How Algorithms Win Over Chinese Audience?
Examining Users’ Motivations, Perceived Merits, and Usage of Chinese
Personalized News Applications based on Algorithms
by Mohan Wang Student ID: 12367737
Master’s Thesis
Graduate School of Communication
Erasmus Mundus Master “Journalism, Media & Globalisation” Supervisor: dr. J.E. Möller
Abstract
Personalized news applications based on algorithms are extremely popular in China now. The aim of the study is to explore firstly how algorithm-based news applications fulfill the needs of different types of news seeker in China. Then, further provide insight on why those applications are popular based on China’s unique political and media contexts. The study conducted an online survey among Chinese users (N=421) and resulted in several findings. Chinese users use
personalized news applications primarily to obtain information. Users who seek news for information value the accessibility while entertainment and social utility seeker prioritize the richness of information from non-mainstream media. Although personalized recommending is the core of such applications, no types of news readers consider it as the crucial merit. While such applications functioning as a platform of non-mainstream media channels, serves an important role for users who read the news to seek information and entertainment. The findings reveal that in a politically totalized state, personalized news applications based on algorithms are creating new media dependency that helps users expose to alternative information.
Introduction
In the era of information overload, news users have to make trade-offs due to limited time and attention. Professional editors used to make news selections for readers based on traditional news value. With the development of the internet and big data, algorithms became a new gatekeeper that empowers readers with a higher level of selective elasticity. (Gillespie, 2014) Algorithms screen from a vast amount of content, make the most precise matching of
recommendation to each reader. (Newman, Fletcher, Levy, & Nielsen, 2016)
According to Reuter digital report 2017(Newman, Fletcher, Kalogeropoulos, Levy, & Nielsen, 2017), more than half of respondents prefer algorithms select news for them rather than editors. Thurman, Moeller, Trilling, and Helberger’s (2019) study also confirmed a general preference for algorithmically selected news among respondents from 26 countries. Especially when the political independence of the news media falls in one country, the tendency is even stronger. Similarly, Wei and his colleagues (2014) found that press freedom level can affect mobile news consumption in one country. The less press freedom, the more mobile news seeking. In China, where all mainstream media are under strict control, personalized news applications based on algorithms have a very high market share, with strong user adhesion. (Zhu, 2018) Three out of every ten people in China today are users of algorithm-based news applications. Among the top ten news applications in 2018, algorithm-based news applications accounted for 50%. (Kew, 2018) In August 2018, two most popular personalized news applications: Headline
Toutiao and Tencent news reached 285 million users and 288 million users, respectively, with an average of 76 minutes of usage per user per day. (BusinessAnalysis, 2018)
However, popular area as it is, little attention has been paid to uncover how users perceive this new media technology in an authoritarian country. (Hyun & Kim, 2015) Current
studies on personalized algorithms are mostly from western countries, which are about the diverseness of content and potential consequences such as filter bubble and echo chambers. (DeVito, 2017; Diakopoulos, 2015; Napoli, 2015) “How audiences perceive the relative merits of algorithmic selection is largely non-existent.” (Thurman et al., 2019) Especially in China, where the media and political context are largely different from western countries, though some scholars researched social media and microblog users (Wang & Shi, 2018), no studies before examined the personalized news field.
To fill up these gaps, this study conducted an online survey among Chinese users (N=421) to undercover what the advantages of the personalized news applications based on algorithms in China are, and how users with different news seeking motivations perceive and use them. This paper combines and integrates factors from several research areas and puts forward four merits of personalized news applications. The result is discussed with regard to studies on Chinese citizen journalism and mobile news consumption and is compared to Western studies on user’s appreciation towards personalized news. As a preliminary study, it provides insight from the users' side, contributes to discussions on personalized algorithms and news seeking motivations in the authoritarian country, and facilitate an enhanced theoretical framing of China’s news media landscape.
Theoretical Background
Perceived merits of personalized news applications based on algorithms in China
As mentioned above, personalized news applications are popular among a large number of Chinese users. These applications, such as Tencent News and Headline Toutiao, take away the information dissemination right and reshuffle the media industry where mainstream journalism used to be the monopolist.
The definition of algorithms itself is simple. Broadly, algorithms are a series of code that convert input data into output results. Weber and Kosterich (2018) defined the algorithms as “a set of rules and a sequence of actions performed to accomplish a specific task.” Wang (2017) defined it as a "series of instructions input to solve a problem.” Algorithms in journalism mainly function as producing and distributing news. In this research, the focus is on the news
dissemination process for many mobile news applications and aggregators, such as Google News, Apple News in Western countries, Tencent News and Headline Toutiao in China. (Zheng, Zhong, & Yang, 2018) Accordingly, algorithms, in this case, are personalization algorithms, which continuously collect and store users’ personal information as input data and output recommended personalized news. (Wang, 2017)
To understand and clarify why people appreciate algorithms as the new gatekeeper and the popularity of personalized news applications, it is necessary to distinguish China from other western countries. (Bohman, 2004; Shao & Wang, 2017) Because Chinese online media and Internet environment are entirely different from democratic countries. (Guo, 2019; Sullivan, 2014) Therefore, based on its media and political contexts, I combined several aspects from previous studies on social media and mobile news in China; and put forward four perceived merits.
Exposure to non-mainstream media
China, as an authoritarian one-party country, the government maintains tight control over mainstream news media. (Chan, Chen, & Lee, 2017) The censorship is strict to media contents and access to many western websites are restricted. (Taneja & Wu, 2014) According to Nan (2013), two parallel public opinion fields in China exist at the same time: one is “mainstream media public opinion field,” and the other is oral or online public opinion field. Sullivan (2014)
found that Chinese users are more familiar with news portals and social media rather than the mainstream media’s website. The user stickiness to mainstream media is low (Xu, Qi, & Li, 2018), due to the homogeneity of content and viewpoint, and the role of being “mouthpiece of the party.” (Sparks, 2008)
Meanwhile, the restriction is less tight for citizen journalism as long as they do not
mobilize people against the government or the party. The government can tolerate some criticism and negative information online, but once it comes to social mobilization, they will delete it immediately. (King, Pan, & Robert, 2013; Poell, De Kloet, & Zeng, 2014) Under such policy, social media, and citizen journalism in China are becoming more and more crucial as they are holding roles of providing and communicating non-sanctioned information and viewpoints. Wang and Mark (2013) found that up to 70% of Chinese citizens received news from citizen journalism; they demonstrate no difference on the level of trust in official and citizen journalism channel and generally low trust in news from the government account. Chin-Fu’s (2013) case study supported that citizen journalism channels provide alternative information in China's propaganda-filled media environment and successfully affected the government’s policy agenda. Chinese users are even more likely to believe in rumors on social media (Gao, 2012) because traditional news media are always silent towards negative news.
Before the boom of the algorithmically distribution, people have to look for and follow citizen journalists or public accounts on different websites. (Luo & Harrison, 2019) What personalized news applications in China are doing solved this problem. They do not only aggregate news from traditional news outlets but also provide a platform for all citizen journalists to publish their work and express their opinions. Algorithms will then select and recommend content from the platform. Therefore, personalized news applications based on
algorithms, on the one hand, save readers’ time to go to different websites to acquire information; on the other hand, help grassroots media channels to get more exposure and enlarge their social influence. Hassid and Repnikova (2016) stated that online dissemination protect journalist when reporting sensitive issues and reduce the risks of political inference. Citizen journalism can even challenge mainstream media with such platforms. (Svensson, 2017) According to Yang (2009), digital media technologies enhanced and mediated Chinese civil discourse as they allow more public opinion visible in the online space.
Thus, exposure to non-mainstream media is considered as a merit for Chinese users as they can receive alternative information other than official media and follow non-mainstream channels. It is a unique indicator to see if personalized news applications in China has the role of evading censorship.
Wide range of content One important feature of mobile news applications is its capability of broadening a user’s news experience, as the internet largely increased available news sources to the audience.
(Chaffee & Metzger, 2001; Tewksbury, 2005) This feature is even stronger in Chinese
personalized news applications because sources are not only limited to traditional media outlets which are mostly under strict control, but also from citizen journalism, video sites, live report social media. (Luo& Harrison, 2019) Taking Headline Toutiao as an example, the total number of public channels exceeded 1.1 million, of which citizen journalism channels reached 900,000 until the end of 2017. (Toutiao, 2019) Through categories on the top page, including real-time hot topics, live news, technology, entertainment, sports, economic, military, international news, fashion, tourism, childcare, food, and history, readers can find whatever topics they interested in the applications.
Still, it should be noted that there are some problems behind this merit. Some scholars have concerns that most of the time citizen journalism in China providing soft news or
entertaining content rather than focusing on political and severe public issues. (Akhavan-Majid, 2004) A large portion of contents on those platforms is soft news, which is attractive, easy to understand, and have click-bait headlines. (Wei & Wan, 2017; Zhang, Liang, & Zhang, 2016). Wei and Wan (2017) analyzed 2924 news articles extracted from a personalized news
application, found that up to one-third are ambiguously misleading.
This tendency is mainly due to the operation of citizen journalism and algorithms as the gatekeeper. Journalists or content producers run public channels for profit-making, yet critical or political contents cannot always reward them financially while publishing entertaining content is much easier to get clicks and politically safer. Meanwhile, algorithms are developed and
controlled by commercial companies; their goal is to obtain more users and make profits. Therefore, algorithms would more likely to distribute and recommend compelling and non-repetitive content to attract users to stay as long as possible. (Tong, 2019)
Accessibility
The merit of convenient access derives from the researches of mobile news applications. (Van Damme, Courtois, Verbrugge, & De Marez, 2015) Caronia (2005) described mobile news consumption as "no-when-times at no-where-spaces", meaning that news reading is no longer scheduled activity. It could happen in undefined time and place as mobile news applications enable users to receive news anywhere anytime conveniently, provide users’ a channel to take full use the fragmented time. (Dimmick, Feaster, & Hoplamazian, 2011; Caronia, 2005) Shim and his colleague (2015) proposed that the accessibility feature is what distinct mobile news technology from other news platforms, and it is associated with online news reading motivations.
Accessibility is considered as one of the key factors that affect users when choosing information products. (Yuan, Fulk, Monge, & Contractor, 2010)
The internet survey in China (CNNIC, 2015) reported that reading news is the second most popular use among millions of Chinese mobile users. Especially for personalized news applications, they aggregate content that enables users to read their news feed without wasting time looking from multiple sources.
Tailored to personal interests
In addition to the above, this study considers a merit from the core feature of personalized algorithms, that is using personalized news applications to receive news of personal interest. According to FT.com’s production manager on personalized news, subscribers expect high relevance of news as they want to keep updated with their interested topic. (as cited in Thurman et al., 2019)
Personalized news applications based on algorithms, refers to mobile news applications that use algorithms as the core to mine and analyze user behavior, preference, potential needs, and automatically recommend news to meet their needs. (Liang, Lai & Ku, 2016) Weber and kosterich (2018) divided the mechanism of personalization algorithms into two levels:
classification and clustering. Algorithms first classify every news article based on its keywords and content, then use regression to complete the precise matching between users and content. From publicly available files of Headline Toutiao (Gao, 2018), similarly, content analysis and user tags are the two cornerstones of the system. To make the recommendations more precise, algorithms analyze users and categorize them with tags. User registration information, self-label, social relationship, browser and click history, geographic location, etc., are all used as the basis for analysis. (Wang, 2017)
China is not strict with privacy and data management and companies are not regulated by the EU General Data Protection Regulation (GDPR). It is easier for news aggregator companies to improve their personalized degree by reading users' information. According to the speech of the product manager in Headline Toutiao (Gao, 2018), basic information such as gender, age, location is automatically obtained when users’ log in with a third-party social media account, which is very common in China. Those data provide algorithms fundamental materials to make customized recommendations to new users who do not have browsing history in the applications before. After the user starts using the applications, algorithms will continuously learn the user's behavior, leading to an even more precise recommendation. Wang (2017) conducted a content analysis and found that no two people received the same news feed; even when the participants of the study were all college students from the same class whose background differences were relatively small. Moreover, gender differences and interests had already influenced the content disseminated by the algorithm. “Who you are "determines" what content is recommended to you."
Uses and Gratification of Mobile News Reading
Clearly, not all types of news readers consider the four merits in above sections are equally important to them. For some readers who follow news to get update of certain event, they may not see the richness of content as an advantage, while for those who want to explore more in vertical area, personalized recommendation feature is valued. Lai and Yang (2016) found that previous studies had conflict findings on Facebook using motivations; some studies showed that social interaction (Cheung, Chiu, & Lee, 2011; as cited in Lai & Yang, 2016) are users’ primary needs, yet some found that users use Facebook more for passing the time. (Lee, Lee, & Choi, 2012; as cited in Lai & Yang, 2016) They proposed that the explaining power of studies
decreased because Facebook have integrated many features, which previous studies didn’t take it into account. Thus, they explored the relationship between motivations and appreciation of
different Facebook features. Same applies to personalized news applications based on algorithms; as the four merits above derive from various functions. According to the motivational
consumption model (Lee & Chiyi, 2014), motivations are the predictor and decisive factor for perceived value and usage. News seeking motivations influence how much users value the merits and the usage.
Many studies on mobile news and social media applications adopted uses and
gratification theory (U&G). Pang (2019) investigated the association between the use of Chinese microblog and the offline civic engagement. Wei and Lo (2015) examined the mobile news consumption of Chinese users by proposing a model based on the U&G theory. Kang (2014) explored the new media technology adoption, resulted that entertainment, information seeking and social connection motivation are predictors of mobile applications usage.
The study adopted the U&G theory to explain news seeking motivations. (Blumler & Katz, 1974) This audience-centered framework assumes that individuals’ use media to satisfy their needs. (Blumler & Katz, 1974; Rubin, 2009). Studies showed that some people need to keep up with what going on in the world (Diddi & LaRose, 2006), while some need news to relax or just to have something to talk with others. (Lee, 2013) Flavián and Guerra (2006) have developed five motivations: to receive the update of issues, to seek information, to seek the pleasure of talking, to relax, to relieve boredom, to simply pass the time and habitual use. Shim and colleagues’ (2015) study divided why people access news with mobile into two dimensions: information-seeking and social utility (including entertainment). You and colleagues’ study (2013), Kang’s (2014) study proposed three dimensions of news seeking motivation:
information-seeking, social utility, and entertainment. In general, they all agree that leisure and entertainment, information seeking, and social utility are essential news seeking motivations. (Lee, Goh, Chua, & Ang, 2010)
Therefore, the study adopts three motivations and set up the following research question: RQ1: How do Chinese users with different news-seeking motivations appreciate the merit of personalized news applications?
RQ2: How do different types of news-seeking relate to usage of personalized news applications among Chinese users?
Entertainment and Leisure
This theme is defined as reading news to get entertainment and enjoyment, to occupy time and relieve boredom and day-to-day stress. (Lin, Gregor, & Ewing, 2008; Whiting & Williams, 2013) Research showed that entertainment motivation is positively associate with the usage of social media. (Nov, Naaman, & Ye, 2009; Shao, 2009) The political environment of China discouraged the discussion of public issues and lacks the education of civic awareness. Contents on social media are often more entertaining and far away from politics. The hot topic rankings on Chinese Twitter (Weibo) are usually dominated by gossip news. According to an internet use survey (Guo & Bu, 2001), more people seek for entertainment online than hard news. In the report, Guo and Bu (2001) concluded that the internet in China is an entertainment
highway rather than an information highway.
One of the uses of news in China is to “kill time” as users may not seriously reading the news but look for “more human-centered” topics in segments of a day. (Chan, 2015)
Personalized news applications based on algorithms suit for those who want to kill time because the algorithms generate endless newsfeed. Users who have spare time may unconsciously and
habitually check the news as they do not need to search for anything actively. Just refreshing the page, interest-oriented content will appear on the screen automatically.
Thus, the present study sets the following hypothesis:
H2a: Leisure and entertainment motivation for news-seeking will be positively related to usage of personalized news applications based on algorithms.
Information seeking
This theme is defined as reading news to seek out information. (Whiting & Williams, 2013) Information seeking includes learning the situation in the country and world (Bondad, Rice, & Pearce, 2012), satisfy curiosity through acquiring knowledge, and keep update with events. Studies showed that users with information seeking motivation follow more news on mobile (Wei et al., 2014) and it is the central reason for social media news consumption (Shim et al., 2015) The cognitive mediation model also proposed that individuals who are willing to be informed about current situation would devote considerable attention to news. (Eveland, Shah, & Kwak, 2003)
What more, Thurman and colleague’s (2019) study in 26 countries shows that users prefer personalized algorithms selection because they believe technology is immune from editorial control from untrustworthy media. It’s not easy for readers to acquire sufficient information from Chinese traditional media (Tai & Sun, 2007). Especially during public emergency events, people look for alternative channels to seek information as they have the sense that the government and mainstream media control and derive information from them. (Xie, Qiao, Shao, & Chen, 2017)
H2b: Information news seeking motivation will be positively related to the usage of personalized news applications based on algorithms.
Social utility
This theme combines social interaction and community utility, which is defined as using personalized news applications to communicate and interact, to share information with others. (Whiting& Williams, 2013; Bondad-Brown, Rice, & Pearce, 2012) Social gratification is used as a predictor when studying social media because socializing and sharing is easy through mobile with a few clicks. (Lee & Ma, 2012). Pew research found that almost half of internet users have shared news on social media. (Lenhart, Purcell, Smith, & Zickuhr, 2010) In authoritarian countries, social media plays an important role in providing a platform for users to express opinions and find like-minded others to interact and share information. (Chen & Chan, 2017) Meanwhile, personalized news applications have integrated the social function which enable users to share the news to their personal page or friends. This function promotes mobile news reading to be a social act.
The study sets the following hypothesis:
H2c: Social utility motivation for news-seeking will be positively related to the usage of personalized news applications based on algorithms.
Methodology
In order to study the media consumption and motivations, data needs to be collected directly from Chinese users. Online survey method was chosen as it is appropriate for only a few questions with clear meaning, and is widely adopted by many U&G studies. (Phellas, Bloch, & Seale, 2011; Lai & Yang, 2016; Pang, 2019). The survey was conducted for two weeks from Apr 4th to Apr 17th, 2019 by using Qualtrics. All participants signed the consent form that they
understand their responses were anonymous, and would only be used this research. The questionnaire is in Chinese, which was directly translated from the original codebook.
The study uses non-probability sampling due to its small research scale. What more, people who are using personalized news applications based on algorithms are hard to reach, because there is no particular online forum for the topic. Therefore, in order to reach more sample, snowball sampling was used. The author first sent the questionnaire through Wechat, to whom are using personalized news applications. After they filled in, they were asked to invite more people whom they think also meet the criteria of the research. In the questionnaire, one question is set to filter whether the respondent is valid for the research.
The link of the survey was opened 735 times and 521 complete responses in total, with a response rate of 70.9%. Of the returned questionnaires, 100 respondents do not use personalized news applications based on algorithms. Ultimately, a total of 421 valid responses were collected. Of those included in the analysis (N=421), more women (59.6%) than men (40.4%).
Respondent’s mean of age is 37, with distribution on all ages. Most of the respondents (37.7%) are in their 20s, 16% in their 30s, 20% in their 40s, and 21% in their 50s, 2.4% are 18 or 19, 2.8% older than 60. Most of the respondents’ highest level of education is bachelor degree (44%), 20% have a master degree, 14% only attended high school or lower, 2% of respondents hold a Ph.D. degree. Regarding their locating city level, most respondents are living in Tier 2 cities (66%), 15% living in Tier 1 cities. 11%, 7%, and 1% of respondents are from Tier 3, 4, 5 cities respectively.
Operationalization of Variables
To ensure that the participants are using personalized news applications based on algorithms, instead a self-reported method, a multi-choice question with options of eight top
personalized news applications were provided. Eight applications were selected from the ranking of top algorithmic news and information applications in China (Kew, 2018), and each has been checked manually to ensure it meets the definition in the paper. Through such way, participants do not need to be extremely clear about what is defined as a personalized news applications based on algorithms, as long as they choose one from the eight options or indicate the name. Those who never used personalized news applications was redirected to the end of the questionnaire.
There are several aspects of measuring the use of news media, such as frequency of use, the intensity of use, use of features, and interaction. This study focuses on two aspects: frequency of access and the time of use per day. Usage time ranges from 1-10 minutes to over 60 minutes (M=3.18, SD=2.05). Usage frequency ranges from 1 to 10 times (M=4.28, SD=3.59). In order to reduce the confusion, the survey provided examples under the question to make the calculation clear: "for example, if you open Headline Toutiao three times a day and Tencent News one time a day on average. Then the frequency is four times." "On average, if you read Headline Toutiao for 40 minutes, Tencent News for 15 minutes on average per day, the total usage time is 55 minutes."
News seeking motivations were measured with six statements that are expressing reasons why the respondent seek news, using a 5-point Likert scale: (1) strongly disagree to (5) strongly agree. Every two items were averaged to form one dimension. Statements are based on Shim et al.'s (2015) study. The first dimension refers to reading news to get leisure and entertainment, which is separated by two statements: “I read the news to have fun and relax.” “I read the news to kill time and relieve boredom.” (M = 3.26, SD = 1.15, Cronbach’s α = .80). The second dimension refers to seeking news to get more information, with: “I read the news to get in-depth
and further information.” “I read the news to obtain up-to-date information.” (M = 3.63, SD = 1.19, Cronbach’s α = .91). The third dimension refers to motivation for social utility: “I read the news in order to be social and to chat with others.” “I read the news to share with others.” (M = 3.42, SD = 1.17, Cronbach’s α = .89). The scale is reliable as all Cronbach's alpha have
exceed .80.
Four perceived merits are measured with the following statements, using a 5-point Likert scale: (1) strongly disagree to (5) strongly agree. “I appreciate personalized news applications because it provides a wide range of content.” (M = 3.61, SD = 1.16) “I appreciate personalized news applications because I can follow non-mainstream and citizen journalism channels.” (M = 3.63, SD = 1.10) “I appreciate personalized news applications because it provides precise and tailored content to my interests.” (M = 3.64, SD = 1.12) “I appreciate personalized news applications because I can access news anytime, anywhere.” (M = 3.73, SD = 1.11)
The survey first asked about the background information of respondents including gender, age, education level, current locating city. The classification of educational level ranges from high school or less, college, bachelor, master, and Ph.D. Cities are categorized into five levels, based on the hierarchical city tier system in China. (Yicai, 2018) The tier system was constructed with five indicators: economic resource, urbanized degree, human activity, lifestyle diversity, and long-term potential in the previous year by the authorized organization.
Statistical analysis
All data were analyzed using SPSS. Hierarchical regression analyses were conducted as it’s an effective framework to see if interested variables have the predictive capability of
dependent variables after accounting for other variables by building several regression models at each step. (Lewis, 2007) This analysis is widely adopted by many U&G studies on media use.
(Shim et al., 2015; Ye, Xu, & Zhang, 2017; Park, Kee, & Valenzuela, 2009) Therefore, the study uses hierarchical regression analyses to examine whether different news seeking motivations can predict users’ usage and perception towards personalized news applications.
Results
To answer RQ1, the association between different news-seeking motivations and how they perceive the merit of personalized news applications is examined. The independent variables were entered in two steps: (1) age, gender, city, education level, (2) news seeking motivations. Four perceived merits are dependent variables and three news seeking motivations as independent variables while controlling demographic variables. (see table 1)
Table 1 shows that for all four dependent variables, demographic variables have very weak prediction ability, could only explain no more than 2% of the variance of each model, with no significant association with the four perceived merits.
Users’ who read news for entertaining (b* = 0.18, t = 3.01, p < .01) and social utility (b* = 0.21, t = 2.66, p < .01) are more likely to appreciate personalized news applications which provides wide range of content. On the contrast, users who seek information from news reading are not significantly associate with the perceived merit of wealth content. News seeking
motivations largely increased the prediction accuracy to 23.6%( R2 change = 0.23, F = 19.49, p <.001), though the strength of prediction is still moderate.
When predicting the appreciation of following non-mainstream media on personalized news applications, gender (b* = -0.15, t = 3.08, p < .01) are significantly associated to this dependent variable. Male users are more likely to perceive this function as a merit. Motivation variables raise the prediction utility of the model to 25.2% (R2 change = 0.23, F = 21.26, p <.001). Seeking for entertainment (b* = 0.23, t = 3.90, p < .001) and information (b* = 0.23, t =
2.99, p < .01) have significant and moderate positive relation with exposure to non-mainstream media as the merit.
Table 1
Result of hierarchical regression analyses on news seeking motivations and perceived merits
Merit 1 Merit 2 Merit 3 Merit 4
b* SE b* SE b* SE b* SE Step 1 Age 0.04 0.01 0.03 0.01 0.01 0.01 0.03 0.01 Education 0.01 0.06 0.03 0.06 0.06 0.06 0.02 0.06 City 0.09 0.07 0.08 0.07 0.04 0.07 0.08 0.07 Gender -0.09 0.16 -0.15** 0.11 -0.09 0.11 -0.08 0.11 R2 0.01 0.02 0.00 0.00 F 1.69 3.08* 1.39 1.45 Step 2 Entertainment 0.18** 0.06 0.23*** 0.06 0.19** 0.06 0.12* 0.06 Information 0.15 0.08 0.23** 0.07 0.21** 0.07 0.28*** 0.07 Social Utility 0.21** 0.08 0.09 0.07 0.15 0.07 0.17* 0.07 R2 0.24 0.25 0.23 0.27 ΔR2 0.23 0.23 0.23 0.26 F 19.49*** 21.26*** 19.43*** 22.72*** ΔF 17.80*** 18.18*** 18.04*** 21.27***
Note. Merit 1 = Wide range of content. Merit 2 = Exposure to non-mainstream media. Merit 3 = Tailored to personal interest. Merit 4 = Accessibility.
Regarding the model of tailored to personal interest as the dependent variable, adding the second block: motivation variables increase the prediction accuracy to 23.5% (R2 change = 0.23, F = 19.43, p <.001). Motivation for social utility (b* = 0.15, t = 1.87, p =0.062) has no
significant relation with the personalized feature as the merit. While users who read news for entertainment (b* = 0.19, t = 3.15, p < .01) and seek information (b* = 0.21, t = 2.68, p < .01) display significant appreciation toward the personalized characteristics.
When coming to model 4, accessibility as the dependent variable, three news seeking motivations entered as the second block increased 26.2% of the accuracy of prediction (R2
change = 0.26, F = 22.72, p >0.05), and all three motivations are significantly associated with the dependent variable. Information seeking motivation (b* = 0.28, t = 3.63, p <.001) has the
strongest and most significant association with accessibility among three motivations.
Entertainment and social utility motivations also show positive association with the dependent variable, with significance at 0.05 degree.
To answer RQ2, usage frequency and time are dependent variables; three news seeking motivations are independent variables while demographic factors are control variables. (See table 2)
The first regression model has users’ access frequency to personalized news applications per day as dependent variable. The prediction utility of demographic factors is weak: only 6.5% of variation in usage frequency can be predicted based on respondents’ age, education, city and gender. (R2 = 0.07, F = 8.33, p <.001) Still, age (b* = 0.18, t = 3.46, p < .01); gender (b* = -0.19, t = -4.04, p < .001), both are significantly and weakly associate with usage frequency. Age is positively related to access frequency and male users are more likely to access personalized news applications in a day more often. News seeking motivations as predictors increase the accuracy
of the model slightly with 3%, up to 10%. (R2 change = 0.03, F = 7.14, p <.001) None of the three news seeking motivations are associated with usage frequency significantly.
Table 2
Result of hierarchical regression analyses on news seeking motivations and usage
Frequency Time b* SE b* SE Step 1 Age 0.18** 0.02 0.29*** 0.01 Education -0.02 0.18 0.01 0.10 City 0.04 0.21 0.04 0.12 Gender -0.19*** 0.35 -0.16** 0.20 R2 0.07 0.07 F 8.33*** 9.18*** Step 2 Entertainment 0.09 0.20 0.04 0.12 Information 0.10 0.26 0.17* 0.15 Social Utility 0.01 0.27 0.05 0.15 R2 0.10 0.12 ΔR2 0.03 0.05 F 7.14*** 9.35*** ΔF -1.19*** -0.17*** Note. * p <.05. ** p <.01. *** p <.001.
When taking usage time as dependent variable, demographic variables explained 7.2% of variance (R2 = 0.07, F = 9.18, p <.001) of the model. Age (b* = 0.228, t = 4.484, p < .001) and
gender (b* = -0.162, t = -3.422, p < .01) of personalized news applications users are significantly associate with their usage time. The second block contains three news seeking motivations, which increase the accuracy of prediction of usage to 12.2% (R2 change = 0.12, F = 9.35, p <.001). Users who read news to seek information (b* = 0.171, t = 2.044, p < .05) show a positive and significant relation to their usage time, therefore, H2b: Information news seeking motivation will be positively related to the usage is supported. Motivations for entertainment and leisure, social utility are not significantly relating to usage time. Hence, H2a: Leisure and entertainment motivation for news-seeking will be positively related to the usage, and H2c: Social utility motivation for news-seeking will be positively related to the usage can be rejected.
Conclusion and Discussion
The aim of the study is to explore firstly how algorithms based news applications fulfill the needs of different types of news seeker in China. Then, further provide insight on why those applications are popular based on China’s unique political and media contexts. The survey provides several meaningful and significant results.
Firstly, users with different news seeking motivations perceive the merit of personalized news applications differently. Information seeking is among the top of three motives, and it is the only motivation that significantly relates to the usage. In other words, personalized news
applications work primarily to fulfill the information needs of users and respondents who read the news for information spent more time on those apps. This result confirmed previous research that surveillance or information-seeking gratification is the strongest predictor of mobile news consumption. (Wei, Lo, Xu, Chen, & Zhang, 2013) Information acquiring has become the norm for Chinese users.
Accessibility is perceived as the central benefit for those who read the news to seek information. This association is supported by the characteristics of mobile news applications that they could update and generate news feed instantly with its temporal elasticity. (Caronia, 2005) Respondents visit these applications more than four times a day, while the average usage time is less than one hour. The usage result supports that fragmented reading is common on personalized news applications. (Van Damme et al., 2015)
Besides, information seekers also value news from unofficial media outlets and personalized recommendations, but they care less about the richness of content. The finding confirms the association between information seeking and appreciation of non-mainstream media; the more Chinese news readers want to be informed, the more they value non-mainstream media. It supports the argument that personalized news applications based on algorithms are functioning as a gateway to escape from traditional media. (Tai & Sun, 2007; Wei& Lo, 2015) This linkage also supports Moeller and Helberger’s (2018) prediction that when the content input is sufficient, and algorithms are sophisticated enough, personalized recommendations plays a key role in audiences’ exposure to information and creates a new media dependency.
Western studies have been debating on whether the algorithmic selection could be a threat to democracy. Because algorithms erode the editorial control of traditional media and could reduce challenging content to news readers, leading to the problem that the audience is not exposed to enough information to make informed decisions. (Beam, 2014) While in China, the situation is opposite. The result confirms that users need algorithm-based news applications to be well-informed because editorial selection in mainstream media is censored. With algorithms as the new gatekeeper which able to screen and recommend content automatically, in a “media-rich
but information-poor country” like China, personalized news applications as a unique news platform enable users to escape from the controlled media landscape. (Wei& Lo, 2015)
For those who read news for entertainment and kill time, they appreciate personalized news applications primarily because they can easily follow non-mainstream media and citizen journalism. The richness of content and personalized selection are also valued, while
accessibility is not the major concern. The high level of appreciation of non-mainstream media for those who seek leisure from news reading confirms concerns mentioned above. Although personalized news applications provide a large number of citizen journalism channels, many of them serve for the entertainment need. (Wei & Wan, 2017; Zhang, Liang, & Zhang, 2016) The commercial drive of algorithm-based news applications are the same in China and the West; both try to engage users as long as possible. (DeVito, 2017) The “capitalization of digital platform” (Langley & Leyshon, 2016) could lead to a tabloidized media environment as entertaining content is more likely to get clicks and revenues than hard news. As Tong (2019) argued, algorithms as the new gatekeeper do not have traditional news value when recommending content. Lacking guidance and gatekeeping, citizen journalism accounts would shift their goal from making good journalism for profit.
A wealth range of content from personalized news applications is perceived to be most beneficial for those who read the news to socialize online or offline. They want to read about a wide variety of topics rather than a tailored selection to their own interest. This result is easy to understand because extensive choices of information provide more topics than the personalized. The result also supports Shim's finding that social utility is not the significant predictor of social media news consumption. (Shim et al., 2015)
Surprisingly, no types of news seekers prioritized the personalized feature as a merit of algorithm-based news applications. Information and entertainment seeking news users regard it as an advantage. This conclusion is contradicting to the selling points and the core of these applications, as the personalized recommendation is what differ these applications from other mobile news products. The study provides two possible explanations for this result. Firstly, regardless of algorithms unique abilities to enable personalized news recommendation, information seeking is still the central needs of individuals’ news consumption; whether personalized or not does not change the fundamental attribute of the applications being an information platform. (Mitchelstein & Boczkowski, 2010) The other speculation is that due to homogenized mainstream media and social ideology, people do not have many options, they passively turn to these applications to fulfill their need. Social media such as Weibo and WeChat applications cannot provide rich and comprehensive news, while most commercial companies are adding algorithms recommendation feature to the mobile news applications to gain more user adhesion.
As for actual usage, only information seeking motivation and usage are positively associated. Gender and age are predictors of usage time and frequency. The older a user is, the more time and frequent a user would spend on personalized news applications daily. The result is conflicting to Logg (2017) and Thurman and colleague’s (2019) study that younger people prefer algorithmic choice to elder users. One possible reason is that younger Chinese news readers tend to be skeptical about the news from mainstream media and disappointed by such censored information, they already have multiple sources from social media and news channels. (Tong, 2019) Mindich’s (2005) study also found that those with lower media skills adopt news recommended through social media as they do not have the ability to discover credible news
channels. The result is contrary to Logg’s (2017) finding that appreciation towards algorithms does not differ by gender. Male users are more likely to invest time in such apps than female users do, and they are more interested in receiving information from non-mainstream media.
The results and findings offered by the survey lead to suggestions for future study as well as observes potential limitations. The major limitation is that data are collected by using non-probability sampling strategies, and the sample size is still small. Because of the snow sampling strategy, there is a possible sample bias. Comparing to the distribution of Chinese internet users (CNNIC, 2018), 67.8% of the population is under 40 years old with more male users, while more female participants are included in the study. Chinese internet users are mainly with high school degree with only 20% attended university (CNNIC, 2018), while over half of respondents from the current study have bachelor and higher degrees. Therefore, the result is less representative and should be interpreted with caution. Besides, as no previous study has examined or proposed merits of personalized news applications in China, the four themes are developed by the author based on research in Chinese social media and mobile news consumption. Additional factors could be added in the future. Fourthly, the usage time and frequency are self-reported, which means that respondents could underestimate how much time they spend on applications. As mentioned above, those applications provide non-repetitive news feed that attracts users to unconsciously and habitually engage on the applications, of which time may not be calculated.
Based on the current method, future studies may move further to create a prediction model by testing mediating factors between motivations and actual usage. Perceived merits variables could, therefore, be tested as mediators. The survey only examined the usage frequency and time as a preliminary study. If future studies examine the actual relationship between
like how many and what kind of citizen journalism channels that one user is following. Results also provide several points that could be further developed. For instance, news readers with information seeking and entertainment seeking motivations all value non-mainstream media source but the quality and credibly of contents from those citizen journalism channels need to be analyzed. Demographic factors show a very limit, almost none effect on how users perceive personalized news algorithms. A future study could repeat the study with other sample groups and include more aspects, such as income, political identification, and intimacy to technology. The study contributes to existing studies on personalized news in many aspects. Firstly, as mentioned above, few studies tested how audience perceive the relative merit of personalized news, especially in China. The study examined personalized algorithms from users’ perspective, and attempted to establish associations between news seeking motivations, perceived merits, and usage based on uses and gratification framework. The empirical finding indicates that users with different news seeking motivation perceive the merit of personalized news applications
differently, information seeking motivation positively relate to the usage. Users who seek news for information value the accessibility while entertainment and social utility seeker prioritize the richness of information from non-mainstream media. It expands the U&G research to the field of personalized news applications. Secondly, the study sorts out and conceptualizes the merits of personalized new applications based on China’s political and media context, providing indicators to understand the role of personalized algorithms in an authoritarian country. The survey result reveals an unexpected finding that regardless of personalized recommendation being the core of such applications, no types of users consider it as key merit. While personalized news
applications being a platform of citizen journalism are valued by both information seeking and entertainment seeking news users. Furthermore, the study links the result to western studies on
personalized news, showing an opposing observation that rather than reducing diversity and creating filter bubble, Chinese users rely on personalized news applications based on algorithms to be well-informed. These findings also have implications for citizen journalist and developer of personalized news applications. They should satisfy the information need of users by providing high-quality content and effective information disseminating feature rather than only focusing on increasing it to a personalized degree.
References
Akhavan-Majid, R. (2004). Mass media reform in China: Toward a new analytical framework. Gazette (Leiden, Netherlands), 66(6), 553-565.
Beam, M. A. (2014). Automating the news: How personalized news recommender system design choices impact news reception. Communication Research, 41(8), 1019-1041.
Blumler, J. G., & Katz, E. (1974). The uses of mass communications: Current perspectives on gratifications research (Vol. 1974). Sage Publications, Inc.
Bohman, J. (2004). Expanding dialogue: The Internet, the public sphere and prospects for transnational democracy. The sociological review, 52(1_suppl), 131-155.
Bondad-Brown, B. A., Rice, R. E., & Pearce, K. E. (2012). Influences on TV viewing and online user-shared video use: Demographics, generations, contextual age, media use,
motivations, and audience activity. Journal of Broadcasting & Electronic Media, 56(4), 471-493.
BusinessAnalysis. (2018, April 23). Headline Toutiao in-depth analysis report. Retrieved May 30, 2019, from https://new.qq.com/omn/20180423/20180423G0UJ0B.html
Caronia, L. (2005). Feature Report: Mobile Culture: An Ethnography of Cellular Phone Uses in Teenagers’ Everyday Life. Convergence, 11(3), 96-103.
Chaffee, S. H., & Metzger, M. J. (2001). The end of mass communication?. Mass communication & society, 4(4), 365-379.
Chan, M. (2015). Examining the influences of news use patterns, motivations, and age cohort on mobile news use: The case of Hong Kong. Mobile Media & Communication, 3(2), 179-195.
Chan, M., Chen, H. T., & Lee, F. L. (2017). Examining the roles of mobile and social media in political participation: A cross-national analysis of three Asian societies using a
communication mediation approach. New media & society, 19(12), 2003-2021. Chen, Z., & Chan, M. (2017). Motivations for social media use and impact on political
participation in China: A cognitive and communication mediation
approach. Cyberpsychology, Behavior, and Social Networking, 20(2), 83-90.
Cheung, C. M., Chiu, P. Y., & Lee, M. K. (2011). Online social networks: Why do students use facebook?. Computers in Human Behavior, 27(4), 1337-1343.
Chin-Fu, H. (2013). Citizen journalism and cyberactivism in China's anti-PX plant in Xiamen, 2007–2009. China: An International Journal, 11(1), 40-54.
CNNIC. (2015). The 35th Statistical report on internet development in China. Retrieved
from https://cnnic.com.cn/IDR/ReportDownloads/201507/P020150720486421654597.pdf CNNIC. (2018). The 43rd Statistical report on internet development in China. Retrieved
from https://cnnic.com.cn/IDR/ReportDownloads/201807/P020180711391069195909.pdf DeVito, M. A. (2017). From editors to algorithms: A values-based approach to understanding
story selection in the Facebook news feed. Digital Journalism, 5(6), 753-773.
Diakopoulos, N. (2015). Algorithmic accountability: Journalistic investigation of computational power structures. Digital Journalism, 3(3), 398-415.
Diddi, A., & LaRose, R. (2006). Getting hooked on news: Uses and gratifications and the formation of news habits among college students in an Internet environment. Journal of Broadcasting & Electronic Media, 50(2), 193-210.
Dimmick, J., Feaster, J. C., & Hoplamazian, G. J. (2011). News in the interstices: The niches of mobile media in space and time. new media & society, 13(1), 23-39.
Elvestad, E., Blekesaune, A., & Aalberg, T. (2014). The polarized news audience? A
longitudinal study of news-seekers and news-avoiders in Europe. A Longitudinal Study of News-Seekers and News-Avoiders in Europe (July 22, 2014).
Flavian, C., & Gurrea, R. (2006). The role of readers' motivations in the choice of digital versus traditional newspapers. Journal of Targeting, Measurement and Analysis for
Marketing, 14(4), 325-335.
Gao, H. (2012). Rumor, lies, and weibo: How social media is changing the nature of truth in china. The Atlantic, 16.
Gao, X. (2018, January 16). Detailed explanation of Headline Toutiao recommendation algorithms. Retrieved May 30, 2019, from https://36kr.com/p/5114077
Gillespie, T. (2014). The relevance of algorithms. Media technologies: Essays on communication, materiality, and society, 167, 167.
Guo, L. (2019). Media Agenda Diversity and Intermedia Agenda Setting in a Controlled Media Environment: A Computational Analysis of China’s Online News. Journalism Studies, 1-18.
Guo, L., & Bu, W. (2001). Survey report of Internet use and its influence: Beijing, Shanghai, Guangzhou, Chengdu and Changsha 2000. Beijing: Chinese Academy of Social Sciences.
Hassid, J., & Repnikova, M. (2016). Why Chinese print journalists embrace the Internet. Journalism, 17(7), 882-898.
Hyun, K. D., & Kim, J. (2015). The role of new media in sustaining the status quo: online political expression, nationalism, and system support in China. Information, Communication & Society, 18(7), 766-781.
Kang, S. (2014). Factors influencing intention of mobile application use. International Journal of Mobile Communications, 12(4), 360-379.
Kew. (2018, April 28). Cheetah Big Data 2018Q1 China App Ranking. Retrieved from https://news.mydrivers.com/1/573/573714.htm
King, G., Pan, J., & Roberts, M. E. (2013). How censorship in China allows government criticism but silences collective expression. American Political Science Review, 107(2), 326-343.
Lai, C. Y., & Yang, H. L. (2016). Determinants and consequences of Facebook feature use. New Media & Society, 18(7), 1310-1330.
Langley, P., & Leyshon, A. (2016). Platform capitalism: The intermediation and capitalization of digital economic circulation. Finance and society, 2(1).
Lee, A. M. (2013). News audiences revisited: Theorizing the link between audience motivations and news consumption. Journal of Broadcasting & Electronic Media, 57(3), 300-317. Lee, A. M., & Chyi, H. I. (2014). Motivational consumption model: Exploring the psychological
structure of news use. Journalism & Mass Communication Quarterly, 91(4), 706-724. Lee, C. S., & Ma, L. (2012). News sharing in social media: The effect of gratifications and prior
Lee, C. S., Goh, D. H. L., Chua, A. Y., & Ang, R. P. (2010). Indagator: Investigating perceived gratifications of an application that blends mobile content sharing with
gameplay. Journal of the American Society for Information Science and Technology, 61(6), 1244-1257.
Lee, J., Lee, M., & Choi, I. H. (2012). Social network games uncovered: Motivations and their attitudinal and behavioral outcomes. Cyberpsychology, Behavior, and Social
Networking, 15(12), 643-648.Influences on TV viewing and online user-shared video use: Demographics, generations, contextual age, media use, motivations, and audience
activity. Journal of Broadcasting & Electronic Media, 56(4), 471-493.
Lenhart, A., Purcell, K., Smith, A., & Zickuhr, K. (2010). Social Media & Mobile Internet Use among Teens and Young Adults. Millennials. Pew internet & American life project. Lewis, M. (2007, Feb). Stepwise versus Hierarchical Regression: Pros and Cons Paper
presented at the Annual Meeting of the Southwest Educational Research Association.
Retrieved from https://eric.ed.gov/?id=ED534385
Liang, T. P., Lai, H. J., & Ku, Y. C. (2006). Personalized content recommendation and user satisfaction: Theoretical synthesis and empirical findings. Journal of Management Information Systems, 23(3), 45-70.
Lin, A., Gregor, S., & Ewing, M. (2008). Developing a scale to measure the enjoyment of web experiences. Journal of Interactive Marketing, 22(4), 40-57.
Logg, J. M. (2017). Theory of Machine: When do people rely on algorithms?.
Logg, J. M., Minson, J. A., & Moore, D. A. (2019). Algorithm appreciation: People prefer algorithmic to human judgment. Organizational Behavior and Human Decision Processes, 151, 90-103.
Luo, Y., & Harrison, T. M. (2019). How citizen journalists impact the agendas of traditional media and the government policymaking process in China. Global Media and China, 4(1), 72-93.
Mitchelstein, E., & Boczkowski, P. J. (2010). Online news consumption research: An assessment of past work and an agenda for the future. New Media & Society, 12(7), 1085-1102. Moeller, J., & Helberger, N. (2018). Beyond the filter bubble: Concepts, myths, evidence and
issues for future debates.
Napoli, P. M. (2015). Social media and the public interest: Governance of news platforms in the realm of individual and algorithmic gatekeepers. Telecommunications Policy, 39(9), 751-760.
Newman, N., Fletcher, R., Kalogeropoulos, A., Levy, D., & Nielsen, R. K. (2017). Reuters Institute digital news report 2017. Reuters Institute for the Study of Journalism. Newman, N., Fletcher, R., Levy, D. A., & Nielsen, R. K. (2016). Digital news report
2016. Reuters Institute for the Study of Journalism.
Nov, O., Naaman, M., & Ye, C. (2010). Analysis of participation in an online photo-‐sharing community: A multidimensional perspective. Journal of the American Society for Information Science and Technology, 61(3), 555-566.
Sullivan, J. (2014). China’s Weibo: Is faster different?. New media & society, 16(1), 24-37. Pang, H. (2018). Can microblogs motivate involvement in civic and political life? Examining
uses, gratifications and social outcomes among Chinese youth. Online Information Review, 42(5), 663-680.
Park, N., Kee, K. F., & Valenzuela, S. (2009). Being immersed in social networking environment: Facebook groups, uses and gratifications, and social outcomes. CyberPsychology &
Behavior, 12(6), 729-733. l College, 18(3), 5-9.
Phellas, C. N., Bloch, A., & Seale, C. (2011). Structured methods: interviews, questionnaires and observation. Researching society and culture, 3.
Poell, T., De Kloet, J., & Zeng, G. (2014). Will the real Weibo please stand up? Chinese online contention and actor-network theory. Chinese Journal of Communication, 7(1), 1-18. Rubin, A. M. (2009). Uses-and-gratifications perspective on media effects. In Media effects (pp.
181-200). Routledge.
Shao, G. (2009). Understanding the appeal of user-generated media: a uses and gratification perspective. Internet research, 19(1), 7-25.
Shao, P., & Wang, Y. (2017). How does social media change Chinese political culture? The formation of fragmentized public sphere. Telematics and Informatics, 34(3), 694-704. Shim, H., You, K. H., Lee, J. K., & Go, E. (2015). Why do people access news with mobile
devices? Exploring the role of suitability perception and motives on mobile news use. Telematics and Informatics, 32(1), 108-117.
Sparks, C. (2008). Media systems in transition: Poland, Russia, China. Chinese Journal of Communication, 1(1), 7-24.
Svensson, M. (2017). The rise and fall of investigative journalism in China: digital opportunities and political challenges.
Tai, Z., & Sun, T. (2007). Media dependencies in a changing media environment: The case of the 2003 SARS epidemic in China. New Media & Society, 9(6), 987-1009.
Taneja, H., & Wu, A. X. (2014). Does the Great Firewall really isolate the Chinese? Integrating access blockage with cultural factors to explain Web user behavior. The Information Society, 30(5), 297-309.
Tewksbury, D. (2005). The seeds of audience fragmentation: Specialization in the use of online news sites. Journal of broadcasting & electronic media, 49(3), 332-348.
Thurman, N., Moeller, J., Helberger, N., & Trilling, D. (2019). My friends, editors, algorithms, and I: Examining audience attitudes to news selection. Digital Journalism, 1-23.
Tong, J. (2019). The Taming of Critical Journalism in China: A combination of political, economic and technological forces. Journalism Studies, 20(1), 79-96.
Toutiao. (2017, October 1). Media partners. Retrieved May 30, 2019, from https://www.toutiao.com/media_partners/
Van Damme, K., Courtois, C., Verbrugge, K., & De Marez, L. (2015). What’s APPening to news? A mixed-method audience-centred study on mobile news consumption. Mobile Media & Communication, 3(2), 196-213.
Wang, H., & Shi, F. (2018). Weibo use and political participation: the mechanism explaining the positive effect of Weibo use on online political participation among college students in contemporary China. Information, Communication & Society, 21(4), 516-530.
Wang, Q. (2017). Open the "black box" of algorithmic distribution: quantitative research based on Toutiao’s news feed. Journalism Review, 9, 002.
Wang, Y., & Mark, G. (2013, February). Trust in online news: Comparing social media and official media use by Chinese citizens. In Proceedings of the 2013 conference on Computer supported cooperative work (pp. 599-610). ACM.
Weber, M. S., & Kosterich, A. (2018). Coding the News: The role of computer code in filtering and distributing news. Digital Journalism, 6(3), 310-329.
Wei, R., & Lo, V. H. (2015). News on the move: Predictors of mobile news consumption and engagement among Chinese mobile phone users. Electronic News, 9(3), 177-194. Wei, R., Lo, V. H., Xu, X., Chen, Y. N. K., & Zhang, G. (2014). Predicting mobile news use
among college students: The role of press freedom in four Asian cities. new media & society, 16(4), 637-654.
Wei, W., & Wan, X. (2017). Learning to identify ambiguous and misleading news headlines. arXiv preprint arXiv:1705.06031.
Xie, Y., Qiao, R., Shao, G., & Chen, H. (2017). Research on Chinese social media users’ communication behaviors during public emergency events. Telematics and Informatics, 34(3), 740-754.
Xu, F., Qi, Y., & Li, X. (2018). What affects the user stickiness of the mainstream media websites in China?. Electronic Commerce Research and Applications, 29, 124-132. Yang, G. (2009). The power of the Internet in China: Citizen activism online. Columbia
University Press.
Ye, Y., Xu, P., & Zhang, M. (2017). Social media, public discourse and civic engagement in modern China. Telematics and Informatics, 34(3), 705-714.
Yicai. (2018, April 27). China's city rankings have changed: Shanghai became the top, Wuxi back to Tier2. Retrieved May 30, 2019, from
You, K. H., Lee, S. A., Lee, J. K., & Kang, H. (2013). Why read online news? The structural relationships among motivations, behaviors, and consumption in South
Korea. Information, Communication & Society, 16(10), 1574-1595.
Yuan, Y. C., Fulk, J., Monge, P. R., & Contractor, N. (2010). Expertise directory development, shared task interdependence, and strength of communication network ties as multilevel predictors of expertise exchange in transactive memory work groups. Communication Research, 37(1), 20-47.
Zhang, H., Liang, S., & Zhang, S. (2016). The current status and the future of self media. News and Writing, 5, 28–31.
Zheng, Y., Zhong, B., & Yang, F. (2018). When algorithms meet journalism: The user perception to automated news in a cross-cultural context. Computers in Human Behavior, 86, 266-275.
Zhu, X. (2018). The impact of personalized recommendation based on algorithmic mechanism in the new media era on Headline Toutiao’s news dissemination. Journal of Hunan Mass
Appendices Appendix-A: Independent and control variables
Variables Survey Items Response range
News seeking motivation 1: Entertainment and Leisure
I read news to have fun and relax.
1 = Strongly disagree 2 = Tend to disagree 3 = Neutral
4 = Tend to agree 5 = Strongly agree I read news to kill time and
relieve boredom.
News seeking motivation 2: Information seeking
I read news to get in-depth and further information.
I read news to obtain up-to-date information.
News seeking motivation 3: Social utility
I read news to in order to be social and to chat with others.
I read news to share with others.
Gender What is your gender? 1=Male, 2=Female
Age What is your age? 18-70
Education
What is your highest level of education? 1=High school 2=College 3=BA 4=MA 5=PhD
City
Which city are you currently living in? 1=Tier 5 city 2= Tier 4 city 3= Tier 3 city 4= Tier 2 city 5= Tier 1 city
(Filter question to check if the respondent valid for the research)
Which personalized news applications based on algorithms are you currently using? (If answer=I don’t use personalized news platform, end survey)
Headline Toutiao Tencent News Netease News Tiantian news Yidian News UC News ZAKER Flipboard
Others: Please specify I don’t use personalized news platform.
Appendix-B: Dependent variables
Variables Survey items Response range
Usage frequency
You said you are using personalized news applications.
How often did you access them per day? 1-10 times
Usage time
How long did you spend on personalized news applications each day?
1= 1-10 minutes 2=11-20 minutes 3=21-30 minutes 4=31-40 minutes 5= 41-50 minutes 6= 51-60 minutes 7= more than 60 minutes Perceived merit 1:
I use personalized news applications because it provides wide range of content.
1 = Strongly disagree 2 = Tend to disagree 3 = Neutral 4 = Tend to agree 5 = Strongly agree Perceived merit 2:
I appreciate personalized news applications because I can follow non-mainstream and citizen journalism channels.
Perceived merit 3:
I appreciate personalized news applications because it provides precise and tailored content to my interests. Perceived
merit 4: I appreciate personalized news applications because I can access to news anytime anywhere.
Appendix-C: Descriptive statistics of variables Items M SD Cronbach's α Gender 1.6 0.49 Age 36.78 12.18 Education 2.76 0.99 City 3.86 0.81 Frequency 4.28 3.59 Usage Time 3.18 2.05
Entertainment and Leisure seeking 3.26 1.15 0.804
I read news to have fun and relax. 3.34 1.25
I read news to kill time and relieve boredom. 3.18 1.27
Information seeking 3.63 1.19 0.913
I read news to get in-depth and further information. 3.57 1.21 I read news to obtain up-to-date information. 3.69 1.26
Social utility seeking 3.42 1.17 0.887
I read news to in order to be social and to chat with others. 3.57 1.21
I read news to share with others. 3.69 1.26
I use personalized news applications because it provides wide range
of content. 3.61 1.15
I use personalized news applications because I can follow
non-mainstream and citizen journalism channels. 3.63 1.10 I use personalized news applications because it provides precise and
tailored content to my interests. 3.64 1.12
I use personalized news applications because I can access to news
Appendix-D: Composition of the sample
Variables Items Frequency Percentage
Gender Male 170 40.4% Female 251 59.6% Age 18-19 10 2.4% 20-29 159 37.7% 30-39 67 15.8% 40-49 84 20.1% 50-59 88 21% 60-69 13 2.8% Education High school 60 14.3% College 84 20% BA 185 43.9% MA 83 19.7% Ph.D. 9 2.1% City Tier 1 63 15% Tier 2 276 65.6% Tier 3 46 10.9% Tier 4 30 7.1% Tier 5 6 1.4% Total N=421 100%
Appendix-E: City tier system (Yicai, 2018) Level Cities
Tier 1 Beijing, Shanghai, Guangzhou, Shenzhen
Tier 2 Chengdu, Hangzhou, Wuhan, Chongqing, Nanjing, Tianjin, Suzhou, Xi'an, Changsha, Shenyang, Qingdao, Zhengzhou, Dalian, Dongguan, Ningbo Tier 3 Xiamen, Fuzhou, Wuxi, Hefei, Kunming, Harbin, Jinan, Foshan,
Changchun, Wenzhou, Shijiazhuang, Nanning, Changzhou, Quanzhou, Nanchang, Guiyang,Taiyuan , Yantai, Jiaxing, Nantong, Jinhua, Zhuhai, Huizhou, Xuzhou, Haikou, Ürümqi, Shaoxing, Zhongshan, Taizhou, Lanzhou
Tier 4 Weifang, Baoding, Zhenjiang, Yangzhou, Guilin, Tangshan, Sanya, Huzhou, Hohhot, Langfang, Luoyang, Weihai, Yancheng, Linyi, Jiangmen, Shantou, Taizhou , Quzhou, Handan, Jining, Wuhu, Zibo, Yinchuan, Liuzhou,
Mianyang, Zhanjiang, Anshan, Quzhou, Daqing, Yichang, Baotou, Xianyang, Qinhuangdao, Zhuzhou, Putian, Jilin, Huai'an, Zhaoqing, Ningde, Hengyang, Nanping, Lianyungang, Dandong, Lijiang, Jieyang, Yanbian Korean Autonomous Prefecture, Zhoushan, Jiujiang, Longyan, Luzhou, Fushun, Xiangyang, Shangrao, Yingkou, Sanming, Handan, Lishui, Yueyang, Qingyuan, Jingzhou, Tai'an, Luzhou, Panjin, Dongying, Nanyang, Ma'anshan, Nanchong, Xining, Xiaogan, Qiqihar
Tier 5 Other cities and countryside
