Civic Action and Media Perceptions within the Wall The (

: Little has been known about China’s policing of the Internet until recently, when researchers began publishing insights on the types of message that gets deleted and permitted on various social media platforms, as well as whether or not such moderations are performed automatically. Many discussions have focused on how such efforts may undermine the democratic potential and civic actions that may be empowered and facilitated by the Internet. Two cases discussed in this paper show a different picture: the aftermath of a train collision in Wenzhou in 2011, and an elaborate plan by a company to take out its competition – both utilizing social media. Structuration theory is used to analyse the types of agency, structures, and power negotiations that can be observed in both cases. The paper then reports a survey carried out with 499 participants on their perceptions of both cases, focusing on how trust propensity and types of information may shape their perceptions of media credibility. Results show that trust propensity was only significant in shaping perceptions of credibility for social media, but the types of information is significant in shaping perceptions of credibility for both mainstream and social media. Implications are drawn for media literacy as well as how civic actions function within China.


China's Internet policy
Especially in the case of China's use of social media platforms, it is important to start the paper by clarifying the context of analysis . The Internet has become an 'unclear concept' (Salter, 2003, p. 118). Its technical definition is clear enough, as a networked platform enabling various information and communication applications such as email, the World Wide Web, electronic databases, and social networks such as Linked In and Facebook. It is its social construction that is unclearquestions of who can use it, and whether such use should be regulated remain unanswered and hotly debated. Governance of the Internet has to recognize the social and cultural contexts working in each society; any attempt at trying to come up with a uniform, unified set of recommendations sets itself up for failure. Notwithstanding the uncertainty about its social construction, the Internet has become a globalized and familiar concept, thanks to the advent of the World Wide Web. The Web, originally innovated as an application to link textual documents together, has now become so accepted as the main access point to the Internet that it is sometimes misunderstood as the Internet itself. With the Web, a new type of application emerged in the late 1990s, as part of a growth of publishing tools to allow non-technical users to publish content on the Web. Originally known as 'web log' to describe the nature of it as personal journals on the web, the term 'blog' was coined by Peter Merholz in 1999(The Economist, 2006. Since the innovation of the blog several other similar platforms have emerged, with design attributes such as the ease of producing, sharing, commenting, following, liking content, and the formation of social networks consisting of both bonding and bridging ties. Friendster, Twitter, YouTube and Facebook are common examples. In many societies the participatory and networking attributes of social media platforms on the Internet have been argued to bring about substantial impacts, such as the diffusion of ideas, transforming political and civic engagement, and bridging digital and developmental divides. It would however be erroneous to think that given these attributes, the Internet is inherently democratizing. As Lessig (1999) argued, the Internet is regulated in some extent by both democratic and authoritarian governments by controlling its code and the legal environment it operates in. To date, studies of the Internet's impacts have been focused on its democratizing potential, and less about its impacts in the context of authoritarian regimes, in part due to the unavailability of access to such sites of inquiry by researchers.
The platforms that are so popular in many societies are however forbidden in China. As Xue (2005, p. 241) puts it, 'the Chinese government has been in a 'love and hate' dilemma towards the Internet'. It was not until the 1990s that China developed a national information policy that included governance of the Internet; before that the government dealt with issues arising from the use of the Internet on a case by case basis. In a white paper released in 2010, the Chinese government acknowledged that the value of 'human wisdom' to be found in the Internet, but asserted its rights to censor by maintaining that potential harm from the 'spread of information that contains content subverting state power, undermining national unity [or] infringing upon national honor and interests' should be curbed (Bristow, 2010). As Xue (2005) noted, the desire by the Chinese government to maximize the positive benefits of the Internet has always been balanced by the conscious aspiration to protect the society from potential negative influences which may destabilize its sociopolitical culture. The operationalization of this principle is manifested by the blocking of sites perceived as sensitive by the Chinese government using a variety of technological tools. Popularly referred to as the Great Firewall of China, the types of posts and sites that get blocked are commonly believed to be those containing criticisms of the government. Studies from Harvard however, revealed that the story is more complex than that. According to Gary King who heads the studies, content and sites that are blocked are instead those focused on collective action (NPR, 2012).
The need to embrace the democratic potential of the Internet, whilst still upholding tight controls over it is manifested by the presentation of alternatives to those blocked by the Great Firewall. The critical tension faced by the Chinese government in having to manage the risks by opening its doors to the Internet is perhaps best reflected in the following quote from the People's Daily in 2005: "As long as we use more ways of properly looking at the Internet, we can make use of the best parts, we go for the good and stay away from the bad and we use it for our purposes, then we can turn it around on them. Just as we can defeat the well-armed American military in the Korean war of yesteryear, we won't be defeated in this huge Internet war by the various intranational and international reactionary ideological trends in the various areas" (Wu, cited in MacKinnon, 2008).
Residents in China are given other choices, such as Sina Weibo (a microblogging platform similar to Twitter), Qzone and Bokee (blogging platforms similar to MySpace and Blogger), and YouKu (a video sharing platform similar to YouTube). Such alternatives are growing rapidly, with more than 500 million Internet users, 100 million bloggers and 300 million active microbloggers (Hassid, 2012) at the time of writing. Through these platforms, residents are still given opportunities to engage in democratic discourse and collective action, although they are moderatedsuch as CC: Creative Commons License, 2013. the deletion of selected messages from Weibo (Bamman, 2012). Bamman's (2012) study affirmed the suspicion that political criticisms were not tolerated and quickly deleted. On the other hand, the action also worked for common social good on other occasions, when false rumors can be quickly dispelled by deleting messages attempting to disinform.
The background and studies behind these policies give greater clarity to the social construction behind the modus operandi of the Internet in China. Although the Internet is open for use by residents, it is not without boundaries that are continuously defined. It is not that the democratic potential of the Internet is underminedthe story is much more multifaceted. Instead, it confirms Drezner's (2010) conclusion that the Internet 'merely reinforces pre-existing dynamics between states and non-state actors ' (p. 39). Especially in the case of China, the use of the Internet and social media platforms requires contextual understanding of the power relations between the state and non-state actors, the latter including residents, mainstream media, business organizations, and non-governmental organizations.
As noted by Sullivan (2013), given that there is still much room for Internet penetration to grow in addition to the fact that Chinese Internet users are exceptionally active in their production and use of online information (Yang, 2009), the need for more research to examine the social impacts of the Internet in China in specific incident-driven contexts. The link to the empirical realities of specific cases is crucial, as they can provide opportunities to develop relevant and valid theories (Glaser and Strauss, 1967). The principle guides the development of the paper, as I start off by discussing details of two cases involving the use of social media in China to provide at least the empirical realities in which subsequent research questions are grounded in. The theoretical framework follows after, giving rise to relevant research questions. Methodological details are then discussed, followed by findings and implications for future research forays into the field.

The bright and dark sides of social media
The story of China's regulation of the Internet is one that is driven by contentboth the diffusion and censoring of it. Internet penetration is encouraged, but there has always been tight control of content. Despite the heavy monitoring of information flow, the use of social media by users is far from primitive, as illustrated by the following cases.

The scandal concerning Mengniu Dairy Company
In July 2010, Mengniu Dairy Company, the largest product manufacturers in China, was involved in a highly publicized smear campaign of spreading false rumors about its competitors, Yili Industry Group and a lesser known competitor Synutra International that their milk products for children and infants contained harmful ingredients which would affect juvenile development. Mengniu was alleged of spreading stories using a variety of social media platforms, sending the message that Yili's infant formula was mixed with Eicosapentaenoic Acid (EPA) that could lead to premature sexual development. In another story the company was accused of spreading, was that Synutra product would lead to the growth of breasts in baby girls.
The stories went viral in July 2010 and there was a sudden surge in articles and news posted online on the various Chinese news portals, discussion forums, blogs, micro-blogs, QQ groups and other social media platforms which was damaging for the two diary companies, particularly for Synutra as its stock took a plunge after the online report. By October 2010, investigations conducted by the local Ministry of Health and the police revealed that products from both dairy companies were in fact, safe for consumption. They were allowed to continue to sell their products. The drama climaxed at the point when local authorities found that the whole saga was engineered by An Yong, a brand manager at Mengniu, and three other employees from BossePR Consulting. The smear campaign was referred to as "731 Plan" by An Yong and his accomplices. It was a strategy to defame and take out its competitors.
The elaborate plan included hiring people who pretended to be pregnant women, and others to post nearly 100 comments online to spread the message that Yili's formula contained EPA which might cause premature sexual development. The '731 Plan' succeeded, as it took advantage of the Chinese consumers' growing mistrust of domestic food companies, as well as the viral capacity of social media. Despite interventions from the authorities intending to correct the rumours spread about Yili and Synutra, the elaborate use of social media to disseminate information has resulted in an erosion of trust in dairy products in China, persisting even today (although the story has been made more complex with a number of other incidents of tainted milk powder), as reported by Xinhua News (Xinhua, 2013).

The train collision in Wenzhou
On 23 July 2011, two high-speed trains travelling on the Yongtaiwen railway line collided on a viaduct in the suburbs of Wenzhou, Zhejiang Province, China. The two trains derailed and caused four cars to fall off the viaduct. The accident killed nearly 40 people and injured 192 others. Rescue efforts ended within a day, with train cars being broken up and buried quickly. Hours after rescue efforts ended, a 2-year old girl was however found alive in the wreckage.
During the press conference, the Railways Minister Wang Yongping explained that the accident occurred due to equipment malfunction caused by lightning. Pictures of the accident and of the wreckage being buried were however, widely circulated by witnesses on the Internet. These led to criticisms from local media that there were inadequate investigations, mishandling of important evidence by the authorities, and rescue efforts were inadequate and ended prematurely. The authorities also attempted to muzzle reports of the event on local Chinese media, by disallowing reporters to enter the scene, and reports should not suggest that it had to do with the high-speed train technology used on both trains.
Despite the media directives issued by the authorities, this event witnessed several key anchors from various Chinese media openly criticizing the government in the way they handled the accident. One news anchor, Bai Yansong, said on CCTV News (a national television channel) in response to Wang's assurance that the train technology itself was advanced (Tan, 2011): "The technology may be advanced, but is your management advanced? Are your standard operating procedures advanced? Is the supervision advanced? Is your respect for people advanced? Are all the minute details advanced? At the end of the day, is your overall operational capability advanced?" (translated) Subsequently by 29 July, all coverage of the event was banned except for the ones focusing on rescue efforts or other 'positive news'. Chinese Internet users, however, continued rallying around the cause. Utilizing a variety of social media platforms, they posted or spread messages varying from calling for the 'truth' to be reported, criticizing mainstream media, parodying Wang's statements about the accident, to demanding for the resignation of Wang.
The actions that took place only on social media platforms such as Renren, Qzone, and Weibo gained international traction, with international mainstream media such as The Guardian reporting the drama that was unfolding amongst Chinese Internet users in the wake of the event. Wang was eventually dismissed from the Ministry of Railways on 16 August 2011.
The two events unfolded two contrasting potential of social media platforms in China. In the Wenzhou train collision disaster, citizens exerted their agency by using social media platforms to express and spread their dissent, share information that were censored from mainstream media to create greater awareness, and rallied together to call for the resignation of the railways minister. Despite attempts to censor such messages from the local social networks (Bamman, 2012), the intensity and speed at which Chinese Internet users posted and spread the messages concerning the Wenzhou train disaster, focused at unraveling the inadequacies of the investigations and the 'incompetence' of Wang Yongping were unrelenting. Wang became notorious online, was eventually dismissed from the post (although there is insufficient evidence to link the case and his dismissal directly, the case was believed to have contributed to his removal from the post) and suffered extreme stress. Whether the toxicity of the messages was necessary is perhaps secondary to the fact that Chinese Internet users were able to demonstrate collective agency when perceptions that investigations and rescue efforts were inadequate were rife. Despite structural mechanisms that were in place to constrain their actions, Chinese Internet users were able to utilize the resources they had to negotiate and renegotiate power relations between them and the state. Yang's (2009) observation of Chinese Internet users in their voracious use of the Internet to share and use information is perhaps best exemplified in this case.
In contrast, the Mengniu Dairy Company scandal unveiled a less desirable aspect of social media. With an elaborate plan in place, Mengniu was able to spread false rumors to the extent that it translated into real damages to the competition. Chinese Internet users fell prey to the rumors, and even after the truth was revealed many remained confounded and skeptical. Paralyzed by fear that these rumors might be right, they continue to remain vigilant about the credibility of information coming from both mainstream and social media. While Mengniu was able to exercise much agency in their utilization of local social media platforms, it might come at the cost of eroding trust in both mainstream and social media platforms.
The lessons that can be drawn from both cases reflect a particular insight offered by Giddens (1984), in that the relationship between human actions and structures (such as the media environment, legal and power institutions) is recursive. Actions and structures being closely and continuously interdependent enable the construction and reconstruction of new knowledge and meanings by people living and working together in societies. Structuration theory views structure as 'both the medium and outcome of the reproduction of practices' (Giddens, 1979, p. 5). This is also referred to as the duality of structure (Giddens, 1984), which sees that the institutional properties of social systems are created by human actions, and in turn shape future actions. Specifically, it is defined by Giddens (1984) as 'structure as the medium and outcome of the conduct that it recursively organizes; the structural properties of social systems do not exist outside of action but are chronically implicated in its production and reproduction' (p. 374). In other words social structure is continuously made and remade as a consequence of people's actions, and at the same time provides the context in which such actions take place. Thus action and structure are inseparably linked in an ongoing interplay through which the institutions or patterning of society, and the knowledge which underlies them, are created and re-created.
Such knowledge is driven through interactions that characterized Chinese Internet users in the Wenzhou train collisionresulting in discursive consciousness about the situation. In the Mengniu Dairy Company scandal, the discursive consciousness created through the messages posted via various social media platforms may have translated into decisions to boycott products of the companies in question.

Trust and the Perceptions of Information
With the increased adoption of alternative social media platforms in China, the quantity of information available to residents in China has increased significantly. Information found on such platforms are characterized by 'conversational media' and are made up of recursive co-production of content, in contrast to the one-way dissemination of packaged information found in traditional and mainstream media.
Although this implies that people may turn to social media platforms for alternative information, mainstream media channels are still instrumental in shaping human perceptions of events as they unfold. This is perhaps best described by McLuhan (1964) as: 'societies have always been shaped by the nature of the media by which men communicate than by the content of the communication'. Studies addressing the trustworthiness of media confirm a positive relationship between the usage of media and the perceived credibility of the media (Rimmer and Weaver, 1987). This implies that the more an individual uses a particular media channel, the greater is his trust in the credibility of the media. The finding may help to explain McLuhan assertion that the media may shape people more than the content itself.
The media channel usage effect may however, be undermined when factors are present to constrain traditional mainstream media's ability to acquire information to meet the information needs of residents. In some cases concerning the risks associated with food produced in China for example, mainstream newspapers' reports were not as timely and contained facts released by the state. Compared to their social media counterparts which contained first-hand accounts and warnings about various food products, the value of information found in mainstream media as perceived by residents may be lower. Although this could mean that there may be more negative information present in social media platforms, Siegrist and Cvetkovich (2001) found that people may trust negative information more than positive information.
Studies have addressed the question of users' perceptions of media credibility (Bracken, 2006;Choi et al, 2006;Lu and Andrews, 2006;Johnson and Kaye, 1998). The issue of trust in a media is particular important for information seeking behavior research, as it implies that individuals may spend more time seeking information from sources that he perceive as credibleconstructing certain trajectories of information seeking involving formal or informal information sources. In their survey of politically-interested Internet users, Johnson and Kaye (1998) found that they had a higher perception of online media's credibility compared to traditional media. However, participants in their study had moderate perceptions of both types of media.
The combined effects of trust in mainstream media channels, and trust in alternative information available on social media platforms can therefore be confounding. The situation is even more complicated given China's policing of the Internet. Indeed, as MacKinnon (2005) inferred, China's policing of the Internet may contribute to the pursuit of nationalism amongst its residents since it effectively provides a moderated and somewhat packaged view of events and the world. Guo et al's study (2005) may have confirmed this inference when it found that most Chinese Internet users actually indicated greater trust in local sources of information than foreign sources. This is one unique feature of Chinese Internet users compared to Internet users from other societies. However, when it comes to differential reports of events on local mainstream media and social media platforms how do Chinese Internet users cope with the dissonance? Specifically, what implications does it have in terms of their trust and credibility of different media platforms?
Given the circumstances in which social and mainstream media were used in both events, the paper focuses on the effects on trust in social media and mainstream media. As Chinese citizens demonstrated agency through their use of social media in the Wenzhou train collision event, it was expected that they might have greater trust in negative information which was circulated on social media in that event. On the other hand, if they were informed about how Mengniu Dairy Company had exploited social media to spread false rumors about their competitors, it could be argued that they would have shown greater distrust in negative information, and greater trust in the credibility of information circulated on mainstream media.

The propensity to trust
In seeking information to meet information needs, individuals often have to depend on cues to help them determine the quality and credibility of information. For example, the source of and peer ratings of information may provide indications about the credibility of information. In a site like Wikipedia, the number of edits and references used to support the information on a page may provide cues to ascertain information credibility. However, users may not always be able to pick up these cues, or be able to use them effectively. Knox (1970) argued that internal values i.e. the individual's propensity to trust may instead determine how they assess the credibility and trustworthiness of the information they are encountering.
The issue of trust has been examined from a variety of disciplines such as communications (Giffin, 1967), management (Scott, 1980), game theory (Milgram and Roberts, 1992), and labout relations (Taylor, 1989). McKnight et al (1998) defined it as the belief in a source or a person, to the extent that would translate into a willingness to depend on them. In the context of the paper, it refers to the extent to which individuals depend on the information being reported in either mainstream or social media platforms as an accurate reflection of the realities in both cases. Defining it as 'reliance upon the characteristics of an object, or the occurrence of an event, or the behavior of a person in order to achieve a desired but uncertain objective in a risky situation', Giffin (1967) highlighted the relevance of trust as a way for an individual to manage uncertain or threatening situations.
Building on this definition, Mayer et al (1995) constructed a model of trust by explicating the antecedents as well as the outcomes of trust. Specifically, they provided the concept that individuals may differ in their propensity to trust a given object, event, or person. Trust propensity may also be the most salient factor shaping trust especially in new or uncertain situations (Bigley and Pearce, 1998). Indeed, as a meta-analysis conducted by Colquitt et al (2007) shows, a direct relationship can be found between trust propensities on trust, although this relationship may be weak.

Methodology
A survey questionnaire was set up online using Survey Monkey, and advertised via Renren, Weibo, Douban, and Qzone between 1 August to 28 September 2012. Participants had to fulfill two screening criteria before they were allowed to complete the questionnaire: a) they use both mainstream and social media platforms in China, and b) they have used them to seek information about both cases (the Wenzhou train collision and the Mengniu Dairy Company scandal). In total 499 responses (N=499) were collected from 10 cities in China, and the mean age of participants was 27.5 years old (SD=6.7) within a range of 18-50 years. Given that the average age of typical Chinese Internet users is 28 years old (Liu, 2010), the resulting sample is representative in terms of age. However, given that it was an online questionnaire, there are sampling limitations as users in certain areas are less likely to have good Internet access to participate in the questionnaire. The limitation was addressed by making sure that there was a decent number of participants (at least 30) in each of the 10 most populated cities in China (Guangzhou, Shanghai, Beijing, Shenzhen, Chengdu, Dongguan, Hangzhou, Shantou, Wenzhou and Ningbo. Wenzhou and Ningbo were purposive, given that the Wenzhou train collision happened in the Zhejiang Provinceof which the two cities are a part of).
Once participants fulfilled the screening criteria, they were presented two articles, each giving a summary of the two cases (the articles provided information on the types of messages reported on both mainstream and social media). These articles functioned as an equalizer, to ensure that participants were on the same level in terms of their knowledge and memories of each event, given that there may be differences in the intensity and time lapse in the use of social media to seek information about both events. They were asked to spend time reading each article, and once they say yes that they have sufficient knowledge of what happened, they were asked to fill up a questionnaire.

Measures
Participants' perceived level of trust was assessed by asking to what extent they believe two statements published (total of four statements extracted from mainstream and social media) on a 4point scale. Each statement reflected a positive or negative valence for information reported about the event. The valence of each statement was first coded by one coder, and then two other student coders coded the same statements separately. They then came together to discuss disagreements and refined their coding definitions. There was 100% agreement for all statements except for statement 2 of the Mengniu Dairy Company scandal. Scott's Pi for this statement was 0.89. Table 1 shows the statements and their coded valence, source of reporting in parentheses (Information in parentheses is not shown to participants). After reading each statement participants were asked to what extent they believe the statements on a 4-point scale. In terms of propensity to trust, a 2-item measure was adapted from Mayer et al (1995).
Participants were asked to what extent do they believe the following statements: 'Others are highly trustable' and 'Generally, people have good intentions' and measured on a Likert scale (1=Strongly Disagree to 4=Strongly Agree). Cronbach alpha for the measure of propensity to trust using these two items in the sample is 0.78.
With regards to media exposure, survey respondents were asked to round up in 5-minute intervals how many minutes a day they spend consuming information on mainstream newspapers, radio, television and their websites, and using alternative platforms such as blogs and social networking sites. From a list of mainstream and alternative platforms, they were also asked to identify the primary sites they used to seek information in both cases. Socio-demographics, including age, gender, educational level, household income, and occupations were asked to conclude the questionnaire.

Findings
Participants indicated that they used Qzone as their primary social media site to seek information in both cases (89% of the participants). This is unsurprising given that Qzone is reportedly the fastest growing social media platforms in China. As of September 2012, it reported having a total of 597 million users (CNN, 2012), making it one of the strongest online communities in China.
To what extent do participants trust both positive and negative statements coming from mainstream and social media (RQ1)? This may be mediated by differences in media exposure. As participants who use social media more may show greater levels of confidence in information circulated from social media (and likewise for those who use mainstream media more), a crosstabulation analysis was done between media exposure and trust in each of the four statements to find out if differences in media exposure may be significant in driving these findings. No significant relationships were found, implying that media exposure was not significant in shaping the extent to which participants trusted information coming from either mainstream or social media for both events. Figure 1 shows responses to one of the statements asked about information circulated on mainstream media. Trends were similar for all statements for mainstream media for both eventsa majority of participants showed greater disbelief with positive statements 70% for the Wenzhou train collision and 65% for the Mengniu Dairy Company scandal). While it was expected that participants might show greater distrust in information reported on mainstream media in the Wenzhou train collision event, it was surprising that the same trend was observed also for the Mengniu Dairy Company scandal. Even though the article they were asked to read contained information about how the rumors were in fact spread by Mengniu itself, they remained skeptical and showed distrust in the information reported by mainstream media.

Figure 1: Trust / distrust in positive information about Wenzhou train collision
Could it be that the participants, who regarded themselves as heavy media users (the study screened for participants who regard themselves as regular users of both mainstream and social media), were naturally skeptical? Results show that this was not the case. Whilst they responded consistently to positive statements reflecting their distrust, they showed themselves to be more confident of negative information. Again, the way they responded to both statements about both events was similar; with a majority believing information reported online (64% for Wenzhou train collision and 56% for the Mengniu Dairy Company scandal). Figure 2 shows an example of the responses to one of the statements about the Mengniu scandal circulated on social media.

Certain dairy products by Yili contain deep sea fish oil ingredients which were mixed up with Eicosapentaenoic Acid (EPA) and would inhibit infants' growth. To what extent do you believe this information?
To address RQ1 fully, cross tabulations were run for all four statements using other variables such as age, gender, education. No significant differences exist for these variables, showing that socio-demographic differences did not have an influence on the extent of trust in these statements.
These findings show that confidence in positive or negative information coming from mainstream or social media was not different even though the usage of social media was distinctive. In the Wenzhou train collision event, the use of social media for collective action was so evident that it was picked up by international newswires around the world. The demonstration of ground-up power agencycould have led to greater trust in the negative information constructed collectively via social media, with the counter effect of eroding trust in positive information reported by mainstream media. In the Mengniu scandal, the situation is a little more complex. The initial action to disinform Chinese Internet users was not collective, but rather, staged (by Mengniu). But the plan was so successful that it resulted in massive boycotting of the products in question, and the eventual stock plunge of one of Mengniu's competitors. The knowledge of what transpired should have led to greater distrust in negative information; yet it was not the case as found in this study. Negative information is trusted more, or perhaps, when presented with conflicting information, users may actually prefer to place greater trust in negative information. The results show that the effect of information may not be discountedparticipants may still choose to believe certain information even after knowing that such information might be false.
To address RQ2, which is focused on investigating on whether or not confidence in either mainstream or social media is affected by a) propensity to trust (high versus low propensity), b) trust in positive versus negative information, and c) the interaction of trust propensity and trust in information, the data was analyzed using a 2 (propensity to trust) x 2 (trust in information) analysis of variance (ANOVA) for each media source. ANOVA was chosen as the method of analysis given the study's intention to examine possible significant differences between people who differed in their propensity to trust (high versus low), and trust in information (positive versus negative).
The analysis yielded no significant results for the dependent variable of perceived confidence in mainstream media in terms of the propensity to trust. This suggests that the propensity to trust had no effects on perceived confidence in mainstream media as a credible source for both events. However, the main effect for information was significant, F(1,495) = 89.25, p<0.001, indicating that participants were more confident in reports that contained negative information than in reports containing positive information. There was however, also a significant interaction effect between the two factors, F(1,495) = 4.98, p<0.05, suggesting that the propensity to trust and types of information contribute to eventual confidence in mainstream media. Results for confidence in social media as a dependent variable differed. The propensity to trust was a significant effect (F(1,495) = 65.55, p<0.05) on confidence in social media. This suggests that participants that had greater propensity to trust were also more confident in social media as a credible source for both events. The main effect for information was also significant, F(1,495) = 119.25, p<0.001, indicating that participants were more confident in reports that contained negative information than in reports containing positive information on perceived confidence in social media. A significant interaction effect was also found between the two factors, F(1,495) = 5.87, p<0.05, suggesting that the propensity to trust and types of information contribute to eventual confidence in social media.

Discussion
It is important to recognize the contextual attributes of each case so as not to isolate findings from the study. Issues of power negotiations and renegotiations, trust and confidence in media, and the types of information circulated were prevalent in both cases. Especially in the case of the Wenzhou train collision, democratic power was renegotiated via civic action by utilizing the facilities afforded by social media. In Giddens' (1984) terms this contributes towards a structure of domination, which is dependent on the 'mobilisation of [authoritative] and [allocative] resources' (p.31). It is shaped by, and shapes in turn the use of power in the interactions between people and facilities allocating resources. At the same time, the communications that were facilitated by social media enables the formation of interpretive schemes -'the modes of typification incorporated within actors' stock of knowledge, applied reflexively in the sustaining of communication' (Giddens, 1984, p. 29). Such interpretive schemes result in collective knowledge constructed by Chinese Internet users about the train collision (the structure known as signification by Giddens, 1984)and it is by drawing from such knowledge that collective actionrallying for deeper investigations, the resignation of the railways ministerwas made possible. Such collective knowledge forms the basis for common pool resourcesthe knowledge accumulated becomes resources used by social actors to mobilise people towards collective action (Tilly, 1978). Although it is not the intention of the paper to address how Chinese Internet users mobilise information as resources for the purpose of collective action, and whether it may shape democratic potential of the Internet in China, it is evident from this analysis that the issue of trust in media sources and the information they carry may mediate the process of resource mobilization.
In the other case, Mengniu was seen to assert its power over its competitors as well as everyday Internet users in China. The company made use of social media as facilities to dominate, and through the strategic employment of confederates (people who pretend to be pregnant women and young mothers) rumors were spread about their competitors. Through their interactions with other unsuspecting Chinese Internet users, they made it possible for the false knowledge to be constructed, shaping the eventual decisions and actions to stop buying the targeted products.
The effects of such signification may be long-lasting, as the study shows. At the time of the survey, two years have already transpired from the first time the rumors were spread. Yet majority of participants in the survey still showed uncertainty towards information intended to correct the wrong information spread, and continued to doubt the safety of local dairy products. In a follow-up conversation with one of the participants it was shared that many parents of young children in China choose to buy infant milk products overseas, and have them shipped back to China, rather than purchasing them locally because of their fear that the rumors might be true. While this affirms Siegrist and Cvetkovich's (2001) finding that people choose to trust negative information more than positive or neutral information, it may also show growing distrust of the credibility of mainstream media, as reflected by the majority of participants who expressed greater disbelief in information reported on mainstream media.
Although Guo et al (2005) have demonstrated that the Chinese place greater trust in local media than foreign media, the study reported here has shown that local social media platforms, even with the sanctions and moderations put in place may mediate the relationship between citizens and state-controlled mainstream media. It will, however, be a mistake to discuss the utilization of social media with a celebratory tone. Indeed, as the Mengniu Dairy Company scandal has proven, the democratic potential of social media may be undermined when certain companies choose to exploit the facilities and culture afforded by social media. Although social media somewhat 'flattens' the world that exists for Chinese Internet users, mainstream media, and the state, their relations are not completely horizontal with some dominating over others. In the case of the train collision in Wenzhou, Chinese news anchors also made use of mainstream media to criticize the authorities until they were forbidden from doing so by the authorities, and such actions were praised by Chinese Internet users.
The individual's propensity to trust was not found to be significant for perceptions of credibility of mainstream media, but was significant for social media. This is perhaps due to the fact that mainstream media is already widely penetrated across households in China. For social media however, the propensity to trust was positively associated with perceptions of social media credibility. This suggests that those with greater propensity to trust were the ones who would place greater trust in the credibility of social media, compared to those with lower propensity to trust. They may also be the ones who are more ready to trust information on social media, and may therefore be dominated by Mengniu's '731 plan'. Some implications for media literacy may be relevant here, as the results show that media literacy especially on social media platforms may also be mediated by individual propensity to trust, which is a personality trait and may not be easily intervened or changed.
Findings concerning the effect of information valence on the perceived credibility of both mainstream and social media are significant. Practically, the implication is that messages containing risks and negative information may be perceived as more credible and trusted than those that are positive or contain little negative information (note that participants were not conscious of the media platforms when they responded to different types of statements, so the media effect is controlled for within the questionnaire). Other studies have found that people may trust certain sources, for example, experts, more for the same information (Frewer et al, 1996). This study however, has shown that information itself can shape the extent by which individuals have trust and confidence in the credibility of both mainstream and social media.
These findings have substantial implications in understanding how civic participation works in China. First, biases towards negative information may shape eventual actions and decisions by citizens, as seen in both cases discussed here. It does not help that more negative information is found on social media, with the norm of media sanctions and directives being issued by the state. The study points to the possible trajectory that, Chinese Internet users, even if they also use mainstream media heavily (as it was with participants in this study), can be looking to social media as a more credible and trustable media source especially in events of great uncertainty. Even with the Chinese government's media interventions, the two cases have also demonstrated that civic participation can still be pursued, although it is not without boundaries.

Conclusion
The circumstances and evolution of the Internet's usage in China challenges the idea that the Internet is all-enabling and promotes the development of democratic civil society. As shown in the paper, new media may be used to pursue civic actions, but it is not without boundaries and requires the negotiation and renegotiation of power relations between state and non-state actors. In some cases, non-state actors can also use the platform to encourage civic participation for their own means and purpose to dominate the competition.
Using two controversial cases where social and mainstream media were respectively utilized by Chinese citizens and authorities or business organizations to pursue their desired agenda, the study has shown the effects of information and trust propensity on perceived confidence in mainstream and social media platforms within China. There are implications for media literacy and understanding how civic participation works in China. Additionally, it also points to the challenges of managing social media for the state. The role of the state in coming up with timely interventions and release of information in the case of the Mengniu Dairy Company scandal is significant; whereas the state is placed under surveillance in the Wenzhou train collision event with Chinese Internet users rallying together to call for greater accountability and transparency. The survey that was conducted based on participants' information seeking about the two cases revealed that participants place greater trust in social media, although such trust is mediated by trust propensity and types of information. Such trust, if they continue to grow, will see greater mobilization of civic action via social media. This is of course, also dependent on the effect of information.
There are some limitations of the study. Due to the fact that approximately a year lapsed between the two events (the train collision happened around a year after the Mengniu Dairy Company scandal), there may be a 'recency' effect as participants may recall the Wenzhou train collision more effectively and therefore responding to the questionnaire with such dominant memories. On the same note, the survey was conducted only in August 2012, about a year after the train collision happened. Although there was still talk about the event on various social media platforms, there may be memories of other events which might intervene with the way participants responded to the questions. As such, the survey began by presenting information about both events with the intention to overcome these limitations. However, it is still important to acknowledge that these factors might present a limitation to the findings.
In addition, participants came from 15 cities in China, and such diversity may present constraints in the representation of the sampling frame. It was not one of the sampling criteria, although the question was asked about where participants were residing at the point when they participated in the survey. The decision not to locate participants in a particular city came from the realization that even for a 'local' event like the train collision which happened in Wenzhou, the issue was 'national'. Citizens in China took up the issue as a social and political cause, questioning the efficiency and governance of the authorities, and whether or not the well-being of their fellow countrymen was being considered. As for the Mengniu Dairy Company scandal, it was triggered solely via social media which was 'national', and various state media responded thereafter. Nevertheless, the possibility that participants that reside in the vicinity of Wenzhou may respond with greater intensity than participants from other cities should be acknowledged. Future work is informed by a case-based program of research, acknowledging the social factors involved in mobilizing social media for civic action. Future studies will also involve other antecedents of trust and confidence in social media, such as beliefs, salience of information, and influence from opinion leaders.