1

Advertising Chinese Politics: The Effects of Public Service Announcements in Urban China

by

Daniela Stockmann Assistant Professor Leiden University

Ashley Esarey Visiting Assistant Professor

Whitman College

Zhang Jie Assistant Professor

Communication University of China

Comments are welcome*

Email correspondence: dstockmann@fsw.leidenuniv.nl

* For fruitful research collaboration we would like to thank Shen Mingming, Yan Jie, and Chai Jingjing from the Research Center of Contemporary China and Song Houze from at Peking University. For insights, suggestions, comments, and criticisms we are grateful to Liu Linqing, Rosario Aguilar, Liz Suhay, Ann Crigler, Michael Delli Carpini, Devra Coren-Moehler, Tang Wenfang, Ted Brader, Peter van Aelst, Tereza Capelos, Nick Valentino, Eric Groenendyk, Pierre Landry, Stefan Landsberger, Michael Meffert, Philip Everts, Kees Brants, Wang Yuhua, Pan Jian, and Carsten Hoppe. This paper is part of a joint research project between Leiden University and Communication University of China, funded by the Royal Netherlands Academy of Arts and Sciences (KNAW). We are also grateful for financial support for conducting focus groups from Whitman College. For superb research assistance we would like to thank Li Zheng, Jin Xi, Xu Yuemiao, Ma Beibei, Lu Hanji, and Zhang Chenlong. For computer programming we are dedicated to Harold Fellermann and Daniel Wunsch.

2

Abstract

During the reform era the Chinese state has modernized the way in which campaigns are conducted. Today, the government focuses on public service advertising on television to actively shape peoples attitudes and behavior on such issues as environmental protection and legal reform. More broadly, public service advertising is aimed at holding together an increasingly diverse and fragmented society. This paper examines the effects of public service advertising on peoples opinions. Based on experiments we investigate the extent to which public service advertising is effective and why. As the first-ever study concerning the effects of public service advertising in the Chinese media, this paper employs multiple research methods to contribute to the understanding of the relationship between the Chinese state and society. The predominant view in Chinese politics is that reforms since the Mao era have weakened the coercive power of the state. Recently, however, scholars have begun to explore the mechanisms through which the Chinese state has been able to adapt and remake institutional forces for reinforcing national unity and integration. This study contributes to this debate by exploring the role of television propaganda in influencing peoples views and behavior and finds that public service announcements can lead to negative views of the state, unless viewers recognize the messages as a state sponsored form of political communication. In the 21st century, Chinese remain surprisingly supportive of the states efforts to guide public opinion through political advertising.

3

Introduction

Political advertising constitutes one of the most important means by which political

leaders communicate with citizens. In the United States, in particular, there is extensive

evidence from experiments, surveys, and political campaigns that demonstrates that

political advertisements can and do have measurable effects on individual voters and

perhaps on the larger political system. Political advertising and campaigns affect citizen

knowledge and information levels (Gelman and King 1993, Hutchings 2001), public

evaluations of candidates and parties (Mendelberg 2001, Valentino, et al. 2004), voting

decisions (Lupia 1994, Brader 2005), and attitudes toward the political system itself

(Ansolabehere and Iyengar 1997).

Compared to political advertising, public service announcements (PSA) play a less

important role in the political science literature. In contrast to political ads, defined as one

means through which parties and candidates present themselves to the electorate (Holtz-

Bacha and Kaid 2006), PSA are supposed to educate and persuade the public on issues of

social significance (Nan 2008). Early studies of PSA have had difficulties detecting

effects on peoples behavior (see, for example, Robertson, et al. 1974), but recent

research has found that the persuasiveness of PSA, particularly on television, depends

strongly on features of the message, such as its perceived realism (Andsager, et al. 2001),

sensationalism (Stephenson 2003), presentation of exemplar versus statistical evidence

(Limon and Kazoleas 2004), overall liking (Nan 2008), and framing (Shen and Dillard

2007). As PSA campaigns have become more sophisticated, they are presumably more

likely to affect citizens attitudes towards political and social issues.

4

Recent changes in campaigning via mass media in China show that PSA (or gongyi

guanggao, in Chinese) play an important role in the way in which political leaders

communicate with and persuade citizens to accept their political leadership. During the

Reform Era (1978-present) the Chinese state modernized its strategy for political

campaigning. Today, state and party units at different levels of government rely

increasingly on political advertising, primarily on television, in order to substitute for

more traditional ways of political communication, such as, films and propaganda posters.

In the past, public service advertising was directed at building a modern set of values and

norms or spiritual civilization in order to hold together an increasingly diverse and

fragmented society. In practice, however, PSA have become part of a wide range of state

efforts to fight corruption, improve treatment of migrant workers, protect the

environment, improve knowledge of laws and regulations, and promote a positive image

of the National Peoples Congress. These commercials are shown everyday during

primetime hours on national and local television, thus reaching approximately 75 percent

of the rural and about 80 percent of the urban population (Stockmann 2009).

Yet the specific effects of PSA in China remain uncertain. Chinese leaders are

convinced that political commercials are well perceived by the Chinese public and,

therefore, are a useful form of propaganda. Officials evaluate the impact of PSAs

primarily based on the number broadcasted, which provides insights into exposure but

not the persuasion of media audiences. There is also agreement among China scholars

that contemporary campaigns have short-term effects on the behavior of the targeted

population, though compliant behavior quickly disappears once a campaign eases off

(Economy 2002; Manion 2004; van Rooij 2006). Prior research, however, has not

5

revealed the causal mechanisms that link campaigning to changes in citizen behavior. In

addition, many scholars have argued that increasing marketization, internationalization,

and technological advancement of the Chinese media have increase the ability of the

Chinese state to propagate its political goals and policies (see, for example, Brady 2008,

Zhao 2008, Landsberger 2009). Yet the effects of new forms of propaganda, including

the use of marketing strategies, on Chinese audiences remains thus far untested

empirically.

Improved circumstances for research have made such an inquiry possible. This paper

draws on unique data to study who is likely to be persuaded by PSA and why. The data

analyzed here are from four focus groups and seven experiments conducted in Beijing in

2009 and 2010. Focus groups provided us with the opportunity to develop hypotheses

about citizen responses to PSAs, while experiments allowed us to investigate the effects

of PSAs on individual levels of knowledge and attitudes. These data are complemented

by 27 semi-structured interviews with government officials, producers, and broadcasters

conducted in Beijing during 2008 and 2009.

Campaigning by Means of Public Service Announcements

Mass campaigning used to be a prominent feature of Chinese politics: the country

experienced at least one major campaign almost every year until Mao Zedongs death in

1976. Today, the Chinese state continues to use campaigning as a policy instrument, but

campaigns do not reach the same degree of coercion and citizen involvement as during

the Mao era. During the Cultural Revolution (1966-1976) the population became

6

exhausted and disillusioned about the purpose of mass movements.1 Chinese political

leaders realized that staying in power required not more but less state involvement in

peoples lives. At present, public officials reject Maoist-style mass mobilization

techniques aimed at brain washing, a Chinese expression for remolding the attitudes of

citizens through periods of intense ideological indoctrination (Lifton, 1969; Manion,

2004). The state designers of campaigns understand that they need citizen cooperation

and voluntary involvement to mobilize the public to expose problems, such as corruption,

and to aid subsequent investigation by state agencies. Instead of relying on coercion,

officials favor intensified publicity and mobilization through established legal channels

and programs to lower the costs of participation through avenues such as hotlines and

reporting centers. The state provides positive incentives for political participation in

campaigns (Manion 2004, van Rooij 2006). For those who do not choose to become

involved, however, there is no longer any fear of negative consequences.

Campaigns in China are aimed at implementing policies and educating citizens. For

example, the government has initiated legal enforcement campaigns and legal

dissemination campaigns, as part of a broader effort to build a rule of law in China.

Legal enforcement campaigns have been periods marked by swifter, stricter, and

concentrated punishment of violations of the law. There are usually three phases to

enforcement campaigns: First, the state raises awareness and mobilizes the public;

subsequently, citizens are encouraged to report non-compliant behavior in order to

1 Ordinary citizens realized that mass mobilization techniques had been used primarily as a means to solve power struggle at the top. See, for example, Mo 1976 on peoples hesitance to initially participate in the Criticize Lin Biao, Criticize Confucius campaign. Desire for change amongst the population was also clearly expressed in the Democracy Wall movement, in which workers called for rapid reforms.

7

identify targets; and finally, the targets of the campaign are punished.2 Such campaigns

include drives against crime and corruption but also against violations of law, including

copyright infringements and pirated goods, pollution, illegal arable land use, the

operation of illegal internet cafs, distribution of pornography, the production of unsafe

food and drugs, and employment of forced labor (Manion 2004, Rooij 2006). During the

early phases, citizens are mobilized to use legal channels to expose violators of the law.

Legal awareness campaigns aim to inform citizens about the law and to actively build a

legal consciousness among the population (Exner 1995). The assumption is that greater

knowledge and positive perceptions of the legal system motivate citizens to use the legal

system in order to solve grievances and thus improve policy implementation. At the same

time, participating in the legal system raises knowledge of the law. Therefore, campaigns

aimed at implementing policies and educating citizens are mutually reinforcing.

Media play an important role in both types of campaigns. Via mass media public

officials aim to create a favorable environment for the implementation of government

policies by raising awareness as well as changing peoples attitudes and behavior. Public

service advertising constitutes one means for government officials to communicate their

political agenda to citizens and gain popular support to legitimize their policies.

PSAs on television are reflective of a new approach towards propaganda work. Along

with greater marketization and internationalization of the Chinese media, the state has

increasingly relied on methods and techniques imported from the West and adapted them

to Chinese conditions and needs. For example, during the 2008 Beijing Olympics and

after the Wenchuan earthquake, the Propaganda Department relied on a public relations

2 Manion (2004) describes those three phases as three peaks: publicity, report, and punishment peak.

8

strategies and PSA campaigning to disseminate propaganda similar to the promotion of a

new product (Brady 2009, Chen 2009, Landry and Stockmann 2009, Landsberger 2009).

This separates PSA from other forms of propaganda associated with the propaganda

system under Mao, such as banners with slogans, propaganda posters, and the posting of

newspapers in public spaces, which did not rely on insights from commercial marketing.

Typically, PSA are initiated by state and party units, produced by TV stations and

advertising companies, and funded by corporations. State and party units as well as social

organizations at all levels of government all have the right to initiate PSA campaigns and

receive public funding to initiate campaigns. Due to the fragmentation of the Chinese

political system (Lieberthal and Lampton 1992), initiators usually collaborate with other

governmental units in order to increase support for campaigns. At the central level,

institutions that frequently promote PSA campaigns also incorporate the institutions that

are in charge of managing and implementing PSA policies, specifically the Guidance

Committee on Building Spiritual Civilization Construction (GCBSC), the State

Administration for Industry and Commerce (SAIC), State Administration for Radio,

Film, and Television (SARFT), and the General Administration of Press and Publication

(GAPP). Due to the Leninist nature of Chinas political system in which the Chinese

Communist Party (CCP) dominates state institutions, it is particularly useful for

campaigns to include party leadership; the GCBSC is part of the broader propaganda

system supervised by the Central Propaganda Department (CPD) of CCP. The SAIC is

the state institution in charge of issuing licenses to companies, including corporate

advertisers and advertising companies, and of implementing advertising policy more

generally. SARFT and GAPP administer broadcasting and print publications,

9

respectively. Depending on the theme of the campaign, these party and state units may

collaborate with other political and societal actors, such as, for example, the Discipline

Inspection Committee on anti-Corruption efforts.

Once the topic has been decided, initiators usually get in touch with broadcasters.

Advertising departments of TV stations then divide up production among advertising

companies, who will then approach corporations as sponsors. One reason for this division

of labor is the lack of funding for PSA campaigning. Although party and state units

receive state subsidies they can spend on propaganda, the funding is not sufficient to

purchase the desired advertising slots when TV ratings are high. The state requires TV

stations to spend at least three percent of commercial time during prime time in the

evenings between seven and nine on PSA campaigning, including holidays. 3 This

reduces opportunities for television stations to make a profit by selling commercial

advertisements. In order to mobilize cooperation, the state has therefore tacitly agreed to

several initiatives by the broader advertising community. Since 1997, media outlets have

been permitted to receive corporate sponsorship for PSA. In return, corporations are

allowed to display their name or brand logo. The PSAs cannot, however, call attention

to the brand or trademark and may not place products or information about the provision

3 Guanyu Zuo Hao Gongyi Guanggao Xuanchuan de Tongzhi (Announcement on How to Do PSA Propaganda Well), CPD, SAIC, SARFT, GAPP, 4 August 1997, available at http://www.gapp.gov.cn/cms/cms/website/zcfgs/layout3/index.jsp?infoId=450908&channelId=782&siteId=48, accessed 24 November 2009. In 2002 this rule was changed towards on average every day between 7 and 9 pm Guanyu Jin Yi Bu Zuo Hao Gongyi Guanggao Xuanchuan de Tongzhi (Announcement on gradually Improving PSA Propaganda), CPD, GCBSC, SAIC, SARFT, GAPP, 12 December 2002, available at http://www.cnlyjd.com/fagui/fagui/bumenguizhang/xuanchuanguiding/200212/fagui_1767947.html, accessed 24 November 2009.

10

of services in the ads. 4 On television corporate logos cannot appear for longer than five

seconds and may not be larger than one fifth of the total screen size.

These regulations are not always strictly enforced. Surreptitious advertising is

common for both corporations and television stations; some PSAs display the telephone

number of the corporate sponsor alongside its logo (Stockmann 2009). The involvement

of state, media, and corporate actors in the production and funding of PSA creates some

difficulty clearly distinguishing PSA from commercial advertising.

Chinese central-level officials are well aware of the use of PSA for commercial

purposes, but acknowledge that broadcasters, producers, and financial sponsors need to

gain some benefit in order to have a positive incentive to participate in campaigning.

When campaigns are of high importance to the Chinese leadership, such as the campaign

in anticipation of the 2008 Olympics or the 60th anniversary of the Peoples Republic of

China, the central government does not allow corporate trademarks to appear behind

PSA, but instead lists government institutions, most of the time GCBSC, SAIC, SARFT,

and GAPP, as sponsors for the advertisement. When broadcast on television PSAs are,

therefore, often accompanied by a commercial logo, but can also appear with references

to the television station or state and party sponsors.

Overall, PSA campaigning is representative of a new approach toward propaganda

that attempts to sell political messages in a manner similar to a commercial product. PSA

are supposed to actively change attitudes and behavior in a way favorable to state

policies. Due to the involvement of different political and societal actors in the

4 Guanyu Zuo Hao Gongyi Guanggao Xuanchuan de Tongzhi (Announcement on How to Do PSA Propaganda Well), CPD, SAIC, SARFT, GAPP, 4 August 1997, available at http://www.gapp.gov.cn/cms/cms/website/zcfgs/layout3/index.jsp?infoId=450908&channelId=782&siteId=48, accessed 24 November 2009.

11

production of PSA, the origin of political messages on television is often difficult for

viewers to determine, especially when corporate trademarks or television logos appear.

The next section discussed the implications of this new context for campaigning by

means of PSA for media effects.

Implications for Media Effects

Communication campaigns are characterized by periods in which the quantity of

media messages increases during a specific period of time. In liberal democracies

communication campaigns are often characterized by the diverse messages competing

against each other, especially when political candidates and parties attempt to convince

the electorate to cast their vote in favor of them. In contrast, authoritarian states exert

more control over the mass media, limit the diversity of political messages in the news,

and impose regularity on information flows (Geddes and Zaller 1989, Egorov, et al.

2009). Because equal access is less guaranteed for political parties, even electoral

campaigns can be largely one-sided, as, for example, during electoral campaigning in

Russia (Oates 2006).

Following a similar pattern, PSA campaigning in China increases the quantity of

messages that support the goals and policies of the government throughout the course of a

campaign, which is typically limited to a few months. During this period, a number of

PSAs on a common theme are broadcast several times a day on Chinese television,

primarily during the golden time before the 7 oclock news. Exposure to PSAs is very

common; they now reach around 80 percent of the urban and 75 percent of the rural

12

population.5 According to CTR market research conducted between January and July

2009, 67 percent of government-initiated PSA were broadcast on city-level television, 24

percent on provincial television, 5 percent on CCTV, and 4 percent by others. Television

stations all over the whole country participate in PSA campaigning and often

simultaneously emphasize several themes of local or national importance.6

Does the states heavy emphasis on broadcasting PSAs on television mean that people

are becoming more supportive of the goals that the government intends to promote? In

the American context, the effects of political campaigning are well-demonstrated.

Political ads and other campaign events can change the priorities and preferences of

voters (Kinder 2003). Most studies reveal that such effects are caused by content and tone

of political information (see, for example, Ansolabehere and Iyengar 1997), though some

focuses on group-related cues (Mendelberg 2001 2001, Valentino, et al. 2002) and

emotions (Brader 2006). In an authoritarian state ruled by a single party, such as China,

political messages that appear in television ads are not associated with different parties,

groups, or social movments as in liberal democracies. Previous research on the

development of PSAs in China suggests that television ads may be highly successful in

persuading audiences (Brady 2008, Landsberger 2009). Chinese have also been found to

5 Percentages are based on the China Survey 2008, conducted by the College of Liberal Arts of Texas A&M University in collaboration with the Research Center of Contemporary China at Peking University. It is based on a random stratified multi-stage probability sample of all Chinese adults, employing the GPS sampling technique and thus included migrant workers. See Pierre F. Landry and Mingming Shen, "Reaching Migrants in Survey Research: The Use of the Global Positioning System to Reduce Coverage Bias in China," Political Analysis 13 (2005). 6 This percentage was 6.7 percent for Beijing TV, 4.4 percent for Guangdong TV, and 3 percent for Qinghai TV. PSA were measured as advertising initiated by party or state units and did not include those PSA financed by companies, thus underestimating the percentage of PSA. Zhongguo Guangbo Yingshi (China Radio Film & TV) Magazine, December 2009.

13

be quite receptive to media influence, except when people are highly aware of an issue

and disagree with the position of the government on it (Kennedy 2009, Stockmann 2010,

Stockmann and Gallagher 2011). Yet with respect to PSA campaigning, political

messages tend to be non-controversial. It is therefore possible that PSA work as intended

by the Chinese state such that we observe a positive relationship between PSA content

and peoples support for the message promoted by the ad, if we observe any effects at all

(Content Hypothesis).

In addition to content, we are particularly interested in the effects of the source labels

attached to the ads. Source characteristics such as a speakers public approval (Page, et

al. 1987), insider status (Carmines and Kuklinski 1990), accuracy and objectivity

(Iyengar and Kinder 1985), expert status and knowledgeability (Page, et al. 1987, Lupia

and McCubbins 1998), likeability (O'Keefe 1990), party reputation (Iyengar and

Valentino 2000), ideology (Zaller 1992), and media label (Baum and Groeling 2009) can

matter for persuasion. In China, there is evidence that source characteristics also figure in

the persuasion of Chinese media audiences. With respect to newspapers people make

sense of a complicated newspaper environment by dividing newspapers into types that

vary in terms of their credibility. Newspapers that are more commercialized are perceived

as representatives of ordinary citizens and as delivering objective information.

Newspapers that are less commercialized are perceived as expert sources of information

pertaining to the government. Depending on the level of the perception of a newspaper as

credible, readers in Chongqing, Wuxi, Shenyang, Foshan, and Beijing were more or less

easily persuaded by news content (Stockmann 2007). With respect to advertising on

Chinese television, there is also some initial evidence that source characteristics affect

14

attitude change. Chan (2000) found that the image of a country as being environmentally

friendly or unfriendly influenced popular attitudes towards green advertising, perceptions

of the brand, and purchase intentions. Since PSA are usually linked with either a

corporate or a state label, we would expect that people are more supportive of the

message, depending on their perceived levels of trust in either the state institution or

corporation that sponsored the ad (Source Credibility Hypothesis). Since much research

on political trust in China has found that trust in the government is generally quite high

(Chen and Shi 2001, Shi 2001, Nathan 2003, Li 2004), we expect that the state label will

be associated with high levels of credibility.

A final goal of PSA campaigning is to change peoples behavior. Chinese policy-

makers assume that attitudinal change will lead to changes in behavior. There is manifold

evidence in social science research that this assumption holds only for certain kinds of

attitudes. In political science, political attitudes are generally viewed as less important in

fostering political action than other factors, such as socio-economic status and political

mobilization (Rosenstone and Hansen 1993, Verba, et al. 1995). In recent years, however,

there has been a revival of the argument that political perceptions, besides those related to

political efficacy, can make a difference. Political empowerment (Bobo and Gilliam

1990), self-interest (Green and Cowden 1992), intense feelings about an issue (Fiorina

2002), conviction (Burns, et al. 2002), group membership in relation to the political

context (Hutchings 2001), and emotional appeals (Brader 2006) motivate people to

become politically active. Market research also emphasizes that people can be motivated

to regularly purchase a particular product when they feel certain that one brand is better

15

than others (McQueen 1990).7 We believe that it is unlikely that exposure to one PSA,

even if resulting in attitude change, has the power to change peoples behavior.

Nevertheless, PSAs may motivate people to consider how they could achieve the goals

promoted by the message or seek information about how do to so. The content and/or

source label of the PSA should make people more likely to come up with concrete ideas

about how to address the problem raised by the PSA and motivate them to change their

behavior and seek additional information about the problem (Behavioral Intention

Hypothesis).

Overall, we would expect to find that the PSA will affect peoples support for the

political message, depend upon their trust in the PSA source, and will have effects

concerning their behavioral intentions. However, since PSA campaigns focus primarily

on valence issues, PSA may play a stronger role in reinforcing attitudes and behavioral

intention than in persuading the skeptics. Those who behave in accordance with state

policies are likely to be people who held congruent attitudes before the campaign, but are

who are mobilized to become politically active when the state signals support through

PSAs, thereby reinforcing pre-existing beliefs.

Experimental Design and Data

We use an experimental design to test the above hypotheses. Political scientists

increasingly recognize the value of the experimental method (Druckman, et al. 2006).

The main advantage of relying on the experimental method is its ability to isolate the

7 Similar results can be achieved by creating identification between a persons life style and the product, or by creating an emotional bond between the product and the activity in which it is used.

16

causal impact of communication factors on message recognition, political attitudes, and

behavior. If respondents in control and treatment groups make different choices, and if all

differences among the respondents have been randomized, then the most likely causal

source of the choice is the attribute that is manipulated by the treatment (see, for

example, Iyengar and Kinder 1987). Despite satisfactory internal validity, one potential

weakness of relying on the experimental method is that this technique is low in terms of

its external validity. Since we solely rely on samples of ordinary citizens in Beijing, we

can only be confident that experimental results apply to these samples and not all Chinese

or even Beijingers. Case studies and surveys from various cities and provinces, such as

Anhui, Yunnan, Chongqing, Liaoning, Guangdong, and Jiangsu have, however, observed

strong relationships between campaigning and citizen mobilization (Manion 2004, Rooij

2007, Stockmann and Gallagher 2011). Here, we seek to provide insights concerning the

missing causal link between campaigning by means of the mass media and citizen

mobilization.

To our knowledge, this is the first time a laboratory experiment has been conducted

among ordinary citizens in mainland China to study Chinese politics. In recent years

scholars have employed natural experimental research designs to study media effects

(Landry and Stockmann 2009, Stockmann 2010). While natural experiments are strong

with respect to their ability to observe how people behave under natural circumstances,

their capacity to isolate causal factors is weaker compared to experiments conducted in a

laboratory setting, where the experimental manipulation can be more easily controlled.

While laboratory experiments are frequently employed by scholars of psychology in

China, experimental methods have only once been applied to study Chinese politics.

17

Gries (Gries Forthcoming) has conducted an experiment among Chinese students at a

state university in Beijing to investigate the sensitivity to symbolic and material gains and

losses among Chinese and Americans. The study showed that American and Chinese

students reacted differently to the same scenarios about material and symbolic gain in

foreign politics. While convenient samples are frequently used in experiments, the use of

students raises concerns about the generalizability of results in the study of campaigning,

since students usually have limited personal experience with the political process and

therefore may be particularly vulnerable to persuasive campaign communication (see, for

example Valentino, et al. 2002). Another reason why we focused on a broad sample of

ordinary citizens and not particular social groups is that Chinese policy-makers and

producers do not have a particular target audience in mind when developing PSA themes

and producing these ads. Public officials define PSA as propaganda instruments targeted

at ordinary people (laobaixing). Similarly, none of the producers interviewed had a

specific social group in mind when creating PSAs. Instead, they indicated that they

always sought to affect a wide audience (Stockmann 2009). It is likely that producers are

indirectly influenced by the style that attracts target audiences of commercial advertising,

including most women, blue collar workers, and the younger generation (Prendergast and

Hwa 2003, Wang 2008). Since PSA are directed at ordinary people, however, we decided

to test our hypotheses with respect to average citizens. For these reasons, we recruited

adults to study the impact of political advertising on reception, attitudes and behavior.

Our study was conducted in the Research Center for Contemporary China (RCCC) at

Peking University, with a total of 210 adults with 30 participants per experimental

18

treatment in late June 2010.8 Subjects were recruited based on the snowball method,

relying on numerous entry points, and with the aim to derive a sample similar in

demographic characteristics to the population of Beijing, as indicated by the most recent

census data. The resulting sample is roughly representative of the adult population of

Beijing in terms of gender, age, and education, including 48 percent women and 52

percent men ranging between 18 and 64 years of age, with education ranging from below

than primary school level to graduate-level university education. On average, participants

were 37 years old and had received a high-school degree. Due to the large number of

migrants in China, our sample includes 43.2 percent of people who live in Beijing

without a local residency permit (hukou). All participants had lived in Beijing for at least

two years and had not participated in focus groups or experiments conducted by RCCC.

Participants were told they would receive a monetary reward for watching television

and filling in a questionnaire to understand their views about television programmes. As

people entered the RCCC facility they were randomly assigned to one of seven

experimental treatments, which are described below. A lab assistant then showed the

person to a viewing room and instructed her how to use a pair of headphones while

watching a video on a computer. All participants saw the same series of commercials and

a recently aired news program into which one of two PSAs, either with one of two source

labels, or without a source label, was inserted. After viewing the eight-minute video,

participants went into a second room to fill in a self-administered questionnaire,

developed based on insights from four focus groups and two pre-tests.9 The questionnaire

8 The sample dropped to 208 after two questionnaires were lost when scanning digital copies of questionnaires. 9 Since we could not assume familiarity with computers among all participants, we preferred self-administered questionnaires to the use of computers for completing the questions. The lab assistant guiding

19

asked people about their habits of watching television, their views on television program

content, concerns and opinions on three issues explained below, trust and satisfaction

with their government, and inclination to participate in politics. There was no mention of

the term PSA until the end of the questionnaire. The questionnaire concluded with a

note that informed participants they could obtain informational materials about how to

save water, fight corruption, and strengthen the nation by technology and science (the

subject of the news broadcast) after they had finished all tasks. To impose a small cost on

the participant to obtain additional information, the room where the materials were

distributed was located one floor above the experimental location. In order to avoid

pressuring people to obtain information as a component in their participation in the study,

participants first received a small fee for their participation upon turning in the

questionnaires, were guided to the exit, and could leave unobserved. People left the

experimental location at a different exit than incoming participants, thus eliminating the

possibility that they could influence other participants. Since the experiment continued

after people left the experimental location, we debriefed participants by mail.

Even though people did not watch television in a natural setting, the experiment

simulated reality in an important way that facilitates generalization: exposure to PSAs

was incidental and occurred during the airing of commercials prior to the start of the 7

oclock news program. This is how PSAs are actually broadcast on China Central

Television. As a result, the ads shown in the study faced the challenge of attracting the

attention of viewers in a manner similar to that of genuine PSAs.

the participant from the viewing room to the place where to fill in the questionnaire offered to provide assistance, if needed. Thirteen post-tests were conducted face-to-face, primarily among the elderly or participants with low education levels. Experimental results do not change when controlling for interviewer effects or face-to-face interviewing. Results can be retrived from the authors at dstockmann@fsw.leidenuniv.nl.

20

Experimental Manipulation

The experiment tested the exposure to one of two PSAs as well as to the source label

affiliated with the PSA. The goal was to investigate if PSA have an effect on recall,

attitudes, and behavior, and whether the source label shown after the PSA makes a

difference. Participants in one control group were shown a video without any PSA;

participants in the remaining six groups saw exactly the same video featuring one of two

PSAs without any source label, with a state label or a corporate label was inserted. In

order to permit comparison among all control and treatment groups, we randomly

assigned people into one of seven conditions as displayed in table 1.

To test whether the effects differ depending on the issue the PSA deals with, we

chose two PSA that differed by issue area. An earlier content analysis of PSAs that

received first, second, and third prices in the biannual National PSA Competition

organized by the GCBSC and SAIC revealed that ads considered important by the central

government primarily focus on valence issues; they tend to propagate building a civilized

society, protecting the environment, fighting corruption, or fostering nationalism issues

about which most Chinese agree (Stockmann Forthcoming). For example, most people

believe that one should save water, should not give bribes, should support the

motherland, and so on. The messages promoted by PSAs are primarily moral in nature,

that is, they aim to make people more supportive of the values and behavior the state

deems to be important and advantageous for society and politics. Therefore, most citizens

are already familiar with the core messages in PSAs. In the four focus groups conducted

in Beijing, we detected broad support among participants of differing age and educational

21

levels for PSA content of the above types as well as for the governments sponsorship of

PSAs, in general. (Esarey et al 2011). While most PSAs considered important by the

central government share these features, they differ with respect to their relevance to

politics. PSAs can promote major political slogans on the agenda of President Hu Jintao,

such as, for example, the 8 desirables and 8 non-desirables (ba rong ba chi, )

or social harmony, but they may also propagate equal treatment of weak groups in

society, including migrant workers or AIDS victims, and environmental protection

(Stockmann Forthcoming). As is the case for most prize-winning PSAs, the two PSAs

chosen for the experimental manipulation focused on valence issues and promoted a

message about how citizens should behave. We selected one PSA about fighting

corruption and one promoting water conservation.

These two ads differed in terms of their relevance to citizen participation in politics.

The anti-corruption PSA pertains to the problem of reining in the bribery of venal

government officialsan activity in which many citizens do not participate and may feel

powerless to prevent. The PSA about water conservation, however, concerns an issue that

can be affected, to some extent, by all citizens. Moreover, viewers of the water

conservation PSA were likely to readily perceive how the ad pertains to their daily life.

This key difference between the content of the two ads allows us to identify whether the

effect of PSAs differs by their political relevance.

Insert Table 1 about here

22

In order to make ensure that these PSAs were authentic and differed primarily in

terms of the issue, we chose PSAs that had received a prize at the biannual National PSA

Competition organized by the SAIC and GCBSC.10 A requirement for the competition

was that the PSAs had been previously broadcast. Competition judges composed of

officials, senior personnel of advertising departments in major media outlets (such as

CCTV), advertising companies, and academics. The water conservation ad, produced in

Guizhou, received a third prize in the 2005 competition and the anti-corruption ad,

produced in Yunnan, a second prize in the 2007 competition. None of the focus group

participants indicated that they had seen either of the ads, even when specifically asked.

This strengthened our confidence that each PSA would be new to experimental

participants, though people may have been familiar with their central messages.

These ads were also similar in terms of how the story was told and the type of the

behavior they sought to affect. The environmental PSA portrays a group of 15 children

arm wrestling after school. One small bespectacled boy defeats all opponents and is later

shown turning off dripping facets. The resulting physical exercise, a viewer assumes, is

the source of his hidden strength. The sentence, I want to conserve water then appears

prior to the sponsors name, Haier, a major corporation producing household electronics.

The second advertisement was part of an anti-corruption drive. The anti-corruption PSA

shows the hands of a bureaucrat on a black desk. The bureaucrat is presented with a

document, which is rejected, until a gift appears, whereupon the document receives a

stamp of approval. Next, documents accompanied by a wrist watch, a credit card, and car

keys receive approval. The next items are a pair of handcuffs and an arrest notice that is

10 We would like to thank our research collaborator for this project, Professor Liu Linqing at Communication University of China, for providing us with these materials. As the organizer of the National PSA Competition he has the right to use these PSA for further research.

23

stamped I approve. The bureaucrats hands are then shown cuffed while a melodious

voice says oppose corruption and promote clean government, know honor and

illuminate disgrace. Each of these ads includes a surprising turn of events at the end of

the ad. Both ads start with a visual narrative, but the true intention of the story is only

subsequently revealed. Similarly, in both cases viewers are supposed to change their

habitual behavior by turning off water faucets and abstaining from giving gifts to

officials. These considerations suggest that any differences we find across experiments

are likely to result from characteristics associated with the issue rather than other features

associated with the ad itself.

In order to manipulate the source label we relied on examples submitted to the

National PSA Competition to construct two source labels. Many source labels in China

have a plain background on which the logo of the corporate or state sponsor is displayed.

Especially state labels usually list several institutions as sponsor. For example, the PSA

campaign to promote the Welcome the 60th anniversary, be civilized PSA campaign to

prepare the 60th anniversary of the Peoples Republic of China in 2009 displayed the four

sponsoring organizations, including the Central Office of Spiritual Civilization (a branch

of the Central Propaganda Department of the Chinese Communist Party), the National

Commerce Administration, the National Administration of Radio, Film, and Television,

and the General Administration of Press and Publications. In-depth interviews with

officials and producers revealed that the number and rank of supporting institutions

displayed after the ad is seen as a sign of government support for the ad and thereby

increases its authoritariveness (Stockmann Forthcoming). In order to be able to compare

the effects of the content of the label we reduced the number of sponsoring institutions

24

displayed on the state label to one. We chose the most highest-ranking institution that is

usually involved in PSA campaigning at the central level for the state label the Central

Office of Spiritual Civilization, a branch of the Central Propaganda Department of the

Chinese Communist Party. For the corporate label we chose to display the logo of Haier,

a nationally recognized manufacturer of freezers, washing machines, televisions, water

heaters, air conditioners, and microwave ovens. In the past, Haier has supported PSA

campaigns. Both names were displayed in white on black background of equal size; in

case of the state label we chose the font that is commonly used in state labels for PSAs

and in case of the corporate label we displayed the actual logo of the company. Both

labels were shown five seconds after the respective PSA, which is the maximum length

permitted for PSA sponsors, according to official regulations.11

Manipulation Check

Did experimental participants receive the PSA messages while watching the

commercials prior to the CCTV news program? In order to determine if they had done so,

we measured the ability of the participant to correctly recall whether she had seen an

advertising on saving water or corruption. People were provided with a list of

commercials, including the explanation that some of them had appeared in the video and

some had not. They were asked to select the ads they had seen from the list. Since some

focus group participants had difficulties distinguishing between commercial

advertisements and PSAs, we deliberately used the term commercial (guanggao) in the

11 Guanyu Zuo Hao Gongyi Guanggao Xuanchuan de Tongzhi (Announcement on How to Do PSA Propaganda Well), CPD, SAIC, SARFT, GAPP, 4 August 1997, available at http://www.gapp.gov.cn/cms/cms/website/zcfgs/layout3/index.jsp?infoId=450908&channelId=782&siteId=48, accessed 24 November 2009.

25

question, not PSA (gongyi guanggao).12 The variables was coded 1 if the participant

correctly recalled to have seen the PSA contained in the video she had seen, otherwise

zero. Results indicate that 87.8 percent of participants who had seen the environmental

PSA could correctly recall the ad; 86.7 percent could correctly recall the anti-corruption

PSA. This small difference in the ability to recall between people who had seen the two

different ads was statistically insignificant (t = -0.08, p < 0.94). These results indicate that

about 85 to 90 percent of the audience during the golden time on television in China may

receive PSAs. It is possible that people paid more attention to commercial time than they

would when watching CCTV news at home, but even when assuming that these results

greatly overestimate peoples reception of PSAs, they still remain impressive.

Compared to recalling PSA content, being able to recall the source of these ads was

more difficult for participants. Most people were not able to recall correctly the content

of the source label displayed after the ads for only five seconds. At the end of the

questionnaire, we asked whether people had heard the term PSA (gongyi ganggao) before

and if they video had contained one. We then asked participants to describe its content

and which organization they thought had initiated the production of the ad.13 Among

people who said they had seen a PSA only four out of 60 people who had seen the state

label were able to recall the name of the Central Office of Spiritual Civilization; another

ten out of 60 named a central-level institution though not the correct one. None of the

participants mentioned the name of the corporation displayed on the corporate label.

These results may have two implications for the interpretation of the studys findings:

12 Curiously, only 28.1 percent of people who watched the anti-corruption ad correctly identified it as a PSA. By comparison, 51.1 percent of people who watched the environmental ad correctly identified it as a PSA. 13 The term organization (jigou) can refer to non-governmental organizations and does not prime participants to think about state organizations.

26

first, if we find effects associated with the source label these effects are not likely to

operate cognitively. That is, if people cannot correctly recall the source of the ads due to

the short display of the source label, but there are still significant differences between

people who saw the same ads but with different labels, these effects must be associated

with mental shortcuts that are associated with the source label itself that is not stored in a

viewers memory. Second, we may find stronger effects of the state label compared to the

corporate label, because it may be easier to link the state label to PSA content. PSA are

typically embedded within a series of commercials, all of them having a corporate

sponsor. Therefore, if a corporate label is displayed immediately after the ad, it may be

identified as a separate commercial advertising rather than as a source label

corresponding to the PSA.

Measurement

This study examines the effects of the anti-corruption and environment ads on two

aspects of attitudes and behavior. The first aspect is persuasion that is directed toward

encouraging the behavior portrayed as desirable in the ad as well as to increase popular

trust and satisfaction with the central government. The second aspect concerns the

motivation of viewers to implement the behavior portrayed as desirable by the ad as well

as to obtain information about how to save water or fight corruption.

Persuasion

A first set of attitudinal questions refers to peoples support for fighting corruption

and protecting the environment. We created two indices based on participants strong

agreement, agreement, disagreement or strong disagreement of the following statements:

27

We should protect the environment even though it may delay economic development;

We should protect water resources even though it will cause inconvenience; Ordinary

citizens should not bribe government officials even if it makes things more difficult to get

done; Government officials should not take bribes even though all others do so. These

attitudinal statements were developed based on focus group respondents interpretation of

the ads. Both scales run from 1 to 16, whereby higher numbers express more support for

each statement. When comparing the two scales people expressed slightly more support

for environmental protection than anti-corruption measures (mean_environment = 11.81;

mean_corruption = 11.24).14

In addition, we developed an index for trust in government based on peoples

response to the question: Generally speaking how much do you trust the following

organizations: CCTV, local television stations, commercial enterprises, the central

government, and the local government.15 These items were based on similar formats

asked in two national public opinion surveys, the World Value Survey and the China

Survey, in order to allow for comparison. The variable runs from 1 to 16, whereby

higher numbers represent greater trust. This indicator can provide additional evidence for

legitimization of government policies by means of PSA and source-effects.16

14 Cronbachs alpha for the environmental protection scale was 0.78, for the corruption scale 0.54. When using individual statements as dependent variables in the analysis, results do not significantly change. Results can be obtained from the author at dstockmann@fsw.leidenuniv.nl. 15 Cronbachs alpha for the political trust scale was 0.72. 16 Since the state label referred to a party institution at the central level, we also examined peoples trust in the central government separately. Results are similar and can be obtained from the author at dstockmann@fsw.leidenuniv.nl.

28

Behavioral Intention and Information-Seeking

To test whether PSA reinforce behavior we rely on subjective and objective

measures. A first question asked about a persons personal behavioral intention: Do you

have a clear idea about how to fight corruption / to conserve water? Responses were

coded as a dummy variable wherein 1 refers to having a clear idea. We also asked

respondents to write down any specific ideas they had. Answers to this open-ended

question helped us to check whether the ideas people had about how to achieve the goals

promoted by the PSAs corresponded to the content of the ads. People mentioned many

measures, but had clearer ideas about how to conserve water compared to fighting

corruption. We regarded answers that mentioned controlling the water flow of water

fossets, abstaining from giving gifts to government officials, and punishment as answers

that related to PSA content. In addition, people frequently mentioned water recycling,

technology, regulation, and education and propaganda by the state as means to save

water. With respect to fighting corruption, people mentioned greater government

supervision, media independence, changing the political system, education and

propaganda by the state, reporting corrupt officials to relevant state institutions, and

increasing salaries of officials as anti-corruption measures. If people mentioned a method

they could implement on their own (for example, reporting corrupt officials or water-

recycling, we examined their responses to the follow-up question that asked whether the

participant planned to implement the methods she mentioned in the future. This variable,

which we call motivation, runs from 1 to 7, whereby higher numbers stand for greater

willingness to implement these methods.

29

In addition, we examined information-seeking as a means to observe the extent to

which people are mobilized to change their behavior as a result of having watched a PSA.

Previous research on PSA has relied frequently on measuring peoples self-reported

motivation and information-seeking, which may overstate actual behavioral effects.

Therefore, we rely on an objective measure. The questionnaire ended with the text: If

you would like to learn more about environmental protection, strengthening the nation

though technology and science, or anti-corruption measures, we provide informational

materials in Room 409 in this building after all tasks have been completed. Room 409

was located one floor above the experimental location to impose a small cost on the

participant to obtain information. As mentioned earlier, we were careful not to pressure

participants to obtain information because of their participation in the study. In Room

409, a lab assistant provided participants with information material and was instructed to

note down which materials were taken and by whom. We interpret the decision by a

participant to seek additional information by requesting materials as a measure for

information-seeking as it requires a person to take the elevator or walk up the stairs and

search for the correct location.

Results on Persuasion

Our first hypothesis predicts that content will be positively related to support for

the message contained in the ad and political trust, if we observe any effects at all. If PSA

content works as intended by the Chinese state, we should also find that appeals to save

water and fight corruption should increase support of the government. Our second

hypothesis leads us to expect that people will become more supportive of the message,

30

depending on their perceived levels of trust in either the state or corporate sponsor of the

ad. If the state is considered more trustworthy than corporations, we should find a more

positive effect, or the reverse. To test these hypotheses, we estimate the impact of PSAs

and source labels on support for environmental protection, anti-corruption measures.

Insert Table 2 about here

Table 2 displays the result of our first test. Each column represents a separate OLS

regression equation where support for environmental protection, anti-corruption

measures, and political trust is regressed upon dummy variables for each treatment

condition, with the control group as the excluded category (reference variable). With this

specification, we can directly compare the size of the effects of the content with those of

the source labels. To guard against the possibility that differences in the distribution of

socio-demographic characteristics might account for differences observed, we control for

gender, education, holder of a local residency permit, and age.17 We also control for the

likelihood of the respondent to try to guess a desirable answer and the perceived source

of the ad. As pointed out above, the perceived source of the PSA differed considerably

from a participants actual exposure to the source label. The functional form of the model

is as follows:

uControlsabelCorporateLStateLabeltalPSAEnvironmenPSACorruptionAntirustPoliticalTSupport

+++

++!+=

43

210/""

"""

Our hypotheses hinge on the magnitude of coefficients for 1! , 2! , 3! , and 4! .

17 There are no significant differences in the distributions of these variables across cells, and excluding these controls does not alter the direction or magnitude of the results.

31

In contrast to the Chinese governments intent when propagating political content in

television ads, the content itself does not have a positive effect on either support for

environmental protection, anti-corruption measures, or political trust. Instead, exposure to

PSAs has a negative effect on these attitudes. This effect is most visible with respect to

trust in the government as displayed in both size and statistical significance of the

coefficients, but even with respect to support for the political messages contained in each

ad the coefficients are going in the same direction. We believe that this negative effect

may be caused by the fact that PSAs deal with problems that the government is trying to

solve by means of campaigning. Watching a PSA may prime people to think about these

problems, thus influencing their evaluation of the government and specific policies.

With respect to source label effects, we find that the display of the state label has a

strong and significant impact on attitudes, particularly trust in government. The

coefficient for groups that have seen the state label makes them two points more trusting

than groups that have not seen any PSA or a PSA without any label. By contrast, the

display of the corporate label is not as impressive in magnitude or statistical significance.

As in case of PSA content, these results are mirrored when analyzing support for

environmental protection and anti-corruption measures, but not statistically significant.

As mentioned earlier, these results already take into account the actual perceptions people

have of sponsorship of the PSA as we control for those perceptions in the analysis.18

Is the positive effect of the state label really a result of positive affect associated with

the central government? We believe so. During cued recall (see above), we also asked

people how much they liked the ad, provided that they recalled having seen it. We found

18 The size of coefficients for these control variables are not statistically significant. Results can be obtained from the authors through email correspondence with dstockmann@fsw.leidenuniv.nl.

32

that people who had seen the state label were consistently more likely to report that they

liked the ad, provided that participants had correctly recalled the ad. Participants who

could correctly recall having seen an ad about environmental protection tended to like the

ad somewhat without any label (mean = 3.22; s.d. = 0.7), but even better when the state

label was displayed afterwards (mean = 3.39; s.d. = 0.64). Similarly, people who could

correctly recall having seen an ad about fighting corruption somewhat liked the ad

without any label (mean = 3.0; s.d. = 0.65), but liked the state label even more (mean =

3.39; 0.57). These results suggest that people transferred some attributions of the state

label onto perceptions of the ad itself.

We are also confident that the positive coefficient for groups that saw the state label

is based on truthful responses to the question rather than coercion. First, people in those

two treatment conditions were quite frank when expressing their opinions on anti-

corruption measures. When asked to explain their ideas about how to solve corruption,

people who had seen a state label responded, for example: the power should not be in

one hand; through opening of the government, media independence, and supervision of

the masses; by establishing independent departments that have absolute

implementation rights, do not need to report to higher levels, and that have the right to

directly investigate. Similarly, in our four focus groups we showed the same

environmental and anti-corruption PSAs and encouraged people to discuss them. Just as

in the experiment, we also found that people were quite opposed to the idea that

corporations, such as Haier, used PSA as a means to promote their brand names. They

overwhelmingly favored the state as the sponsor of PSAs. This was true across age

groups, education levels, and gender (Esarey, et al. 2011).

33

Furthermore, we explicitly take into account the possibility that participants may give

responses to questions that they feel are desirable for study. Although we created an

elaborate and credible cover story to divert the participants attention away from the true

goal of the study, some participants still tried to guess what they perceived as the right

answer. We could compare peoples answers to the recall of ads and their descriptions of

the PSA in the end of the questionnaire with the actual ads they had seen. We also took

into account whether people were able to provide a description of an ad when stating that

the video they watched had contained a PSA. Based on these responses we constructed a

scale to assess a persons likelihood to guess the right answer and controlled for it in

the analysis. About 50 percent of people did not attempt to guess, 38 percent guessed a

little bit, 11.1 attempted to guess, and 1.4 were highly likely to guess.

Behavioral Intention and Information-Seeking

Our third hypothesis predicts that content and/or source label of the PSA should make

people more likely to come up with concrete ideas about how to address the problem

raised in the PSA and to seek information about it. If PSA content works as intended by

the Chinese state, we should find that appeals to conserve water and fight corruption will

motivate people to implement the measures displayed as solution to the problems of

water shortage and corruption. To test this hypothesis, we estimate the impact of PSAs

and source labels on peoples likelihood to have a clear idea about how to save water and

fight corruption. If people had ideas that they could implement in their personal lives,

they should also express motivation to employ these measures in the future.

34

Table 3 displays the result of our second test. Each column represents a separate

regression equation where having a clear idea about how to save water, fight corruption,

and peoples motivation to implement their proposed solutions is regressed upon dummy

variables for each treatment condition, with the control group as the excluded category.

Since having a clear idea is a dummy variable, we employ probit maximum likelihood

estimation, while ordinary least-squares regression analysis is used to examine effects on

motivation. In all cases, positive coefficients indicate a higher behavioral intention. As

before, we use the same functional form of the model and include the same control

variables.

Insert Table 3 about here

In contrast to persuasion, we find that PSA content and labels have different effects,

depending on which issue the PSA deals with. As mentioned earlier, environmental

protection is more closely related to peoples personal lives than corruption, which relates

primarily to government officials. Therefore, PSA content on water conservation inspires

people more than the anti-corruption PSA to think about how they personally could

contribute to addressing the problem.

The average Beijinger in our experimental sample (a man with a high-school degree,

37 years of age with a Beijing residency permit) was about 35 percent likely to have a

clear idea about how to conserve water without having seen the ad. Yet just having

watched the environmental PSA made him 58 percent likely to express a clear idea. The

content of the ad made 23 percentage points of a difference. By contrast the source label

35

had the reverse effect. While we still observe a positive effect for the state label it is this

time not statistically significant. Instead, the corporate label turns out to make people

more resistant to the message contained in the ad. The average Beijinger in our sample

was only about 19 percent likely to express a clear idea about how to save water when

having seen a corporate label, 16 percent less likely than when not having seen it. For the

environmental PSA, the corporate source label mitigated the positive effect of the

message contained in the PSA. It is important to note that the ideas expressed about how

to solve the problem of water conservation were not directly taken from the ad itself.

While many people mentioned turning off the water faucet, they were not more likely to

give this response when having seen the environmental PSA. Therefore, we believe that

the ideas displayed as desirable in the ad served as a stimulant, but not necessarily as the

blueprint for behavior. Somewhat surprisingly, this stimulation to think about solutions

did not lead to higher motivation among participants. None of the treatment conditions

stood out as motivating people to express an interest in seeking information about how to

conserve water.

With respect to corruption, we observed almost the reverse effects. As in case of

persuasion, the content of the anti-corruption PSA related negatively and the state label

positively with people having ideas about how to solve corruption. Yet none of these

results were statistically significant. This is probably related to the fact that most people

did not see the content of the anti-corruption PSA as relating to their personal lives. Only

few respondents mentioned anti-corruption measures that people could initiate

themselves, such as abstaining from giving gifts to officials or reporting corruption. Most

people raised measures that required the state (primarily the central government) to take

36

the initiative through regulation, political reform, or raising the salaries of officials. This

disconnect between the anti-corruption PSA and ordinary citizens was consistent with our

four focus groups in which people welcomed the corruption ad, although many

considered it to be aimed at deterring officials from taking bribes (Esarey et al, 2011).

Despite the small numbers of participants who had some ideas about how to address

corruption themselves, once people expressed such a notion they felt more motivated to

implement it, provided that they noticed a source label. Comparatively, the effect of the

state label was more powerful than the corporate label, although just having seen a well-

known brand name after the PSA made people about 1.6 points more motivated, which is

about 23 percent on the 7 point Likert-scale. On the whole, viewing the PSA made people

less motivated to implement anti-corruption measures.

Once we imposed a small cost on peoples motivation to change their behavior or

seek information about how to do so, even the strong effect of the state label with respect

to anti-corruption measures fades away. Among our sample of 210 Beijingers, only two

people took the initiative to ask for additional information. One was a 22-year-old highly

educated male had seen the environmental PSA with a state label and the other one a 44-

year-old less educated woman who had only seen the anti-corruption PSA, although

without any label. Only the latter took the time to find the room and get information. She

picked two reports that explained how to fight corruption, three about how to conserve

water, and one about how to strengthen the country through advancements in science and

technology (the topic of the news report). Neither person had indicated in the

questionnaire that they had concrete ideas about how to address water shortage or

corruption.

37

Overall, while we found some effects of PSA content and source labels on peoples

behavioral intention and motivation to solve problems in the future, these effects face

hurdles when citizens are making choices about how to behave in practice. While this

study shows that one individual PSA may not have the power to make people give up

something else (in this case time) in order to learn more information about saving water

and fighting corruption, these hurdles may be able to overcome when PSA campaigns

create an environment in which failure to comply with the behavior displayed as

desirable in the ad becomes costly. This may explain why case studies of specific anti-

corruption campaigns or the Welcome the Olympics, be civilized campaign in 2008

have found a relationship between PSA campaigning and aggregate changes in behavior

among ordinary citizens (Manion 2004, Brady 2009).

Implications for Political Trust in China

Our findings on the positive effect of the state label on trust in the central

government provide a definite answer to a key question that is debated among researchers

of regime legitimacy, satisfaction with government, and political trust in China. It is a

well-established fact that the Chinese government, particularly the central government,

enjoys high levels of political support as expressed in public opinion polls general. In a

survey conducted in 1993, 94 percent of the respondents expressed trust and willingness

to obey the government (Chen and Shi 2001). Surveys conducted in Beijing in 1995,

1997, 1999, 2001, and 2002 are roughly consistent with these findings (Chen, et al. 1997,

Chen 2004, Wang 2006). Trust in local governments tends to be somewhat lower,

38

whereby it decreases when moving from the center to provinces, counties, townships,

towards villages (Li 2004).

A first group of researchers treats these survey responses as truthful responses

reflecting a support for the rule of the CCP and explores the sources of popular support

for the government. For example, (Shi 2001) argued that traditional values such as filial

piety and respect for hierarchy exert a strong influence on high levels of political trust in

both China and Taiwan. (Nathan 2003) speculated that diversification of the media plays

an important role in explaining Chinas high levels of political trust. (Chen and Shi 2001)

found that government propaganda has not influenced popular levels of confidence in the

government, while (Kennedy 2009) showed a strong positive relationship. (Stockmann

2010) argued that these differences can be explained by varying levels of media

credibility and that marketized media remain powerful propaganda instruments in China.

In addition to traditional values and propaganda, scholars have also found that financial

well-being and improved living standards has made citizens satisfied with government

performance (see, for example, Wang 2006).

Yet many of these studies have received strong criticism - less in publications

than at conferences - by researchers that suspect a large social desirability bias in Chinese

survey responses to questions on the political legitimacy of the Chinese government.19

This criticism is often based on a view that the Chinese political system relies strongly on

coercion to sustain its rule. As a result, citizens are afraid of the consequences when

revealing dissatisfaction and are therefore hesitant to reveal the true opinions about their

19 On a similar rationale underlying this argument see the dictators dilemma (Wintrobe, 1998).

39

satisfaction with politics and the degree to which they support the regime.20 In response,

researchers of the first group usually bring up measures of political fear based on survey

questions (Li 2004, Wang 2006). Unfortunately, without systematic list experiments

survey questions alone cannot measure political fear accurately, since people who are

fearful are unlikely to give truthful answers.21

Our experimental results clearly dispute such an interpretation of the results. In

our experiment, affect towards the Chinese central party-state is automatically activated

upon mere exposure to the name of a central party institution, leading to greater

persuasive effects of PSA on political trust. In our experiment, people were engaging in

online-processing of information: when people receive a piece of information, for

example a political candidate, the affective component affiliated with a concept (for

example, the candidate) is stored in long-term memory, but the actual content that lead to

the final evaluation is lost (Lodge, et al. 1995 1995). When people were asked how much

they trusted the central government, they gave more positive answers when the concept

its associated affective component was prompted. They even transferred positive

attributions of the state label onto their evaluations of the PSA message.

These results are significant and strong even though our manipulation check

revealed that people were unable to correctly recall the source of the PSA, even when

looking for general references to central government institutions more broadly. The

inability to recall the state as a source makes the affective response affiliated with the

20 See Zhengxu Wang, "Explaining Regime Strength in China," China: An International Journal 4, no. 2 (2006). for a summary of this argument. 21 On list experiments see Daniel Corstange, "Sensitive Questions, Truthful Answers? Modeling the List Experiment Multivariately with Listit," Political Analysis 17, no. 1 (2009), Adam Glynn, "What Can We Learn with Statistical Truth Serum? Design and Analysis of the List Experiment," in Midwest Political Science Association Conference (Chicago2010)..

40

state label extremely difficult to be overturned by a later conscious, cognitive assessment.

Yet this cognitive assessment is arguably required, if people would overstate their trust in

the central government. Social desirability bias occurs after a person has made up their

mind on a survey question when reporting the opinion to the interviewer (Berinsky 2004).

To over-report trust in government would arguably require that people who sat the state

label would remember having seen it in the ad, thus being more sensitive to questions

related to the government. In addition to our results of liking, the inability to recall the

state label further strengthens our confidence that over-reporting occurs as a result of

political fear.

Insert Table 4 about here

This is not to say that we did not observe any social desirability bias. By contrast,

we actually discovered that about 50 percent of participants were giving overly positive

responsive to our questions. Since we knew the specific content each control and

treatment groups video, we could check the answers people gave to a set of recall

questions with the actual content they had been exposed to. We then created a dummy

variable for people who gave overly positive responses, including people who said they

recall watching an ad on corruption or environmental protection while in fact they had

not seen it, people who said they had seen a PSA, but guess in an open-ended follow-up

question that they must have seen a PSA on the environment or corruption because there

were many questions on these issues in the survey questionnaire, but in reality they had

not seen the ad. 46.6 percent of our sample gave at least one of the above responses,

41

which we believe was motivated by a desire to please the researchers, even though the

true intention of the experiment was only revealed to participants afterwards.22 Therefore,

we believe that high levels of political support for the central government in Chinese

surveys may be partially inflicted as some participants have a tendency to give positive

responses to survey questions. However, overly positive responses are unlikely to result

from political fear: When we examined the characteristics of participants who engaged in

this bias, we found that having watched the state label did not prompt such a response.

Instead, people were most likely to engage in guessing the right (and positive) response

when they did not have a Beijing hukou or had low levels of income (see table 4).

Overall, these findings lead us to conclude that high levels of political trust in

China is indeed exaggerated, but not due to fear of repression, but possibly because of

participants intention to please the interviewer. In our experiment, the average level of

reported trust in the central government was 3.38 (s.d. = 0.69), while the average level of

precisely the same survey question asked in the World Value Survey 2001 was 3.36.23

Although the experience of participating in a survey conducted by face-to-face interviews

may be less artificial than in this experiment, these results suggest that guessing may also

influence survey responses. To our surprise, people who were trying to please us, by

guessing the right answer did not report higher levels of trust in government.24 Instead,

the average person participating in our experiment was less likely to be supportive of

22 A scale developed based on these questions was included as a control variable in the statistical analyses presented in the previous section. 23 Calculated based on descriptive statistics in table 1 in Wang, "Explaining Regime Strength in China," 221.. 24 In table 2 and 3, guessing was a control variable. In these statistical models, guessing is only substantially significant for dependent variables displayed in table 1 (support and political trust). Coefficients are -3.967*** (s.e. = 1.452); -2.903** (s.e. = 1.324); -1.510 (s.e. = 1.120) on support for anti-corruption, environmental protection, and political trust, respectively.

42

either anti-corruption or environmental protection measures and to express support for

central-level political institutions. We are not sure why guessing worked in favor of

positive responses in one instance, but not in the other. It may be the case that this

particular group of people only tried to guess when asked about the content of the video

they had seen, but gave honest responses to questions unrelated to the content of the

treatment. We are highly confident, however, that guessing was not caused by political

fear prompted by exposure to the concept of the Chinese central party-state.

Conclusion

Assessing the effects of political advertising on attitudes and behavior in China is

a challenging enterprise. Nevertheless, an approach drawing upon multiple methods,

including interviews, focus groups, and experimental testing has promise for advancing

scholarly understanding of the extent to which modern, marketized propaganda affects

the worlds largest television audience. An advantage of the experimental component of

this study is that it allowed us to determine that political advertising, to which the

Chinese state is deeply committed, actually does affect public opinion and to some extent

how. This objective was particularly important because our focus group data suggested

that viewers may instinctively tune out advertisements by changing the channel or

engaging in other tasks. The finding that more than 85 percent of experiment participants

could recall the ads they saw suggests that people do receive the messages of PSAs,

although in the case of the experiments the participants did not have the choice of

switching the channel. PSAs appear, therefore, to serve as a powerful form of political

communication.

43

The experiments revealed a complex picture of PSA effects. Contrary to our

expectations, advertisements that mention problems, such as water conservation or

corruption, actually reduce trust in government, which is the opposite of the effect

desired by the government. This finding can perhaps be explained by the fact that in a

country ruled by a one-party regime with hegemonic powers there is often only the state

to blame for political problems in the absence of clear avenues for efficacious citizen

participation in politics. Yet when viewers recognize PSAs as state sponsored messages,

the PSA seemed more likely to increase trust in government. Ostensibly, the viewers

assumption was that state support for issues such as anti-corruption efforts reflects the

genuine desire to improve governance. This indicates that PSA effects may swing based

on audience attentiveness and the presence of a state label. Counter-intuitively, when

observers do not identify the state as the source of an ad, broadcasting PSAs that

highlight social ills may well prove counter productive. Effective manipulation of

propaganda to increase public support for the state, in other words, may be a more

delicate task than the regime comprehends.

The difference between the effects of the PSA treatments concerning anti-

corruption efforts and water conservation was also instructive. People seemed to view the

former as the purview of the government and as a problem that only the state could

address. In fact, bribery is a two-way street that intimately involves some, although not

all, citizens. If people are willing to bribe officials, in a context in which there is little fear

of punishment on either side, officials are likely to accept bribes. In China, however, the

public places blame on the corrupt official who, social mores suggest, should be above

such conduct. The PSA pertaining to water conservation did not spur many people to

44

assume the minor cost of collecting additional information, but it did get people to think

about solving the problem, even if it was not clear they would upon act upon the

solutions. One explanation for this could be a sense of apathy among the Chinese public

that is rooted in the view that individual actions related to water conservation will have

little impact on the problem.

In both the focus groups and the experiments, we found compelling evidence th