As the discussion surrounding misinformation and polarization takes center stage, one space that has been largely overlooked is the Chinese media sphere within the United States, where questionable content has garnered a loyal following among Chinese-speaking immigrants. While headlines with false claims such as “Illegal immigrant started wildfire in Sonoma County” mirror types of misinformation found in American, English-language media, what’s especially worrisome in information ecosystems like these is their central influence on the first-generation immigrant experience and integration with US society. Yet, they exist very much as unknown, parallel universes for most researchers, journalists, and media watchers. With the rise of conservatism among first-generation Chinese immigrants, one of the fastest growing populations in the United States, this media space—led by mobile platform WeChat—offers key clues to how political information and misinformation are constructed for and distributed among the emerging political constituency.
This report sheds light on the nature of misinformation and political polarization in the WeChat news sphere. The conclusions drawn from this preliminary investigation into WeChat mirror more general challenges with misinformation and polarization in the digital news ecosystem, but also highlight the specificity of the information problem for the immigrant population.
WeChat has a highly specific issue agenda in its coverage. Compared to English-speaking media and immigrant Chinese media, affirmative action/census data disaggregation and unauthorized immigration are the disproportionate focus of WeChat content, while jobs, the economy, and healthcare received little attention during the period of study. This divergence in issue priorities—without counteracting narratives—leaves room for misinformation to fester around dominant topics on WeChat. Political discourse on WeChat can be described as asymmetrically polarized, with the right leading in volume, reach, and skewed issue agenda. Both conservative and liberal discourses on WeChat centered around race relations and the ambivalent role of Chinese Americans. But it was the conservative narrative, invoking zero-sum calculations and the neglect of Chinese Americans by liberals, that seems to resonate more widely.
Low barrier to entry on WeChat has generated a profusion of content publishers native to the platform and intense competition for attention. The abundance of revenue-driven content published, coupled with partisan forces, makes WeChat especially vulnerable to political misinformation. Emotionally stirring, sensational stories become amplified through the replication and embellishment of a long tail of WeChat outlets, which creates repetition and familiarity.
Misinformation inside WeChat takes on a pronounced local expression. Local news and information outlets contribute heavily to the spread of misinformation. Local stories and policies especially fall prey to distorted information, when lack of local news coverage on issues of particular interest to immigrant Chinese leaves a vacuum for misinformation to flourish.
The information problem on WeChat intersects with and draws rhetorical and ideological resources from both US and China-based platforms and online publics. While misinformation can be relayed from the far-right in the US, it can also be strengthened by discourses and online communities on Chinese equivalents of Reddit and Quora.
Chat groups, especially large ones where members are acquaintances or complete strangers connected by a common affiliation, are abundant and central in information dissemination. At a time when mobile messaging apps are increasingly utilized for news, WeChat highlights the significance of socially driven, organically spread information mostly unaffected by algorithms and technological manipulation.
WeChat’s information problem highlights the challenge of immigrant political integration, where difference in values and interests and a distinct communication system compound the distribution and consumption of misinformation. More than technological solutions, combating misinformation for immigrant audiences requires engaging with and ultimately bridging their experiences and perspectives.
“George Soros backed the violent clash in Charlottesville.”
“Illegal immigrant started wildfire in Sonoma County.”
“You could be receiving HIV positive blood in California now!”
Headlines such as these, some conspiratorial and emotionally charged and other verifiably false, have become a rampant and well-recognized feature of today’s information environment. But instead of the usual habitats for misinformation such as Facebook and Twitter, these examples are translations of articles found on WeChat, the ascendant mobile messaging and content publishing app among immigrant Chinese in the United States. These headlines do not represent the prevailing discourse within Chinese-language news, but there is no doubt that the divisive and strident political messages in the past election cycle have found their way into the immigrant Chinese news sphere and gained an independent existence. Their presence begs the question of what kind of misinformation exists in this space, as well as how it is generated and distributed.
Ethnic media—media produced by and for an ethnic group—plays an especially pivotal role for immigrant acculturation and integration.12 The cumulative evidence from the research literature suggests that ethnic media is crucial in introducing immigrants to social norms and political knowledge in their host society, while also helping immigrants to maintain connections with, and consolidate, their ethnic identity and community.34. In multiethnic America, overlooking non-English-speaking media in the discussion of misinformation means missing an opportunity to understand how political discourse is constructed for a significant segment of the political constituents, and to build media narratives that are sensitive and inclusive.
The added impetus for studying ethnic Chinese media stems from a rise in conservatism among first-generation Chinese. Although no quantitative assessment exists to date, activism by conservative Chinese Americans has been ardent and vocal, and as a curious phenomenon has already garnered the attention of mainstream media.56 To be sure, Asian Americans still overwhelmingly lean liberal Democrat.7 And I do not wish to suggest that misinformation is inherently a phenomenon on the right. Instead, this report underscores the rise of conservatism as a significant shift in Chinese-American politics that suggests heightened polarization. Polarization and misinformation manifest as two entangled dimensions: misinformation can drive polarization, and polarization creates ripe opportunities for the generation and spread of misinformation.
This study focused on WeChat because it is a distinctive infrastructure for content production and dissemination. With its massive reach, it likely shapes the landscape of political information for immigrant Chinese. What follows is a preliminary assessment of this Chinese-language platform and manifestations of its information problem. The conclusions drawn mirror general challenges with misinformation and polarization in the digital news ecosystem, but also highlight the specificity of the information problem for the immigrant population.
This report draws on a survey conducted in August 2017 with 407 US-based Chinese WeChat users to understand how they encounter political information within WeChat. The survey helped to identify twenty-five WeChat outlets that its respondents deemed important to their understanding of current affairs and politics in the United States. Survey participants were recruited from Chinese organizations and Chinese-language school mailing lists, as well as within WeChat through snowball sampling. Twenty-three follow-up interviews were conducted to further examine how users interact with WeChat outlets and participate in chat groups.
The findings are also informed by a content analysis of twenty-five WeChat news outlets that emerged from our user survey as top venues for social and political news about the United States. The analysis includes all published content in these WeChat outlets—3,837 articles in total—for the period of September to November 2017. (See Appendix B for a full list of these outlets.) Lack of API and web archives limited the ability to analyze a larger time frame. To serve as a comparison, front-page articles on three long-standing Chinese media outlets in the US (World Journal, Singtao Daily, and China Press) were also included in the analysis.
While many of these WeChat outlets provide a mix of gossip, information, and news, a few specialize in politics and have come to occupy opposite ends of the political spectrum. These included three right-leaning news accounts and four left-leaning accounts, which will be introduced in more detail later. I focused on these outlets and analyzed their content between January and November 2017.
Using computer-assisted coding, the articles were analyzed for their coverage of key social and political issues, as well as connections between these issues, political parties, and groups; racial and identity-based groups; and terms denoting values. Care was taken to include code words and terminologies peculiar to the Chinese-language political discourse. Some examples include shabaitian (“sweet and dumb,” which can be roughly equated with libtards) and hepingjiao (“religion of peace,” a derogatory phrase for Islam). (For a list of terms, see Appendix A.) Using Gephi, semantic networks were constructed to demonstrate the co-occurrence of these terms.
In addition, two research assistants and I flagged and tracked questionable content pertaining to US social and political issues on WeChat from September to November 2017, as an exercise in teasing out overarching features of misinformation. The idea of misinformation is fraught with definitional issues and troubled by ideological interpretation.8 Because our approach was not meant to be exhaustive, we worked with a loose definition of misinformation that includes false, highly biased, and hyperbolic claims. With stories that got flagged, we used a specialized external search engine (Sogou) to search for similar occurrences on WeChat and mapped their trajectories within and outside WeChat.
An Overview of WeChat
With 889 million monthly active users,9 WeChat is the social media of choice in mainland China, governing aspects of life from social networking and messaging to takeout orders and personal finance. Offering features reminiscent of WhatsApp, Facebook, and Twitter all at once, it combines the intimacy of mobile messaging and small-group interactions with the capacity for viral dissemination. The significance of WeChat for the present analysis lies in its infrastructure for the generation and distribution of native content.
In the last official count in April 2017, there were as many as ten million official accounts (OA), or what I refer to in this report as WeChat outlets.10 They publish a staggering amount of content for users, and some have come to rival established media in influence.11 From individual bloggers and citizen journalists to brands and media companies, many have leveraged the low barrier to entry and vastness of the user base to create content, making WeChat an increasingly central source of news for Chinese-language users. Unlike platforms for social news like Facebook and Twitter, WeChat operates as a more enclosed and private ecosystem. As the report will discuss, information sharing and discovery takes place within networks of friends and acquaintances, with minimal influence from hashtags, trending news, and other forms of technological manipulation. Even interactions between users and WeChat outlets are mostly private—commenting on an article is not always enabled, and outlet managers determine which comments get displayed publicly. These design features are central in defining the generation and dissemination of information and misinformation on WeChat.
The WeChat news outlets selected for analysis in my research all focus on current affairs in the United States. They vary greatly in terms of scale of operation and quality. On one end, there are individual or boutique operations, with one writer or a small base of individual contributors delivering a mix of news, analysis, and opinions. Larger operations have more substantial, full-time editorial teams that consistently churn out content. These could be overseas outposts of established China-based media (e.g., Insight China, a subsidiary of Global Times) or WeChat startups, the most notable example of which is College Daily, a formidable player backed by angel funding.12 Six local news and information outlets topped user nominations as influential sources of news, with geographical focus on cities with major concentrations of immigrant Chinese such as New York, Houston, Atlanta, Seattle, and cities in California. I highlight these locally oriented outlets here because they make a special reappearance later in this report as a type of content publisher aiding the propagation of misinformation.
Issue Salience: Core Concerns and Divergence
Out of the many issues defining the contours of political discourse in the United States, twelve were examined for their relative prominence in the WeChat media ecosystem. Muslims/Islam, terrorism, and affirmative action/census data disaggregation topped the focus of WeChat content, followed by jobs/the economy and undocumented immigration (Figure 1).
Of course, issue agenda could be subject to changes in the news cycle. To put this in perspective, the same search terms were used to query a collection of English-language media in Media Cloud,13 including mainstream media, regional media, online news, digital natives, and political blogs, for the same period (September–November 2017). Leading the English-speaking media agenda were jobs and the economy, healthcare, Muslim/Islam, terrorism, and climate change (Figure 2). It should be noted that the English-speaking media agenda here serves as a benchmark for the Chinese counterpart and should not be interpreted on its own, given that results were subject to the specific time period analyzed and search terms used. For example, “immigration reform” was not included as a search term to ensure focus on undocumented immigration and exclude topics such as H1B visas, which likely explains the relatively scant coverage of undocumented immigration in English-speaking media.
The divergence in issue priorities between English- and Chinese-language media is stark. Without delving into how these issues are framed, the salience of an issue agenda in the media often signals to audiences what to think about.14 Particularly striking was the discrepancy between the two media spheres’ attention on affirmative action and census data disaggregation. This was the third most covered topic in WeChat with coverage almost equal to the two top issues (Muslim/Islam and terrorism). Meanwhile, affirmative action and census data disaggregation hardly registered a blip on the English-speaking media radar. Searching for census disaggregation as a topic by itself returned zero results in this time period.
At the intersection of education and race relations, affirmative action and census disaggregation have become two signature issues that first-generation Chinese rally around and debate. Census disaggregation refers to bills introduced in states like California, Massachusetts, and Rhode Island, which proposed to distinguish different Asian American Pacific Island (AAPI) sub-groups in demographic data collection. Given the vast linguistic and economic differences among the AAPI population, doing so helps expose disparities that have long existed within this group. For opponents of the bill, this paves the way for affirmative action to further disadvantage Chinese Americans—who have higher education attainment compared to some other Asian groups—in college admissions. Complexities of these issues aside, the fixation on affirmative action and census disaggregation on WeChat is conspicuous. The striking invisibility of these issues in English-speaking media signals a disconnect in mainstream coverage.
Unauthorized immigration was also a more prominent issue on WeChat, compared to how issues lined up in English-language media. Although some content touched on the DACA debate, it was sanctuary laws, appearing in seventy percent of the WeChat content on unauthorized immigration, that anchored the discussion of this issue. The salience of sanctuary laws in the WeChat media agenda could also be explained by its application in California, where a large concentration of first-generation Chinese resides, and the sensational stories that produced the trope of California as a liberal haven for criminals. I dissect this point with a concrete case on misinformation later in this report.
It is also worth noting that WeChat’s issue agenda does not align with or represent that of immigrant Chinese media overall. In Chinese ethnic press coverage of jobs, the economy, and healthcare far outpaced other topics (Figure 3). Arguably, operating as traditional newsrooms, ethnic press generated a more conventional news agenda. WeChat, separated from conventional media practices and structures, put forth content more divergent from the coverage by legacy ethnic press and English-language media. The specificity of WeChat’s content may contribute to a gap in issue agenda between immigrant Chinese and the wider public. When coverage of an issue is dominant on WeChat but not in Chinese ethnic media or English-language media, such as in the case of affirmative action and undocumented immigration, this also leaves room for misinformation to fester without counteracting narratives.
The presence of partisan outlets on WeChat also played a role in constructing these issues and bolstering their salience. In the next section, I discuss political polarization on WeChat to set the scene for the dynamics of misinformation on the platform. Although partisan WeChat outlets only constitute one part of the platform and cater to more politically active audiences, their reach is significant and the political discourses constructed by these outlets intersect with and have resonances in the WeChat ecosystem more broadly.
Political Polarization on WeChat
For the politically initiated on WeChat, the phenomenon of polarization is difficult to miss and manifests rather radically through a blend of punditry and provocation. Seven WeChat outlets were identified by the survey respondents as vehicles of partisan politics. This report refers to them as right-leaning and left-leaning outlets. These labels broadly mirror the political divide in US politics, albeit with some distinctions in emphasis in immigrant Chinese politics, as we will see. The conservative WeChat sphere is led by two associated accounts, Voice of North American Chinese (VNAC) and Civil Rights, as well as a similar but distinct account, confusingly named Voice of Chinese Americans (VCA). In the months leading up to the 2016 election, these accounts quickly solidified readership among pro-Trump Chinese.
The left-leaning accounts include yet another confusingly named outlet, Chinese Americans. These names in fact testify to one of the defining characteristics of partisan politics for immigrant Chinese: activism. These outlets formed in response to several key events in Chinese activism, including rallies against talk show host Jimmy Kimmel’s China joke15, mobilizations opposing SCA-5, California’s bill to reinstate affirmative action, and protests against the indictment of Peter Liang, a Chinese-American police officer charged with the shooting of an African-American man. Before they became content-producing outlets, they were WeChat networks calling for action, which convened a base of supporters ready to be engaged as readers. The naming of these accounts and their connection with activism in response to discrimination reflects the primacy of the Chinese-American identity in defining their politics. And these politics are very contentious, as reactions to discrimination have engendered both conservative and progressive visions.
The other three outlets on the left—iAmElection, Anti-Rumor, and NoMelonGroup—were started by a collective of writers who intensified their efforts in response to the popularity of the right-leaning discourses on WeChat. Content generated by these partisan outlets and my personal exchange with their editors show that the two sides are keenly conscious of each other and engage in an ongoing rhetorical battle.
Looking at these outlets, WeChat may be described as asymmetrically polarized, with conservative content leading the scoreboard in volume and reach, as well as narrower and more aggressive in its ideological expression. On average, the right-leaning outlets on WeChat churned out 384 articles per month, drawing an average of 6,060 views per article in the period analyzed. In comparison, left-leaning outlets generated less content (eighty-seven articles per month), with even the most-read account trailing VNAC by a few thousand (Table 1).
Issue agenda was highly skewed on the right (Figure 4). The scope of issues clustered around Islam and affirmative action/census disaggregation, with the two accounting for about forty-four percent of its content. Negligible attention was paid to gun control, climate change, and reproductive rights. As an issue that has motivated Asian-American Republican voters in the past,16 jobs and the economy only made up six percent of the content on the right. Healthcare garnered even less attention. This is not to suggest that the economy and healthcare do not constitute the ideological expression of Chinese conservatism, but rather demonstrates the singularity of content focus by right-leaning outlets.
In contrast, left-leaning outlets had a more even distribution of topics. The economy and healthcare were covered on par with the other top issues. Climate change and reproductive rights also received substantial attention. To illustrate further, the ten most viewed stories from these outlets spanned taxation, education policy, healthcare, the travel ban, and the Women’s March. The headlines in Table 2 and Table 3 provide a taste of the style and perspective in these partisan outlets.
*WeChat displays view count larger than 100,000 as 100,000+
In particular, the right on WeChat devoted almost one-fourth of its content to covering Islam. Not surprisingly, this is a topic plagued by different genres of misinformation. For example, an article with 48,664 views was headlined, “The Muslim takeover of America. Part 1.” It opened by describing a congressman from Minnesota swearing in on the Quran, and went on to caution against the rising number of Muslim voters and Muslims running for office. Obvious prejudice aside, this article exemplifies the complex choreography of misinformation. The photo of the congressman swearing in was accompanied by a link from Snopes, which rated the claim as mostly true, but clarified that religious texts are not actually used for swearing in per se; and in the case of the Minnesota congressman, he posed for the photo with a copy of the Quran (owned by Thomas Jefferson) after the ceremony. Quoting from a fact-checking site did not preclude ideologically motivated use of information. In the entire sample of stories published by partisan WeChat outlets, there were only several instances of direct sourcing from English-speaking alt-right media, such as The Daily Caller and Zero Hedge, but the generation of misinformation did not hinge on referencing biased and questionable sources.
While polarizing rhetoric on both sides revolved around discrediting the other side, the right was more concerted in its delegitimization of liberal Democrats. Mentions of liberals dominated right-leaning outlets while conservatives and liberals were more evenly mentioned on left-leaning outlets. As many as nine percent of the content generated by right-leaning outlets made references to liberals using the code words huazuo (“Chinese left”), baizuo (“White left”), and shabaitian (“sweet and dumb”). These terms, usually used in a derogatory way, connote an understanding of liberals as hypocritical and superficial in their obsession with equality, multiculturalism, and political correctness.17 This narrative has familiar streaks of right-wing populism, echoing terms such as “social justice warriors” and “libtard,” but contains an added element of perceived double-standard, where the liberal version of social justice applies to a host of minorities—African Americans, Latinos, Muslims, LGBTQ, undocumented immigrants, and felons—while Chinese Americans are either neglected or sacrificed. The strong co-occurrence of “liberals” and “discrimination,” much stronger in the conservative WeChat sphere than in left-leaning outlets (Figure 5), is one illustration of this baizuo narrative.
Figure 5: The size of the nodes is proportional to the frequency that a term is mentioned, and the thickness of the edges indicates the frequency of co-occurrence between two terms.
Race relations is key to the partisan political discourse on both sides. Thirty-two percent of the content on the left and twenty-four percent of the content on the right contained references to more than one of the five racial or ethnic groups coded in this analysis (White, African American, Latino, Muslim, Chinese/Asian), with the right being particularly concerned with the relationship between Muslim and Chinese (Figure 5). The centrality of race relations in WeChat’s partisan political discourse has a lot to do with the origin of these outlets and the progression of Chinese activism that they are part of, which has been defined by a strong response to perceived discrimination and injustice.
On the right, this resolved into a social Darwinist, zero-sum conception of racial politics, where Chinese empowerment ultimately means getting what they have always deserved but have been given to other minority groups unfairly. This argument has been applied to issues such as undocumented immigration (Chinese immigrants migrated legally), affirmative action (merit-based admissions), and law and order (hardworking taxpayers being terrorized by criminals).
To be clear, this narrative is not the only understanding of how to situate Chinese Americans in the matrix of race, power, and equality. The left on WeChat has offered interpretations of affirmative action and census disaggregation that acknowledge the potential for discrimination without diminishing its importance for all minority groups, revisited the significance of the civil rights movement for Chinese Americans, and encouraged a form of political participation that emphasizes shared interest with other minority groups. I found support for both visions among my interviewees. Of course, the extent of this conservative turn among first-generation Chinese remains to be assessed on a larger scale, but it is clear that right-leaning outlets are extremely visible on WeChat, and their politics do find resonance. With this visibility comes the potential for political socialization. Several interviewees suggested that they were introduced to US politics by the right-leaning outlets in the months leading up to the 2016 election, using terms such as “enlightening” and “educational” to describe their encounter with these outlets.
It is beyond the scope of this report to unpack the complex politics emerging among first-generation Chinese (although it would be a very worthwhile enterprise). The foregoing analysis is intended to highlight two features of political polarization on WeChat: first, right-wing discourse on WeChat is both forceful and extreme, creating a highly singular set of ideologies with no moderate voices or alternative perspectives; second, both sides are parsing through the role of Chinese Americans in the racialized political landscape, with the version on the right tapping into a zero-sum conception. With its visibility, the right could exert palpable influence on first-generation Chinese politics. In fact, as I discuss in the next section, right-leaning outlets do play a role in the presence of misinformation on WeChat.
Two Case Studies and Three Takeaways
Unsurprisingly, efforts to flag and track misleading WeChat content uncovered various types of misinformation in contentious issue areas, including affirmative action, undocumented immigration, Muslim/Islam, and law enforcement and public safety, many of which are linked to right-leaning outlets on WeChat. For example, a story by VNAC stoked fear by attributing the devastating wildfire in Sonoma County to an intentional act by an undocumented immigrant. However, rather than detailing each, it would be most instructive to discuss two widely circulated stories that help illustrate what I consider archetypal elements of misinformation on WeChat. This is not meant to be an exhaustive description of misinformation on WeChat, but an exercise in unpacking the dynamics of its origins and manifestations. I lay out the main building blocks of the two case studies before summarizing a few key takeaways.
Antifa-Led Civil War
In early November 2017, a story warning readers of mass riots and Antifa-led civil war appeared in different corners of WeChat. In the English-language media world, the story’s rise featured a familiar cast of characters: subculture communities, satirists who were taken as truths, and conspiracy-inclined far-right outlets such as Infowars and the Gateway Pundit. 18 Headlines of these stories used different variants of “Antifa plans ‘Civil War’ to overthrow the government” and “Mass riots may happen across cities in the US.”
VNAC, one of the conservative outlets examined earlier, was the first to publish the story, and its original article was viewed more than 100,000 times (again, this is when WeChat stops counting). Here, connection with the English-language alt-right sphere was evident. The Antifa story appeared as part of VNAC’s continued engagement with the “truth” of Charlottesville and liberal media’s one-sided demonization of the alt-right. It also herded conspiracy theories popular with the alt-right that suggested George Soros funded Antifa and the Black Lives Matter movement.
Subsequent circulation of this story on WeChat has been mostly stripped of far-right ideology. It was aided by a process in which sensational information can catch on through multiplication and piracy by decentralized outlets. On WeChat, we logged thirty different WeChat outlets that published this information, which together accumulated at least 472,625 views. Twenty-two outlets either directly pirated the content or made minor modifications for their specific audiences. The content was almost identical across the thirty outlets: a tweet by Trump USA (@MADE_USA), a list of cities where mass riots would take place, and a description of Antifa and Black Lives Matters as organizations that practice violence (Photo 1).
Notably, seventeen of the thirty outlets that shared this story were locally oriented, ranging from all-purpose local news accounts to those narrowly focusing on housing or classifieds. These are by no means inconsequential players. As an example of their influence, the story on the account Chinese in LA accrued a view count of 100,000, on par with VNAC. While many profit-driven WeChat outlets rely on advertising for revenue, these local news accounts have an especially ostensible display of banner ads, ranging from realtors, immigration attorneys, Chinese schools, and other local Chinese businesses. (Figure 6)
This case illustrates how a conspiratorial idea that was relayed all the way from the English-language far-right by way of partisan outlets on WeChat could be amplified through the replication and multiplication of diffuse outlets eager for shareable content. At a time when political news could easily become fodder for drama and entertainment, outlets that are not overtly political can tag on and become conveyors of political misinformation. Local outlets, in particular, contributed heavily to this amplification process.
The Curious Case of Haissam Massalkhy
If the Antifa story was relatively straightforward, another case took a more meandering trajectory. It revolved around an alleged DUI case, in which a Lebanese motorist fatally struck a Chinese jogger in Walnut, a city west of Los Angeles. What attracted eyeballs was the claim—often embedded in the headline—that the motorist, Haissam Massalkhy, was undocumented and had intentionally committed the act to extend his stay in the US. This story, discussed in the context of California’s sanctuary state status, triggered intense reactions and has mutated into claims such as “sanctuary state protects felons” and “crime becomes a pathway to green card in California.” Here too, as many as forty-three different WeChat outlets, of which thirty-four were local information outlets, published articles that named Massalkhy’s alleged motive, adding up to 319,581 views.
The strength of the narrative partly derives from its emotional appeal to the absurdity of liberal excess. It has become one of the stories exemplifying the unthinkable ramifications of sanctuary state, along with the Kate Steinle verdict, which has received national attention. California’s other proposed legislations served as part of the extended Exhibit A. These included a recently passed bill reducing penalty for knowingly exposing others to HIV, as well as a new law scaling back lifetime registration for some sex offenders, both of which were subject to distorted coverage. California, in this narrative, becomes a “sunken place” where liberal values have run amok, disregarding the safety and interest of law-abiding citizens to protect illegals (Photo 2). The Massalkhy case, with its unfathomable ludicrousness, was the ultimate proof.
Yet, substantive misinterpretation of the sanctuary state bill aside, basic facts about this story remain elusive. Tracing its origin proved to be a cumbersome task. It was a local story that received very little coverage in English-language media. One source suggested that Massalkhy’s green card, sponsored by his ex-wife, may be set to expire at the time of the crash.19 Some reported that the police initially believed the act was intentional.20 But no English-speaking source reported the alleged motive of the defendant. World Journal, a Chinese ethnic press with local reporters in Southern California, was the single source that covered the story, reporting that Massalkhy acted to extend his stay the US.21 Short of accessing court documents, it is impossible to resolve this discrepancy between World Journal’s reporting and news coverage in English. It took peeling away layers of repackaged content to find the original source of the story, and even then the facts are beyond recovery.
In contrast to the relative obscurity of this story in English-language news, the story appeared across Chinese-language media. In addition to China-based media and Chinese ethnic presses in the US, all major message boards, including Wenxuecity, Mitbbs, and Creaders.net, also picked up the story. These are formidable players in the Chinese-language information ecosystem. Wenxuecity, for example, boasts close to three million unique visitors in a thirty-day period, with 60 percent of total traffic originating from within the United States.22 On Zhihu, a Quora-like knowledge-sharing site, the topic was viewed 987,045 times. Known for the caliber of its members, Zhihu is commonly perceived as a high-end knowledge-sharing community, where the most upvoted answers can stand in for authoritative explanations. More recently, it has become a gathering place for Trump enthusiasts.23 On the Massalkhy case, the top answer piled onto the case even more untruths and bias. It related “yet another case” in which a Muslim (which Massalkhy may not have been) purportedly was given a lenient trial after hitting a Chinese girl in the “deep blue state” of New York. In the end, attempts at sensemaking would likely circle back to the same narrative, possibly with amplified outrage.
Decentralization and Content Replication
These two case studies first of all demonstrate the logic of attention economy and the decentralization of content generation that more broadly define the digital information environment of today.2425 Here we see familiar incentives and tactics at work, where intense competition among WeChat publishers creates an ecosystem that rewards speed and sensationalism, contributing to the prevalence of question headlines,26 emotional hyperbole, and rapid replication of content. A report by WeChat’s research team released in April 2017 showed the crowded marketplace of content publishing within WeChat. Of the staggering ten million WeChat outlets, 76.1 percent have fewer than 10,000 subscribers.27 In this hypercompetitive environment, not jumping on a captivating story such as mass riots rocking US cities means not getting a share of the attention pie. While better-resourced WeChat outlets can generate what the platform labels “original content,” others rely on tweaking headlines and editorializing to make the story stand out among replicas.
The case studies uncovered a long tail of outlets that wield minimal influence by themselves, but together their abundance gives misinformation ample opportunity to multiply, distorting or masking the original source. As Jonathan Stray explains, quantity does matter, as “receiving a message via multiple modes and from multiple sources increases the message’s perceived credibility, especially if a disseminating source is one with which an audience member identifies.”28 Combined with the abundance of weak-tie chat groups, which I examine in more detail in the next section, decentralization and replication of content production on WeChat creates the conditions for repeated exposure, familiarity, and “impressions” that may work cumulatively and subconsciously to shape attitudes.29
Efforts to debunk these stories did exist but did not do nearly as well. In both these cases, NoMelonGroup, introduced in the previous section, published debunking articles. Counting reposts by partner outlets, its article on Antifa-led civil war was viewed about 11,000 times, and its article clarifying California’s sanctuary laws in reference to the Massalkhy case was viewed a little over 4,000 times. Counter-narratives, already having to swim upstream,3031 do not benefit from the same long tail of outlets latching onto provocative content.
It also needs to be mentioned that the ease of self-publishing has generated excellent research and analysis on social and political issues which serve to connect the immigrant audiences with social and political discussions in the U.S. But the cumulative influence of cheap and fast information can be formidable.
Localization of Misinformation
Local information outlets tapping into overseas Chinese audiences for revenue heavily contributed to the spread of misinformation as they jumped onto the bandwagon sometimes steered by partisan outlets. The same competition among WeChat publishers dictates the work of startups focusing on North-American users. For example, for Los Angeles alone, there are four different WeChat outlets vying for local audience. It goes without saying that there are financial incentives for cheaply creating locally relevant but emotionally resonant messages for the immigrant Chinese audience. But here the overt bias and politicization displayed by many of these supposedly non-political local outlets could suggest an assumption on their part about the beliefs and values of their imagined audience.
These outlets convey some very useful information, including events, business and consumer information, crime, and political information. Social ties and information are especially crucial for immigrants as they navigate the new places, customs, institutions, and social processes that define their new reality.32 The practicality of local information outlets could create a strong dependency relationship with immigrant Chinese, especially when they rush in to fill a gap in local news. There is already some initial evidence of consolidation of power. One of the bigger operations is a network of twenty local outlets, which could adapt and cross-post the same content to a dispersed audience.
Misinformation also takes on a geographical expression as Chinese-concentrated areas become the focus of news coverage. As cities and suburbs33 become immigration destinations, they also create media markets for locally-oriented WeChat outlets. Especially on issues such as sanctuary cities and census disaggregation, local policies have become focal points of coverage and action.
Relatedly, what the Massalkhy story in particular throws into relief is the lack of credible local reporting—lack of local reporting, period3435—that could serve as a countervailing force to dominant narratives. While the Antifa story could be easily contradicted by mainstream English-language media coverage, the construction of the Massalkhy case was cemented by Chinese-language reporting, with insufficient clues to confirm or challenge basic facts in the narrative. Many have cautioned that the decline in local news weakens civic infrastructures for communities and could leave a vacuum for misinformation to flourish.3637 This study provides a stark illustration of this possibility.
A Different Set of Players and Discourses
While the Antifa story dovetails with ideas and players the English-speaking media world is acquainted with, not all Chinese-language misinformation has a familiar ring. The Massalkhy case is an example of a distorted story that was previously unknown to non-Chinese audiences. WeChat is situated in a transnational media space that intersects with both English-speaking and Chinese-language players in the information ecosystem. Like the hybrid place many immigrants occupy, political information in immigrant Chinese media also sits at the confluence of two public spheres and draws rhetorical and ideological resources from both. In addition to players and discourses originating from the US, the larger Chinese online public supplies its own subcultures, influencers, and ideologies. The response to the Massalkhy case on Zhihu reflects this intersection.
As an additional example, the conservative WeChat sphere has been able to leverage and import Islamophobia from Chinese internet to heighten the sense of fear and panic. Several alarmist articles published by VNAC and Civil Rights cited the banning of non-Halal food on Chinese flights and university campuses in China, and the construction of lavish Mosques in poverty-stricken areas in Western China, where concentrated Chinese Muslim population reside. In China, Islamophobia has been finding a sounding board in Chinese versions of Twitter, Quora, and Reddit.38 These discursive agents and dogmas formed around them—relatively unfamiliar to researchers and media practitioners—also contribute to the circulation of misinformation and shape its meanings as well as resonances. More systematic efforts are needed to identify agents and discourses involved in the Chinese-language information problem, and to unravel the relationship among them.
The Primacy of Weak-Tie Social Curation
One final and distinctive feature of the information problem on WeChat is its relatively algorithm-free design. On WeChat, users encounter content in three primary ways where the influence of technological manipulation is markedly absent. First, users can subscribe to news accounts directly, and all subscriptions show up in a tab in the order in which they were last updated. Second, content sharing can be found in Moments, akin to Facebook’s News Feed but always sorted chronologically. Third, information circulates in the many private, invitation-only chat groups populating WeChat. This set of content exposure and curation process makes information flows in WeChat a distinctly socially driven experience, where the paths and fault lines of information, as well as misinformation, are determined by organic networks and their varying degrees of affinity and trust.
Compared to the two forms of social curation on WeChat, personal curation plays a more minor role in exposing users to information. Eighty-seven percent of the respondents reported reading an article posted by a friend in the week prior to the survey, and seventy-nine percent had read an article shared in a chat group, compared to fifty-seven who had actively browsed a subscribed news account (Photo 3). This is not surprising. Since all subscriptions show up in a separate tab, customizing and browsing content from subscription accounts becomes a much more intentional act. Perhaps even more so than other social media platforms, the strength of this personal curation process inside WeChat depends on the individual’s level of interest and motivation.
Chat groups emerged as a central mode of organization on WeChat. As many as ninety-two percent of the survey respondents reported belonging to a chat group. These range from smaller, more intimate groups formed with family members, friends, and colleagues, to groups organized around shared interests, affiliations, or locality where membership is subject to approval by the administrator. Examples given by survey respondents included parenting groups, local Chinese groups, professional networks, alumni groups, and hometown groups. Groups can also be formed ad hoc, around a certain event or topic. The abundance of these chat groups makes them a key process in the spread of information on WeChat.
It is chat groups that make the outcome of information flows on WeChat much more indeterminate. When one thinks of social networks driving information sharing, emails and friends networks on social media like Facebook may come to mind. These have been most commonly studied as a source of social curation.3940 But WeChat groups operate differently from Facebook and email networks: they can be larger and connected by much weaker relations. For example, group members could be the 200 odd graduates of the Peking University who currently live in Boston. These weak ties—acquaintances or mere individuals connected by a common affiliation—could potentially convey a larger and broader body of information. Among survey respondents, strong-tie groups with fewer than fifteen people are most common, but as many as seventy-one percent reported belonging to a group larger than 100 people. My interviews and observations point to a few features of WeChat chat groups that could complicate existing understandings of how social curation shapes information flows.
First, the degree of information exposure could depend on discussion and elaboration. In larger groups, interactions can quickly get buried. What prolongs the visibility of a shared article is engagement from other group members. As one interviewee said, “I do not always check messages from chat groups, but if I see a series of message alerts, I know something must be happening and I would be curious to read.” In other words, if a shared article triggers discussion, it may be viewed by more group members. This suggests contentious topics may get viewed more.
Second, sharing within a chat group does not necessarily signal “endorsement and social recommendation”/
Messaging apps similar to WeChat, such as Whatsapp, Line, and Kakaotalk, are increasingly becoming venues for sharing and discussing news.41 More like chain emails than social media, these apps are principally organized around affinity-based ties and private networks. The dynamics of these group networks have significant implications for how misinformation spreads and how echo chambers manifest in a more private platform like WeChat. This is a piece of the puzzle that warrants much more systematic research.
In May 2017, WeChat added the feature Top Stories, a tab with personalized news recommendations. This means WeChat’s content exposure mechanisms may see a gradual shift toward algorithmic curation. But for now, the spread of misinformation on WeChat, in the absence of bots,42 micro-targeting,43 and other technologically derived “attention hacking” tactics,44 serves as a reminder of the importance of organic social networks.
Conclusion and Lessons
Misinformation on WeChat is the joint outcome of ideological forces on the platform and dynamics of content production and distribution. These forces and dynamics bear parallels with digital news ecosystems more broadly. Decentralization of content publishing and the logic of the attention economy drive rampant clickbait headlines, emotional hyperbole, editorializing, and disregard for source verification, providing conditions for misinformation to circulate. But more specific to immigrant Chinese media, WeChat’s information problem takes place more on the local level, manifests on a different set of issues, and intersects with players and discourses both in the United States and China. Information flows on WeChat are still by and large a product of personal and social curation. Chat groups, diffuse and abundant, provide countless nodes for the distribution of information through weak-tie networks.
Without diminishing algorithmic accountability and the role of platforms in shaping information flows, findings from WeChat in many ways reinforce the important point that danah boyd and others have made about the cultural tensions underlying the spread of misinformation.4546 Many of the issues prone to misinformation harken back to the position of Chinese Americans in the social and political hierarchy of the US. While right-wing political discourse in the Chinese media ecosystem is highly problematic, underlying its strident tone on affirmative action, census disaggregation, and treatment of Muslims, it’s an appeal to unfairness and invisibility that could find wider resonance among immigrant Chinese. Most importantly, this invisibility can be seen as systemically produced, where Chinese Americans have been overlooked and disempowered, even vis-à-vis other racial minorities, in the social and political agenda. Studies have shown that Asian Americans are chronically underexposed to mobilization by political parties.47 A national survey on intergroup attitudes revealed that Asian Americans, while least described with negative traits by other groups, are also the least understood—the perpetual foreigner.48
Herein lies the primary cause of concern: perception of discrimination and anxiety among first-generation Chinese exist and are tied to very legitimate concerns and experiences; while they do not always express themselves in radical politics, news accounts—including but not limited to the partisan ones—could leverage these sentiments, aided by an ecosystem that is populated by opportunist content producers and designed to enable the circulation of misinformation. When political socialization is conducted within this information environment, it could fuel a type of politics hinging on more ethnocentric and zero-sum conceptions of interests.
Part of the answer to the information problem, I think, entails understanding these concerns and experiences. It includes asking, for example, why affirmative action as it is practiced now can be uncomfortable for even progressive Chinese. Or how being associated with wealth and targeted for crime may not be the best introduction to racial dynamics. It involves addressing the disconnect between Chinese and American issue agendas and news coverage. Fiona Ng, a reporter and producer for the Los Angeles-based KPCC, described herself as the only representative from English-language media in a press conference hosted by Chinese Americans for Trump.49 She has since started following their discussions on WeChat’s chat groups.
This brings me to the next point: more English-speaking media, government agencies, and community organizations should utilize WeChat as tool for two-way communication. I say two-way because the misinformation problem identified in this study requires listening as much as delivering correct information. Using WeChat means more than just joining the app. If closed chat groups are where information circulates and conversations take place, there is no easy workaround other than making real connections with individuals and communities. In this sense, WeChat serves as a reminder that dedicated engagement, more than technical solutions, is needed to combat misinformation.
In addition, substantial and credible local news is more crucial than ever for immigrant groups, especially on stories and policy issues prone to misinterpretation and distorted, or informationally incomplete. And these local news need to be accessible—in the right language as well as delivered through the right channels—to the immigrant population. The local news nonprofit Alhambra Source, which grew out of a research initiative at University of Southern California’s Annenberg School for Communication and Journalism, is an example of local journalism taking seriously the work of immigrant integration. Operating in a majority-minority suburb of Los Angeles, it has cultivated an ethnically and linguistically diverse base of resident contributors from the local area, which allows surfacing of issues and stories relevant to immigrant audiences. In addition to efforts in trilingual reporting (English, Spanish and Chinese), it also runs a dedicated WeChat account to connect Chinese residents with local issues and institutions. While such an extensive model of engagement may not be viable for all, ethnic press and existing WeChat outlets can be leveraged as partners for disseminating verified information. For example, laudable efforts in rumor tracking and counter-narratives have already been initiated by WeChat natives such as NoMelonGroup and iAmElection. Local chat groups would also be helpful networks to tap into.
Traditional ethnic media should not be discounted when new, WeChat-based players seem to have seized sizeable audience and discursive authority. Historically, ethnic media have had formidable influence on Chinese politics. From not endorsing Judy Chu’s city council bid in the 1990s50 to its editorials on affirmative action, ethnic media can also sway Chinese politics in a partisan direction. But these outlets could be vehicles of verified information. Others have observed that the explosion in “news” being generated and shared across digital and social media could lower the bar for news organizations.51 Ethnic media are already known to face challenges in accessing sources.52 With the increasing number of information outlets vying for attention, we need to ask how ethnic presses have responded to the change in their content and audience engagement strategies, and how they can be allies in combating misinformation.
I want to conclude by emphasizing that this is as much a case study of the Chinese media sphere and its particular narratives, players, and processes as it is an opportunity to reflect on immigrant Chinese and their media system as part of the larger political landscape in the United States. Similarly, we need to learn about what the Tagalog, Spanish, and Hindi news ecosystems look like and how narratives are constructed within those spheres. Many of the most divisive political issues are defined in relational terms. It is through understanding divergences and intersections in the political discourses of different groups and political constituencies that we can identify the sources and gaps of understanding and misunderstanding, information and misinformation.
Appendix A. List of Keywords
Appendix B. List of WeChat Outlets and View Count*
*WeChat displays view count larger than 100,000 as 100,000+
Many thanks to the Tow Center for funding this research project and all the staff—Kathy Zhang, Jon Keegan, Nausicaa Renner, and Abigail Hartstone in particular—for their patience and support along the way. This project would not have been possible without the teams at the Alhambra Source and Asian American Advancing Justice. Dr. Sandra Ball-Rokeach, Evelyn Moreno, Karin Wang, and Jessica Jinn provided tremendous insight and were instrumental in the design and implementation of the project. Thanks also to my research assistants: Charlie Chan Woo Hwang, Yifei Wang, and Qiyao Peng for their dedication and hard work. April 2018
- Matthew Matsaganis, Vikki Katz, and Sandra Ball-Rokeach, Understanding Ethnic Media: Producers, Consumers, and Societies, Journalism (Thousand Oaks, CA: Sage Publications, 2011).
- K. Viswanath and Pamela Arora, “Ethnic Media in the United States: An Essay on Their Role in Integration, Assimilation, and Social Control,” MassCommunication & Society, no. 1 (2000): 39–56.
- Adrián Félix, Carmen González, and Ricardo Ramírez, “Political Protest, Ethnic Media, and Latino Naturalization,” American Behavioral Scientist, no. 4 (2008): 618–634.
- Min Zhou, Guoxuan Cai, “Chinese Language Media in the United States: Immigration and Assimilation in American Life,” Qualitative Sociology, no. 3(2002): 419–41.
- Bill Turque, “These First-Generation Chinese Americans are Vigorously Opposing Sanctuary Laws,” Washington Post, March 20, 2017, https://www.washingtonpost.
com/local/md-politics/these- first-generation-chinese- americans-are-loudly-opposing- sanctuary-laws/2017/03/17/ 92728e94-09db-11e7-93dc- 00f9bdd74ed1_story.html?utm_ term=.a831f91efb04.
- Esther Wang, “Conservative Chinese Americans are Mobilizing, Politically and Digitally,” Pacific Standard, October 11, 2017, https://psmag.com/social-
justice/conservative-chinese- americans-are-mobilizing- politically-and-digitally.
- Gorbunova, Yulia. Laws of Attrition: Crackdown on Russia’s Civil Society After Putin’s Return to the Presidency. New York: Human Rights Watch, 2013. Accessed February 11, 2013. http://www.hrw.org/reports/2013/04/24/laws-attrition.
- Karthick Ramakrishnan, Janelle Wong, Taeku Lee, and Jennifer Lee, (2016). National Asian American Survey Pre-Election Survey [data file].
- China Tech Insights (2018). WeChat User & Business Ecosystem Report 2017. [online] Available at: https://www.chinatechinsights.com/report/21370582.html [Accessed 23 Apr. 2017].
- Mia Shuang Li, “How WeChat Became the Primary News Source in China,” Columbia Journalism Review, January 10, 2018, https://www.cjr.org/tow_
- Matt Sheehan, “Here’s What All The Chinese Students At Your School Are Reading,” Huffington Post, September 16, 2015, https://www.huffingtonpost.
com/entry/heres-what-the- 300000-chinese-students-in- the-us-are-reading_us_ 55f9b409e4b0e333e54c3e22.
- “Mediacloud,” https://mediacloud.org.
- Maxwell E. McCombs, Donald L. Shaw, “The Agenda-setting Function of Mass Media,” Public Opinion Quarterly, no. 2 (1972): 176–187.
- Jimmy Kimmel Live! “October 16,” no. 11.59, directed by Jonathan Kimmel. Written by Joelle Boucai, Josh Holloway, Jonathan Kimmel, Brian Paulk, and Danny Ricker. ABC, October, 2013.
- Karthick Ramakrishnan, “How Asian Americans Became Democrats,”The American Prospect, July 26, 2016, http://prospect.org/article/
- Chenchen Zhang, “The Curious Rise of the ‘White Left’as a ChineseInternet Insult,” Open Democray, May 11, 2017, https://www.opendemocracy.net/
digitaliberties/chenchen- zhang/curious-rise-of-white- left-as-chinese-internet- insult.
- Jason Wilson, “Why the Far Right Believes a US Civil War Will Start on Saturday,” The Guardian, November 1, 2017, https://www.theguardian.com/
world/2017/nov/01/far-right- conspiracy-theory-us-civil-war .
- Stephanie K. Baer, “West Covina, Walnut Leaders Say ‘Sanctuary State’Bill Would Let Felons Go Free,” San Gabriel Valley Tribune, March 29, 2017, https://www.sgvtribune.
com/2017/03/29/west-covina- walnut-leaders-say-sanctuary- state-bill-would-let-felons- go-free/.
- Brian Day, “West Covina Man Sentenced for DUI Crash That Killed Walnut Father,” The Daily Bulletin, March 10, 2017, https://www.dailybulletin.com/
2017/03/10/west-covina-man- sentenced-for-dui-crash-that- killed-walnut-father-2/.
- Minyi Zhang, “邵琪命案/蓄意撞死華人判四年不平之聲四起 (The Shao ChiTrial-Cries of Injustice After Driver Received 4-Year Prison Sentence for Inten-tionally Running Over Chinese Man),” World Journal, March 11, 2017, https://www.worldjournal.com/
4871606/article-%E9%82%B5%E7% 90%AA%E5%91%BD%E6%A1%88%EF%BC% 8F%E8%93%84%E6%84%8F%E6%92%9E% E6%AD%BB%E8%8F%AF%E4%BA%BA%E5% 88%A4%E5%9B%9B%E5%B9%B4-%E4% B8%8D%E5%B9%B3%E4%B9%8B%E8%81% B2%E5%9B%9B%E8%B5%B7/.
- “Alexa,” http://www.alexa.com.
- Tianjie Ma, “People’s Republic of Spiritual Rednecks,” Chublic Opinion, November 15, 2016, https://chublicopinion.com/
- Craig Silverman, (February 2015), https://towcenter.org/wp-
content/uploads/2015/02/ LiesDamnLies_Silverman_ TowCenter.pdf.
- Linda Lai and Audun Farbot, “What Makes You Click? The Effect of Question Headlines on Readership in Computer-Mediated Communication,” Social Influence, no. 4 (2013): 289–299.
- Jonathan Stray, “Defense Against the Dark Arts: Networked Propaganda and Counter-Propaganda,” Medium, February 27, 2017, https://medium.com/tow-center/
defense-against-the-dark- artsnetworked-%20propaganda- and-counter-propaganda- deb7145aa76a.
- Gordon Pennycook, Tyrone Cannon and David G. Rand, “Implausibility and Illusory Truth: Prior Exposure Increases Perceived Accuracy of Fake News but Has No Effect on Entirely Implausible Statements,” December 11, 2017, https://ssrn.com/
- Craig Silverman, “This Analysis Shows How Viral Fake Election News Stories Outperformed Real News on Facebook,” Buzzfeed, November 16, 2016, https://www.buzzfeed.com/
craigsilverman/viral-fake- election-news-outperformed- real-newson-%20facebook?utm_ term=.aaMXl5lvgG#.wtVW3y3jYn.
- Emily Thorson, “Belief Echoes: The Persistent Effects of Corrected Misinformation,” Political Communication, no. 3 (2016): 460–480.
- Jim Sanders, Victor Nee, and Scott Sernau, “Asian Immigrants’ Reliance on Social Ties in a Multiethnic Labor Market,” Social Forces, no. 1 (2002): 281–314.
- Wei Li, “Ethnoburb: The New Ethnic Community in Urban America,”December 2008.
- Yemile Bucay, Vittoria Elliott, Jennie Kamin, and Andrea Park, “America’s Growing News Deserts,” Columbia Journalism Review, 2017, https://www.cjr.org/local_
- . “Informing Communities: Sustaining Democracy in the Digital Age,” Washington, D.C.: The Aspen Institute, 2009.
- Claire Wardle and Hossein Derakhshan, Information Disorder: Toward an Interdisciplinary Framework for Research and Policymaking (September 27, 2017), https://shorensteincenter.org/
information-disorder- framework-for-research-and- policymaking/.
- Alice Marwick and Rebecca Lewis, “Media Manipulation and Disinformation Online,” Data & Society, May 2017. https://datasociety.net/output/media-manipulation-and-disinfo-online/
- Bo Liu, “中国人染上了“伊斯兰恐惧症” ?(Chinese Infected with”Islam-ophobia“?)” The New York Times, October 27, 2016, https://cn.nytimes.com/
- Solomon Messing and Sean J. Westwood, “Selective Exposure in the Age of Social Media: Endorsements Trump Partisan Source Affiliation When Selecting News Online,” Communication Research, no. 8 (2014): 1042–63.
- Eytan Bakshy, Solomon Messing, and Lada A. Adamic, “Exposure to Ide-ologically Diverse News and Opinion on Facebook,” Science, no. 6239 (June 5,2015): 1130–1132.
- Nic Newman with Richard Fletcher, Antonis Kalogeropoulos, David A. L.Levy, and Rasmus Kleis Nielsen, (2017), https://reutersinstitute.
politics.ox.ac.uk/sites/ default/files/Digital%20News% 20Report%202017%20web_0.pdf.
- Chengcheng Shao, Giovanni Luca Ciampaglia, Onur Varol, Alessandro Flammini, and Filippo Menaczer, “The Spread of Misinformation by SocialBots,” July 24, 2017, https://arxiv.org/abs/1707.
- Justin Hendrix and David Carroll, “Confronting a Nightmare for Democracy: Personal Data, Personalized Media and Weaponized Propaganda,” Medium, May 4, 2017, https://medium.com/@
profcarroll/confronting-a- nightmare-for-democracy- 5333181ca675.
- danah boyd. “Hacking the Attention Economy,” Data & Society: Points, January 15, 2017, https://points.datasociety.
- danah boyd. “Google and Facebook Can’t Just Make Fake News Disappear.” Backchannel, Medium, 27 Mar. 2017, “https://medium.com/backchannel/google-and-facebook-cant-just-make-fakenews- disappear-48f4b4e5fbe8”.
- Janelle Wong, Democracy’s Promise: Immigrants and American Civic Insti-tutions (Ann Arbor, MI: University of Michigan Press, 2006).
- J. Eric Oliver, The Paradoxes of Integration: Race, Neighborhood, and CivicLife in Multiethnic America (Chicago, IL: University of Chicago Press, 2010).
- Fiona Ng, “The Pro-Gun, Anti–Illegal Immigration Chinese Americans WhoLove Trump,” Vice, November 2, 2016, https://www.vice.com/en_us/
- John Horton, The Politics of Diversity: Immigration, Resistance, and Change in Monterey Park, California (Philadelphia, PA: Temple UniversityPress, 1995).
- Matsaganis, Matthew D, and Vikki S Katz. “How Ethnic Media Producers Constitute Their Communities of Practice: An Ecological Approach.” Journalism: Theory, Practice & Criticism, vol. 15, no. 7, 2013, pp. 926–944.