and territorial
integrity are sacred and inviolable. If the Philippines cannot fully understand this, it will
not only harm the national interests of both sides, but also run counter to the enduring
peace, prosperity and stability of the entire Asia-Pacific.
China-Philippines: Control The Republic of Philippines, referred to as the Philip-

pines, is located in the western Pacific Ocean. It is a multi-ethnic archipelago country in
Southeast Asia, covering an area of 299,700 square kilometers. Its population exceeded
100 million in 2016. Inhabitants of the Philippines were migrants from the Asian conti-
nent, who established the kingdom of Sulu, a maritime power around the fourteenth cen-
tury. In 1946, the Philippines became independent. On June 9, 1975, China established
diplomatic relations with the Philippines. Disputes around the sovereignty of the South
Island Reef and the demarcation of the sea area are the main obstacle in the development
of China-Philippines relations.
[Foreign Policy Expert Interview] In the relationship between China and the Philip-
pines, the South China Sea issue is the key. On the stage of international relations, na-
tional interests are clearly the eternal theme. The relationship between China and the
Philippines is impermanent like soldiers and water. The same is true of geopolitics. The
diplomatic relations between China and the Philippines have actually undergone many
historical transformations. To further develop the China-Philippines relations, both par-
ties must ensure that the relationship is headed in the right direction.
When dealing with the Philippines, China must consider the situation, not only main-
taining national sovereignty, but also adopting flexible strategies at the right time. China
should seek common points while reserving difference, make the best use of the circum-
stances, and push forward China-Philippines relations.
A-7

A.3. Predisposition Questions
Panel A of Table A1 shows questions used to measure predisposition on market economy,

along with their corresponding signs (1 or -1) used to construct the index (bigger means
more pro-market). Panel B of Table A1 shows questions used to measure predisposition
on national sovereignty, along with their corresponding signs used to construct the index
(bigger means more dovish/less nationalistic).
TABLE A1. Q UESTIONS M EASURING P REDISPOSITION

No. Question Sign
Panel A: On Market Economy
1 Privately-run schools should be restricted. -1
2 Private capital should be encouraged to set up private hospitals to provide 1
convenient and high-quality services to those willing to pay high prices.
3 Private ownership and sale of land should be allowed. 1
4 When the market price of grain is low, the government should purchase -1
grain from farmers at a price higher than the market price.
5 People should be allowed to freely exchange foreign currency. 1
6 Chinese enterprises’ overseas investment should not be solely for profit, -1
but should be guided by national strategies.
7 The defense industry should be open to private capital. -1
Panel B: On National Sovereignty
1 The government should prohibit artists who support Taiwan independence -1
and Hong Kong independence from performing in the Chinese mainland.
2 Foreign journalists who regularly publish negative reports about China 1
should be allowed to enter China.
3 China’s core diplomatic strategy should be to break through the contain- -1
ment of China by Western countries led by the United States.
4 When military conditions are sufficient, China can consider unifying Tai- -1
wan by force.
5 There should be a statutory holiday to commemorate Chairman Mao’s -1
birthday.
6 Those who publicly disrespect the national flag and national anthem -1
should be subjected to criminal punishment.
7 Chinese citizens should be allowed to hold foreign citizenship. 1
A-8

A.4. Related Policy Questions
Panel A of Table A2 shows example questions on topics related to SOE reform, along
with their corresponding signs (1 or -1) used to construct the index. Panel B of Table A2
shows example questions on topics related to foreign policy.
TABLE A2. E XAMPLES OF P OLICY Q UESTIONS

No. Question Sign
Panel A: On SOEs
1 Both SOE and private enterprises are pillars of the Chinese economy; 1
hence, there is no need to give special treatment to SOEs.
2 Sectors related to national security and important to the national economy -1
and people’s livelihoods must be controlled by SOEs.
3 The government should not use fiscal resources to give subsidies to SOEs 1
that are on the verge of bankruptcy but hire many employees.
4 For government procurement projects, SOEs should receive more prefer- -1
ential treatment than private enterprises.
Panel B: On Foreign Policy
1 The government should safeguard national unity and territorial integrity 1
through diplomatic and economic measures as much as possible to avoid
military conflicts.
2 If it is in the nation’s interest, China can unilaterally impose economic or -1
trade sanctions on other countries.
3 When faced with military conflicts, China should not use preemptive mil- 1
itary tactics.
4 China should forgo the diplomatic principle of “to hide one’s capacities -1
and bide one’s time” and defend its rights more resolutely in the interna-
tional arena.
A-9

A.5. Summary Statistics
TABLE A3. S UMMARY S TATISTICS

W1 Outcomes and Checks W2 Outcomes and Checks
Obs Mean Std. Dev. Obs Mean Std. Dev.
Agree to market reform (1, 2 or 3) 830 2.16 0.96 354 2.22 0.96
An index of market-oriented reform 823 0.00 1.00 351 0.12 1.03
Agree to more cooperation (1, 2 or 3) 830 1.57 0.74 354 1.53 0.73
An index of dovish foreign policy 825 0.00 1.00 353 0.00 0.98
Pass SOE manipulation check 830 0.83 0.37 354 0.94 0.23
Identify SOE media content as pro-market 830 0.44 0.50 354 0.46 0.50
Identify SOE media content as pro-state 830 0.28 0.45 354 0.30 0.46
Agree policy toward SOE reform is correct (1, 2 or 3) 794 2.30 0.51 350 2.26 0.51
Pass FP manipulation check 830 0.98 0.15 354 0.89 0.31
Identify FP media content as hawish 830 0.07 0.26 354 0.12 0.32
Identify FP media content as dovish 830 0.46 0.50 354 0.50 0.50
Agree policy toward South China Sea is correct (1, 2 or 3) 800 2.46 0.53 347 2.36 0.54
Covarites Covarites (W2 Sample)

Obs Mean Std. Dev. Obs Mean Std. Dev.
Female 830 0.41 0.49 354 0.41 0.49
Age 830 31.1 8.71 354 31.8 8.68
High school 830 0.24 0.43 354 0.19 0.39
Junior college 830 0.13 0.34 354 0.13 0.33
College or above 830 0.43 0.49 354 0.59 0.49
Minority 830 0.03 0.18 354 0.04 0.19
Religious 830 0.17 0.37 354 0.18 0.39
Single 830 0.45 0.50 354 0.40 0.49
Having worked 830 0.77 0.42 354 0.82 0.38
Public sector worker 830 0.16 0.37 354 0.17 0.38
Being reported to 830 0.38 0.49 354 0.47 0.50
CCP member 830 0.13 0.34 354 0.15 0.36
Having used VPN 830 0.37 0.48 354 0.40 0.49
English proficiency [1, 5] 830 2.91 1.13 354 3.13 1.09
Income category [1, 10] 830 4.60 2.03 354 4.95 1.92
Self-reported social class [1, 10] 830 5.21 2.06 354 5.64 1.94
Note: W1 and W2 refer to Wave 1 and Wave 2, respectively.
A-10

A.6. Balance Tables
TABLE A4. BALANCE TABLE : WAVE 1

Self-
Public Being reported

High Junior College Having sector reported CCP Having English Income social
Female Age school college or above Minority Religious Single worked worker to member used VPN proficiency category class
Outcome Variables (0 or 1) [18, 50] (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1) [1, 5] [1, 10[ [1, 10]
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)
Treatment1: Market 0.018 0.491 0.011 -0.020 0.013 0.014 0.010 -0.039 -0.016 -0.009 0.028 0.020 0.058 -0.077 -0.056 -0.045
(0.042) (0.733) (0.037) (0.028) (0.042) (0.016) (0.031) (0.042) (0.035) (0.031) (0.042) (0.029) (0.041) (0.097) (0.172) (0.175)
Treatment1: State -0.048 -0.201 -0.024 0.028 0.008 0.004 0.022 -0.001 -0.030 0.014 -0.035 -0.000 0.018 0.090 -0.060 0.238
(0.042) (0.745) (0.036) (0.030) (0.042) (0.014) (0.032) (0.043) (0.036) (0.032) (0.041) (0.028) (0.041) (0.097) (0.172) (0.174)
0.079* 0.462 0.017 0.008 0.052 0.004 -0.030 -0.066 -0.015 -0.024 0.021 0.014 0.018 0.137 -0.122 0.151
Treatment2: Dove (0.042) (0.758) (0.036) (0.028) (0.042) (0.015) (0.033) (0.042) (0.036) (0.031) (0.041) (0.028) (0.041) (0.097) (0.176) (0.175)
0.012 -1.244* 0.027 0.024 0.033 0.015 -0.060* 0.029 0.023 -0.020 0.016 0.034 -0.002 0.169* -0.322* 0.258
A-11
Treatment2: Hawk (0.041) (0.730) (0.036) (0.029) (0.042) (0.016) (0.032) (0.043) (0.035) (0.032) (0.041) (0.029) (0.041) (0.097) (0.170) (0.173)
Constant 0.386*** 31.311*** 0.230*** 0.120*** 0.391*** 0.023** 0.188*** 0.479*** 0.785*** 0.172*** 0.371*** 0.111*** 0.345*** 2.802*** 4.780*** 5.021***
(0.038) (0.693) (0.033) (0.026) (0.038) (0.010) (0.030) (0.039) (0.032) (0.029) (0.038) (0.026) (0.037) (0.090) (0.154) (0.164)
Observations 830 830 830 830 830 830 830 830 830 830 830 830 830 830 830 830
R-squared 0.008 0.008 0.002 0.004 0.002 0.002 0.005 0.007 0.002 0.002 0.003 0.003 0.003 0.007 0.004 0.006
F test p -value 0.147 0.142 0.790 0.503 0.796 0.666 0.377 0.173 0.735 0.845 0.612 0.711 0.681 0.207 0.433 0.298

TABLE A5. BALANCE TABLE : WAVE 2
Public Being
High Junior College Having sector reported CCP Having English Income
Female Age school college or above Minority Religious Single worked worker to member used VPN proficiency category
Outcome Variables (0 or 1) [18, 50] (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1) (0 or 1) [1, 5] [1, 10[
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15)
Treatment1: Market 0.050 -0.681 0.070 -0.014 -0.052 -0.001 0.058 -0.009 -0.019 -0.059 -0.024 -0.005 0.021 -0.176 -0.196
(0.064) (1.135) (0.053) (0.041) (0.064) (0.026) (0.049) (0.063) (0.048) (0.051) (0.065) (0.049) (0.063) (0.143) (0.244)
Treatment1: State 0.006 -1.538 -0.030 0.035 0.015 -0.014 0.047 0.040 -0.048 -0.097* -0.092 -0.055 0.041 0.073 -0.019
(0.064) (1.115) (0.048) (0.045) (0.065) (0.022) (0.051) (0.065) (0.051) (0.051) (0.066) (0.047) (0.065) (0.145) (0.253)
Treatment2: Dovish 0.110* 0.826 -0.024 0.013 0.048 -0.010 -0.032 -0.178*** 0.030 -0.008 0.067 -0.031 -0.054 0.075 -0.148
(0.064) (1.141) (0.048) (0.045) (0.064) (0.022) (0.053) (0.062) (0.050) (0.051) (0.065) (0.046) (0.063) (0.143) (0.257)
Treatment2: Hawkish -0.022 -2.048* 0.054 -0.035 0.040 0.017 -0.103** 0.003 0.024 -0.072 0.058 -0.009 -0.008 0.078 -0.291
(0.064) (1.139) (0.054) (0.042) (0.065) (0.025) (0.050) (0.066) (0.051) (0.049) (0.066) (0.049) (0.065) (0.145) (0.251)
A-12
Constant 0.363*** 32.921*** 0.166*** 0.128*** 0.571*** 0.039** 0.193*** 0.446*** 0.829*** 0.250*** 0.465*** 0.183*** 0.402*** 3.116*** 5.165***
(0.058) (1.085) (0.047) (0.040) (0.060) (0.018) (0.047) (0.060) (0.045) (0.051) (0.061) (0.048) (0.059) (0.141) (0.226)
Observations 354 354 354 354 354 354 354 354 354 354 354 354 354 354 354
R-squared 0.016 0.023 0.019 0.008 0.005 0.005 0.017 0.032 0.004 0.016 0.010 0.006 0.004 0.010 0.006
F test p -value 0.232 0.0697 0.185 0.594 0.772 0.721 0.150 0.0164 0.853 0.266 0.468 0.733 0.874 0.442 0.709

A.7. Main Results
TABLE A6. T REATMENT E FFECTS ON P OLICY P REFERENCES : SOE R EFORM

Agree to more market competition as An index of market-oriented reform
Outcome variables opposed to state control (1, 2 or 3) (mean = 0; sd = 1)
(1) (2) (3) (4) (5) (6)
Treatment1: Market 0.332*** 0.336*** 0.307*** 0.290*** 0.292*** 0.316***

(0.076) (0.076) (0.078) (0.086) (0.086) (0.088)
Treatment1: State -0.344*** -0.354*** -0.363*** -0.194** -0.196** -0.199**
(0.082) (0.081) (0.082) (0.083) (0.083) (0.084)
Treatment2: Dove -0.180** -0.190** -0.175** -0.122 -0.131 -0.131

(0.079) (0.079) (0.080) (0.087) (0.087) (0.089)
Treatment2: Hawk -0.159** -0.149* -0.157** -0.194** -0.182** -0.188**
(0.077) (0.077) (0.078) (0.085) (0.085) (0.086)
Female -0.110 -0.100 -0.139** -0.149**

(0.068) (0.069) (0.069) (0.071)
Age/10 0.081 0.022 0.789** 0.747**
(0.350) (0.352) (0.372) (0.371)
Age squared/100 0.005 0.012 -0.083 -0.077
(0.049) (0.049) (0.053) (0.053)
High school 0.047 0.075 -0.126 -0.111
(0.104) (0.106) (0.113) (0.115)
Junior college 0.207* 0.222* 0.088 0.074
(0.114) (0.116) (0.127) (0.129)
College or above 0.241** 0.246** 0.147 0.147
(0.108) (0.111) (0.118) (0.120)
Minority 0.158 0.131 -0.291* -0.360**
(0.174) (0.189) (0.166) (0.167)
Religious -0.073 -0.043 -0.161* -0.161*
(0.083) (0.086) (0.088) (0.092)
Single -0.088 -0.107 0.141 0.175*
(0.095) (0.098) (0.099) (0.101)
Having worked 0.067 0.054 -0.060 -0.051
(0.101) (0.101) (0.104) (0.106)
Public sector worker -0.023 -0.032 -0.177* -0.193*
(0.092) (0.092) (0.101) (0.104)
Being reported to -0.228*** -0.238*** -0.108 -0.106
(0.081) (0.084) (0.088) (0.089)
CCP member -0.167* -0.140 0.049 0.028
(0.101) (0.102) (0.107) (0.109)
Having used VPN -0.029 -0.032 0.103 0.088
(0.071) (0.072) (0.075) (0.076)
English proficiency 0.005 -0.000 0.074* 0.067
(0.037) (0.038) (0.042) (0.041)
Income category 0.006 0.019 0.006 0.013
(0.021) (0.022) (0.023) (0.024)
Self-reported social class -0.017 -0.017 -0.039** -0.034*
(0.018) (0.019) (0.020) (0.020)
Outcome Variable Mean: ∼Full sample 2.165 2.165 2.165 0.000 0.000 0.000
∼Treatment1 = Neutral 2.167 2.167 2.167 -0.035 -0.035 -0.035
Control variables No Yes Yes No Yes Yes
Provincial fixed effects No No Yes No No Yes
Observations 830 830 830 823 823 823
R-squared 0.088 0.126 0.160 0.044 0.100 0.125
A-13

TABLE A7. T REATMENT E FFECTS ON P OLICY P REFERENCES : S OUTH C HINA S EA
Agree to policies that emphasize more
on cooperation than sovereignty (1, 2 An index of dovish foreign policy
Outcome variables or 3) (mean = 0; sd = 1)
(1) (2) (3) (4) (5) (6)
Treatment1: Market -0.003 -0.013 -0.008 0.054 0.055 0.047

(0.061) (0.059) (0.059) (0.083) (0.084) (0.086)
Treatment1: State -0.000 -0.011 -0.024 -0.032 -0.037 -0.059
(0.061) (0.060) (0.060) (0.083) (0.083) (0.084)
Treatment2: Dove 0.165*** 0.149** 0.155** 0.189** 0.166** 0.178**

(0.063) (0.063) (0.064) (0.083) (0.083) (0.084)
Treatment2: Hawk -0.318*** -0.307*** -0.307*** -0.266*** -0.288*** -0.264***
(0.058) (0.059) (0.059) (0.087) (0.088) (0.090)
Female -0.043 -0.045 0.175** 0.164**

(0.050) (0.051) (0.071) (0.071)
Age/10 0.304 0.379 0.179 0.088
(0.288) (0.293) (0.387) (0.396)
Age squared/100 -0.032 -0.042 -0.034 -0.021
(0.041) (0.042) (0.055) (0.056)
High school -0.057 -0.073 -0.193* -0.193*
(0.082) (0.082) (0.114) (0.113)
Junior college 0.053 0.020 -0.085 -0.069
(0.086) (0.088) (0.121) (0.123)
College or above 0.128 0.108 0.048 0.069
(0.085) (0.088) (0.117) (0.119)
Minority -0.224** -0.277** -0.129 -0.191
(0.091) (0.115) (0.150) (0.154)
Religious 0.057 0.038 -0.073 -0.125
(0.066) (0.067) (0.093) (0.093)
Single -0.086 -0.083 -0.006 -0.033
(0.076) (0.076) (0.106) (0.109)
Having worked 0.005 0.007 0.037 0.025
(0.079) (0.080) (0.110) (0.114)
Public sector worker -0.109 -0.107 -0.139 -0.108
(0.073) (0.074) (0.103) (0.103)
Being reported to -0.033 -0.025 -0.140* -0.126
(0.063) (0.065) (0.084) (0.087)
CCP member -0.147** -0.143* -0.026 -0.007
(0.073) (0.078) (0.117) (0.123)
Having used VPN 0.131** 0.112** 0.094 0.068
(0.053) (0.056) (0.075) (0.078)
English proficiency -0.011 -0.011 0.009 0.009
(0.029) (0.030) (0.043) (0.044)
Income category -0.011 -0.013 -0.018 -0.022
(0.015) (0.015) (0.021) (0.022)
Self-reported social class 0.025* 0.028* 0.032 0.027
(0.015) (0.016) (0.020) (0.021)
Outcome Variable Mean: ∼Full sample 1.578 1.575 1.575 0.000 0.000 0.000
∼Treatment1 = Neutral 1.573 1.573 1.573 -0.013 0.024 0.024
Control variables No Yes Yes No Yes Yes
Provincial fixed effects No No Yes No No Yes
Observations 830 830 830 825 825 825
R-squared 0.074 0.130 0.156 0.035 0.069 0.094
A-14

A.8. Robustness Checks of Main Results
TABLE A8. T REATMENT E FFECTS ON P OLICY P REFERENCES : ROBUSTNESS C HECK
Identify Pass age Agree policy

media consistency Agree policy toward the
source as filter and Pass both toward SOE South China
National not a manipulatio reform is Sea is
Criterion Full Sample TV Station speeder n checks correct correct
Panel A An index of market-oriented reform (mean = 0; sd = 1)

(1) (2) (3) (4) (5) (6)
Treatment1: Market 0.313*** 0.363*** 0.320*** 0.313*** 0.312*** 0.294***

(0.088) (0.097) (0.096) (0.100) (0.092) (0.088)
Treatment1: State -0.184** -0.201** -0.183** -0.204** -0.197** -0.168*
(0.084) (0.093) (0.089) (0.096) (0.087) (0.086)
Treatment2: Dovish -0.136 -0.152 -0.173* -0.105 -0.090 -0.099

(0.089) (0.099) (0.096) (0.101) (0.093) (0.089)
Treatment2: Hawkish -0.182** -0.266*** -0.205** -0.210** -0.173** -0.173**
(0.085) (0.094) (0.091) (0.096) (0.088) (0.084)
Outcome variable mean 0.000 0.012 0.014 0.030 -0.013 -0.019

Control variables Yes Yes Yes Yes Yes Yes
Provincial fixed effects Yes Yes Yes Yes Yes Yes
Observations 823 708 757 679 770 786
R-squared 0.122 0.137 0.125 0.120 0.125 0.124
Panel B An index of dovish foreign policy (mean = 0; sd = 1)

(1) (2) (3) (4) (5) (6)
Treatment1: Market 0.021 0.043 0.019 0.025 0.038 0.020

(0.086) (0.092) (0.092) (0.094) (0.090) (0.085)
Treatment1: State -0.072 -0.058 -0.069 -0.045 -0.074 -0.059
(0.084) (0.091) (0.091) (0.091) (0.087) (0.087)
Treatment2: Dovish 0.187** 0.222** 0.196** 0.154 0.204** 0.214**

(0.084) (0.091) (0.091) (0.094) (0.088) (0.085)
Treatment2: Hawkish -0.261*** -0.278*** -0.282*** -0.345*** -0.258*** -0.242***
(0.090) (0.097) (0.096) (0.097) (0.093) (0.088)
Outcome variable mean 0.000 0.001 -0.018 -0.015 -0.008 -0.016

Observations 825 710 758 680 772 787
R-squared 0.093 0.118 0.101 0.121 0.097 0.099
Note: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1
A-15

A.9. Treatment Effects on Other Outcomes
TABLE A9. T REATMENT E FFECTS ON OTHER O UTCOMES

Self-reported Self-reported
Trust in the Trust in the Self-reported voting in discussion on
central local complaint- NPC government
Patriotism Nationalism Efficacy government government making elections policies
index index index [1, 10] [1, 10] (0 or 1) (0 or 1) (1, 2, 3 or 4)
(1) (2) (3) (4) (5) (6) (7) (8)
Treatment1: Market -0.156* 0.016 0.033 -0.169 -0.007 -0.006 0.013 0.047
(0.087) (0.085) (0.071) (0.166) (0.178) (0.034) (0.031) (0.050)
Treatment1: State 0.044 0.101 -0.050 0.084 0.069 -0.015 0.001 0.031
(0.084) (0.082) (0.070) (0.159) (0.188) (0.034) (0.030) (0.051)
Treatment2: Dove 0.018 0.042 0.129* 0.033 -0.036 -0.025 -0.034 0.076
(0.085) (0.080) (0.070) (0.159) (0.187) (0.034) (0.031) (0.053)
Treatment2: Hawk 0.056 0.134 0.062 0.080 0.119 -0.005 -0.040 0.087*
(0.083) (0.083) (0.070) (0.174) (0.186) (0.033) (0.031) (0.049)
Outcome Variable Mean 0.000 0.000 0.000 8.145 6.731 0.207 0.169 2.218
Control variables Yes Yes Yes Yes Yes Yes Yes Yes
Provincial fixed effects Yes Yes Yes Yes Yes Yes Yes Yes
Observations 828 823 822 830 830 830 830 830
R-squared 0.118 0.188 0.404 0.077 0.113 0.147 0.159 0.261
Note: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. A full set of individual characteristics and provincial fixed
effects are being controlled for in all regressions.
A-16

A.10. Marginal Effects of the Treatments
F IGURE A2. M ARGINAL E FFECTS OF THE T REATMENTS G IVEN P REDISPOSITION
0.5
Marginal Effects of Treatments

0.0 Treatment
T1: Control
T1: Market
T1: State
−0.5
−1.0
0 25 50 75 100
State Control <− Predisposition on Economy −> Free Market
(a) Treatment 1 on Agreeing to Market-oriented Reform
0.5
Marginal Effects of Treatments
0.0
Treatment
T2: Control
T2: Dove
T2: Hawk
−0.5
0 25 50 75 100
Hawkish <− Predisposition on Sovereignty −> Dovish
(b) Treatment 2 on Agreeing to More Cooperation
Note: The (stacked) histograms at the bottom of each plot show the number of respondents under each
treatment condition given different values of the moderator (in percentile).
A-17

A.11. 48-Hour Follow-up Results
Table A10 presents the estimated treatment effects on the full and recontact samples.
TABLE A10. T REATMENT E FFECTS ON P OLICY P REFERENCES IN 48 H OURS

Panel A SOE Reform
Agree to more market competition as An index of market-oriented reform
opposed to state control (1, 2 or 3) (W1: mean = 0; sd = 1)
Two-wave Two-wave Two-wave Two-wave

Full Sample: Subsample: Subsample: Full Sample: Subsample: Subsample:
w1 outcome w1 outcome w2 outcome w1 outcome w1 outcome w2 outcome
(1) (2) (3) (4) (5) (6)
Treatment1: Market 0.307*** 0.202* 0.139 0.316*** 0.317** 0.022

(0.078) (0.118) (0.117) (0.088) (0.135) (0.142)
Treatment1: State -0.363*** -0.378*** -0.525*** -0.199** -0.287** -0.458***
(0.082) (0.130) (0.131) (0.084) (0.125) (0.141)
Treatment2: Dovish -0.175** -0.254** -0.238* -0.131 -0.116 -0.172

(0.080) (0.125) (0.131) (0.089) (0.143) (0.139)
Treatment2: Hawkish -0.157** -0.178 -0.192 -0.188** -0.037 -0.123
(0.078) (0.126) (0.128) (0.086) (0.132) (0.139)
Outcome Variable Mean: ∼Full Sample 2.165 2.274 2.218 0.000 0.040 0.128
Observations 830 353 353 823 351 350
R-squared 0.160 0.207 0.201 0.125 0.245 0.194
Panel B Foreign Poicy
Agree to policies that emphasize more on An index of dovish foreign policy
cooperation than sovereignty (1, 2 or 3) (W1: mean = 0; sd = 1)
Two-wave Two-wave Two-wave Two-wave
Full Sample: Subsample: Subsample: Full Sample: Subsample: Subsample:
w1 outcome w1 outcome w2 outcome w1 outcome w1 outcome w2 outcome
(1) (2) (3) (4) (5) (6)
Treatment1: Market -0.008 -0.071 0.049 0.047 0.076 -0.068

(0.059) (0.087) (0.095) (0.086) (0.133) (0.141)
Treatment1: State -0.024 -0.043 0.136 -0.059 -0.125 -0.065
(0.060) (0.095) (0.101) (0.084) (0.138) (0.137)
Treatment2: Dovish 0.155** 0.117 0.063 0.178** 0.198 0.050

(0.064) (0.093) (0.101) (0.084) (0.138) (0.138)
Treatment2: Hawkish -0.307*** -0.162* -0.055 -0.264*** -0.053 -0.200
(0.059) (0.091) (0.102) (0.090) (0.147) (0.149)
Outcome Variable Mean: ∼Full Sample 1.578 1.541 1.533 0.000 -0.047 0.000
∼Treatment1 = Neutral 1.573 1.568 1.474 -0.013 -0.066 0.001
∼Treatment2 = Neutral 1.630 1.521 1.512 0.024 -0.136 -0.004
Observations 830 353 353 825 352 352
R-squared 0.156 0.236 0.151 0.094 0.137 0.093
Note: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. A full set of individual characteristics and
provincial fixed effects are being controlled for in all regressions.
A-18

A.12. Variable Definitions
A-19

TABLE A11. VARIABLE D EFINITIONS
Variable Question Options
Individual Characteristics
Female What is your gender? 1 = female; 0 = male.
Age Which year were you born?
Province Which province is your permanent residency?
1 = primary school; 2 = secondary school; 3 = vocational middle school; 4 =

Education What is your highest level of education attainment?
high school; 5 = junior college; 6 = college; 7 = master's; 8 = Ph.D.
Minority What is your ethnicity? 0 = Han; 1 = otherwise.
Single What is your marital status? 0 = married; 1 = otherwise.
1 = do not speak English at all; 2 = can say a few sentences; 3 = can speak
English proficiency [1, 5] What is your level of English proficiency? and read a little; 4 = can manage conversations, but not fluently; 5 = can
speak fluently.
Income category [1, 10] What is your total disposable income, including salaries, stipends, and allowances? 10 categories, from low to high.
Religious Do you have a religion? 1 = yes; 0 = no.
CCP member What is your political affiliation? 1 = Chinese Communist Party; 0 = otherwise.
Self-reported social class [0, 10] What do you think is your social class on a 0-10 scale? 0 = the bottom of the society; 1 = the top of the society.
Having worked Have you ever had a full-time job? 1 = yes; 0 = no.
Public sector worker What is the nature of your employer? 1 = SOEs or the government; 0 = otherwise.
A-20
Being reported to In your job, are there any people who report to you? 1 = yes; 0 = no.
Questions Measuring Predisposition

Pre-disposition on market economy:
Privately-run schools should be restricted. 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
component 1
Pre-disposition on market economy: Private capital should be encouraged to set up private hospitals to provide convenient
5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
component 2 and high-quality services to those willing to pay high prices.

Private ownership and sale of land should be allowed. 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
component 3
Pre-disposition on market economy: When the market price of grain is low, the government should purchase grain from
component 4 farmers at a price higher than the market price.

People should be allowed to freely exchange foreign currency. 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
component 5
Pre-disposition on market economy: Chinese enterprises' overseas investment should not be solely for profit, but should be
component 6 guided by national strategies.
The defense industry should be open to private capital. 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
component 7
Pre-disposition on national The government should prohibit artists who support Taiwan independence and Hong
sovereignty: component 1 Kong independence from performing in the Chinese mainland.
Pre-disposition on national Foreign journalists who regularly publish negative reports about China should be
sovereignty: component 2 allowed to enter China.
Pre-disposition on national China's core diplomatic strategy should be to break through the containment of China
sovereignty: component 3 by Western countries led by the United States.
Pre-disposition on national
When military conditions are sufficient, China can consider unifying Taiwan by force. 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
sovereignty: component 4
There should be a statutory holiday to commemorate Chairman Mao's birthday. 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
Pre-disposition on national Those who publicly disrespect the national flag and national anthem should be
sovereignty: component 6 subjected to criminal punishment.

Chinese citizens should be allowed to hold foreign citizenship. 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
Questions regarding the SOE reform

According to the video, what is the percentage of assets controlled by the SOEs in
Pass SOE manipulation check (W1) 1 = 30%; 0 = 20%; 40%; 50%; do not know.
China?
1 = SOE reform; 0 = reform of private enterprises; reform of joint ventures;
Pass SOE manipulation check (W2) What is the video that you saw recently mainly concerned about?
do not know.
1 = The key to the SOE reform is top-level design; state-led SOE reform
can be successful. 2 = The key to the SOE reform is marketization and
SOE media content What is the main content of the video? adhering to the principle of making SOEs competitive and independent
market players. 3 = Many problems exist in the SOE reform; we should be
A-21
ready for a long and difficult battle.
Agree policy toward SOE reform is 0 = Completely incorrect; 1 = a little bit incorrect; 2 = in general correct;
Do you think the policy toward the SOE reform described in the video is correct?
correct (0, 1, 2 or 3) 3 = completely correct.
1 = SOEs should be bigger and stronger; the SOE reform should be primarily
led by the government; the key to its success is top-level design. 2 =
Marketization is the key to the SOE reform; we should allow inefficient SOEs
Agree to market reform (1, 2 or 3) Which of the following is closest to your opinion toward the SOE reform?
to be driven out by the market. 3 = SOEs are state-owned assets; as long as
the value of these assets is preserved and increases, any measure is
justifiable.
Index of market-oriented reform: Both SOE and private enterprises are pillars of the Chinese economy; therefore, there
component 1 is no need to give special treatment to SOEs.
Index of market-oriented reform: SOEs bears a lot of policy burden; therefore, it is totally reasonable for the state to
component 2 support them with (special) policies.
Index of market-oriented reform: The goal is to have the market to play a decisive role in resource allocation; hence,
component 3 to maintain a high or low proportion of SOEs in the economy is not the goal.
Index of market-oriented reform: Private enterprises contribute 80% of the employment in China; therefore, they need
component 4 more support than the SOEs.
Index of market-oriented reform: Sectors related to national security and important to the national economy and
component 5 people's livelihoods must be controlled by state-owned enterprises.
Index of market-oriented reform: The government should not use fiscal resources to give subsidies to companies that
component 6 are on the verge of bankruptcy but hire many employees.
Index of market-oriented reform: The government should use various measures to help SOEs go out and have them
component 7 hold advantageous positions in international competition.
Index of market-oriented reform: For government procurement projects, SOEs should receive more preferable
component 8 treatment than private enterprises.
Index of market-oriented reform: The government should introduce market competition to sectors in which SOEs are
component 9 monopolies in order to strengthen the dynamism of SOEs.
Questions regarding the South China Sea

Pass FP manipulation check The previous video concerns the relationship between China and what country? 1 = the Philippines; 0 = Vietnam, the United State; Indonesia; do not know.
1 = The irresponsible attitudes of the Philippines on the issue of the South

China Sea makes the Sino-Philippines relations increasingly tense. 2 = The
common interests between China and the Philippines are bigger than the
FP media content What is the main content of the video?
disputes; the main theme of the relations is cooperation. 3 = The China-
Philippines relations fluctuate and the key factor is the issue of the South
China Sea.
Agree policy toward South China Do you think the policy toward the South China Sea described in the video is 0 = Completely incorrect; 1 = a little bit incorrect; 2 = in general correct;
Sea is correct (0, 1, 2 or 3) correct? 3 = completely correct.
1 = China's sovereignty and territorial integrity is sacred and inviolable;
China cannot concede an inch on the issue of sovereignty. 2 = The
international situations are changeable; the way China handles the China-
Agree to more cooperation (1, 2 or
Which of the following is closest to your opinion toward the South China Sea? Philippines relations should change with environments and situations; 3 = In
3)
A-22
order to maintain peace and stability of the region, China should table
sovereign disputes with the Philippines and the two countries should jointly
extract natural resources.
Index of dovish foreign policy: On the issue of South China Sea, sovereignty is the premise; only under this premise,
component 1 cooperation on economic and trade issues can be discussed.
Index of dovish foreign policy: Economic interest is an important component of the national interest. (We) should
component 2 table the disputes on sovereignty and maximize economic interests.
Index of dovish foreign policy: To maintain a long-term stable cooperative relationship with the ASEAN countries
component 3 including the Philippines is more important than de facto control of a few islands.
Index of dovish foreign policy: Economy is a short-term interest and territory is a long-term interest. (We) should
component 4 seize any opportunity to strengthen the control of islands in the South China Sea.
Index of dovish foreign policy: The government should safeguard national unity and territorial integrity through
component 5 diplomatic and economic measures as much as possible to avoid military conflicts.
Index of dovish foreign policy: If it is in the nation's interest, China can unilaterally impose economic or trade
component 6 sanctions on other countries.
Index of dovish foreign policy: The government should attach as much importance to the development military
component 7 power as it does to the development of the economy.
Index of dovish foreign policy:
When faced with military conflicts, China should not use preemptive military tactics. 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
component 8
Index of dovish foreign policy: China should forgo the diplomatic principle of "to hide one's capacities and bide one's
component 9 time" and defend its rights more resolutely on international platforms.
Media source What do you think is the source of the two videos? 1 = CCTV; 0 = local TV news; Internet news; others.
Other indicators
Patriotism: component 1 I am happy that I am a Chinese. 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
Patriotism: component 2 In general, China is better than most other countries. 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
Patriotism: component 3 When other people criticize China, I feel that they are criticizing me. 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
Even if I can choose to become a citizen of another country, I prefer to be a Chinese
Patriotism: component 4 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
citizen.
Patriotism: component 5 I feel proud and lucky to be Chinese. 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
Nationalism: component 1 Unless China achieves modernization, foreign countries will manage to exploit China. 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
The modern diplomatic history of China is a history of China being victimized and
Nationalism: component 2 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
beaten by foreign powers.
"The century of humuiliation" describes not only China's past, but also how foreigners
treat China today.
Foreign criticisms of China always have ulterior motives and we as Chinese should
refute them resolutely.
China should be the leader of East Asia because of its long history and splendid
civilization.
Trust in the central government Do you trust the central government? 10-point Likert scale. 1 = not trust at all; 10 = trust completely.
Trust in the local government Do you trust the local government? 10-point Likert scale. 1 = not trust at all; 10 = trust completely.
Efficacy: component 1 People like me do not have the ability to evaluate government performance. 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
Efficacy: component 2 Government officials do not care about opinions of people like me. 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
A-23
Efficacy: component 3 The government can timely respond to opinions and suggestions of ordinary citizens. 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
Efficacy: component 4 Politics is too complicated for people like me to understand. 5-point Likert scale. 1 = strongly disagree; 5 = strongly agree.
Having used VPN Have you used a VPN during the past year? 1 = yes; 0 = otherwise.
Have you provided suggestions to the government using various channels during the
Complaint-making 1 = yes; 0 = no.
past year?
Voting in local NPC elections Have you voted in a local NPC election during the past year? 1 = yes; 0 = no.
Discuss government policies Do you discuss topics concerning government policies? 1 = almost never; 2 = rarely; 3 = often; 4 = almost every day.

The Effects of Television News Propaganda: Experimental Evidence From China

Uploaded by

Copyright:

Available Formats

You might also like

The Effects of Television News Propaganda: Experimental Evidence From China

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Effects of Television News Propaganda: Experimental Evidence From China

Uploaded by

Copyright:

Available Formats

The Effects of Television News Propaganda:

Experimental Evidence from China∗

Jennifer Pan† Zijie Shao‡ Yiqing Xu§

Keywords: propaganda, media, policy, China, experiment

Word Count: 8,544

Electronic copy available at: https://ssrn.com/abstract=3579148

Electronic copy available at: https://ssrn.com/abstract=3579148

Electronic copy available at: https://ssrn.com/abstract=3579148

Electronic copy available at: https://ssrn.com/abstract=3579148

Electronic copy available at: https://ssrn.com/abstract=3579148

Electronic copy available at: https://ssrn.com/abstract=3579148

Television Propaganda in China

Electronic copy available at: https://ssrn.com/abstract=3579148

Electronic copy available at: https://ssrn.com/abstract=3579148

Electronic copy available at: https://ssrn.com/abstract=3579148

Electronic copy available at: https://ssrn.com/abstract=3579148

Survey sample: We recruited an online sample of 830 respondent from 26 provinces in

Electronic copy available at: https://ssrn.com/abstract=3579148

Electronic copy available at: https://ssrn.com/abstract=3579148

Electronic copy available at: https://ssrn.com/abstract=3579148

Electronic copy available at: https://ssrn.com/abstract=3579148

Electronic copy available at: https://ssrn.com/abstract=3579148

TABLE 1. M ANIPULATION C HECK

Identify Identify Identify Identify

Treatment1: Market 0.060* 0.510*** -0.049* 0.011 -0.038 0.029

Treatment2: Dove -0.077** -0.024 0.023 -0.001 0.501*** -0.021

Outcome Variable Mean:

Electronic copy available at: https://ssrn.com/abstract=3579148

F IGURE 2. T REATMENT E FFECTS ON P OLICY P REFERENCES : SOE R EFORM

(a) Agree to Market Reform (b) An Index of Market-oriented Reform

Electronic copy available at: https://ssrn.com/abstract=3579148

F IGURE 3. T REATMENT E FFECTS ON P OLICY P REFERENCES : S OUTH C HINA S EA

(a) Agree to More Cooperation (b) An Index of Dovish Foreign Policy

Electronic copy available at: https://ssrn.com/abstract=3579148

Electronic copy available at: https://ssrn.com/abstract=3579148

2.5 2.5 2.5

2.0 2.0 2.0

1.5 1.5 1.5

0 25 50 75 100 0 25 50 75 100 0 25 50 75 100

State Control <− Predisposition on Economy −> Free Market

Electronic copy available at: https://ssrn.com/abstract=3579148

2.0 2.0 2.0

1.5 1.5 1.5

1.0 1.0 1.0

0 25 50 75 100 0 25 50 75 100 0 25 50 75 100

Hawkish <− Predisposition on Sovereignty −> Dovish

heterogeneous treatment effects.

Electronic copy available at: https://ssrn.com/abstract=3579148

Full sample ● Full sample ●

Age 30 or below ● Age 30 or below ●

Age above 30 ● Age above 30 ●

Below college ● Below college ●

College or above ● College or above ●

Low income ● Low income ●

High income ● High income ●

Private sector ● Private sector ●

Public sector ● Public sector ●

We do not find differences in the amount of attention respondents of various subgroups

Treatment2: Dove -0.077 -0.024 0.023 -0.001 0.501* -0.021

Note: Robust standard errors in parentheses. * p<0.01, p<0.05, * p<0.1.

Note: Robust standard errors in parentheses. * p<0.01, p<0.05, * p<0.1.

Treatment1: Market 0.332* 0.336* 0.307* 0.290* 0.292* 0.316*

Treatment2: Dove -0.180 -0.190 -0.175** -0.122 -0.131 -0.131

Female -0.110 -0.100 -0.139 -0.149

Treatment2: Dove 0.165* 0.149 0.155 0.189 0.166 0.178

Female -0.043 -0.045 0.175 0.164

Treatment1: Market 0.313* 0.363* 0.320* 0.313* 0.312* 0.294*

Treatment2: Dovish 0.187 0.222 0.196 0.154 0.204 0.214**