Anonymous sources hurt credibility of news stories across cultures: A comparative experiment in America and China

Abstract

This cross-cultural experiment (N = 620) tested the impact of the use of anonymous sources on perceived news story credibility in America and China, two countries with assumed different journalistic standards. Both Americans and Chinese rated news stories with only anonymous sources as less credible than stories with identified sources. The attitude of Americans towards news stories was found to be more positive. The study represents the first comparative research on the topic with rigorously established cross-cultural equivalences.

Keywords

America anonymous sources China cross-cultural communication news credibility

Introduction

Ever since The Washington Post uncovered the Watergate scandal using the information from Deep Throat in 1973, journalists and researchers have debated the value of the unnamed sources in news articles (e.g. Blankenburg, 1992; Culbertson, 1978; Duffy and Freeman, 2011; Martin-Kratzer and Thorson, 2007; Smith, 2007; Sternadori and Thorson, 2009). Defenders argued that, if used carefully, confidential sources could be a ‘valuable tool’ that could help bring to light important stories that otherwise would have not surfaced (Shepard, 1994). On the other hand, the critics postulated that the practice opens the door to speculation by ‘spin doctors’ and ‘leakers’ (Haiman, 1998; Sternadori and Thorson, 2009). In extreme cases, such as in the KNBC-TV coverage of the O.J. Simpson trial, use of anonymous sources can lead to inaccurate reporting or even to fabrication of reports, as in the case of Janet Cooke, whose Pulitzer Prize-winning story written for The Washington Post was discovered to be made up (Shepard, 1994).

As the public has limited means of judging the legitimacy of information provided by anonymous sources, their usage could create media credibility problems. Research has found that newspaper articles, in which information was based on unidentified sources, were perceived as less credible than those in which veiled sources were avoided (Sternadori and Thorson, 2009). Practitioners themselves contend that the method undermines journalistic credibility. Allen H. Neuharth, founder of USA Today, thought that damaging examples of the use of unnamed sources in journalistic articles highly exceeded the number of positive ones and even called for the prohibition of anonymous sources. ‘On balance, the negative impact is so great that we can't overcome the lack of trust until or unless we ban them,’ Neuharth said (Shepard, 1994: 20).

Although newer research has found that the usage of anonymous sources has significantly declined in American newspapers and on TV networks (Duffy and Williams, 2011; Martin-Kratzer and Thorson, 2007), the credibility issues remain. In the United States, usage of veiled sources in news articles makes audiences suspicious of journalistic competence (Sternadori and Thorson, 2009). Inability to properly attribute a quotation, a decades-long journalistic norm, is seen as a ‘sloppy’ practice or laziness to corroborate (Strupp, 2005). However, in non-democratic nations, journalistic standards tend to be lower than in nations with the long tradition of liberal democracy. For example, lack of public access to information and sources, lack of freedom of expression, strong governmental pressures against investigative reporting, as well as frequent non-application of journalistic ethical rules, make reliance on unnamed sources a common occurrence in news articles in non-democracies (IREX, 2008, 2010, 2011). Do then audiences across countries with different journalistic standards perceive anonymous sources as cues for credibility assessment? This experimental study tests how the use of anonymous sources in factual news stories affects readers' perceptions of credibility in America, where journalistic professionalism is strongly developed (Hallin and Mancini, 2004), and in China, where such standards are lacking (Chen, 2004; Chan et al., 2006). In this regard, this study represents the first effort to verify whether news credibility cues are equivalent across nations with traditionally different journalistic practices.

The concept of anonymity

Anonymity is one value of a broader dimension of identifiability. According to Marx (1999), to be fully anonymous means that a person cannot be, in any context, identified according to the seven types of identity knowledge: his or her name, location, pseudonyms (nicknames but also numerical symbols such as a social security number), other symbols that could be linked back to a person (e.g. numbers assigned to individuals tested for AIDS), behavior patterns (e.g. riding a subway each day at 8 a.m.), social categorization (e.g. gender, race, religion, age, class, education, sexual orientation, etc.), or possession of certain knowledge (e.g. secret passwords or keys), or skills (e.g. ability to swim). In news articles, journalists have used different levels of anonymity when referring to their sources. The full anonymity has been usually granted by phrases such as, ‘according to our source that wanted to stay anonymous.’ Partial anonymity, as it will be referred to in this research, has been attributed by the usage of one or several types of identity knowledge determinants when describing a source, such as ‘a company spokesman’ or ‘a senior U.S. government official who wanted to stay anonymous’ (Blankenburg, 1992; Smith, 2007). Although in the academic literature, full and partial anonymity have been studied, the term ‘anonymous sources’ has often referred simultaneously to both of them and has been used interchangeably with the terms ‘unnamed sources’ (e.g. Adams, 1962), veiled sources (e.g. Culbertson, 1975), and even confidential sources (e.g. Strupp, 2005).

Use of anonymous sources and journalistic standards

After Bob Woodward and Carl Bernstein relied on the information provided by anonymous sources to expose Nixon in 1973, researchers have found intensified use of veiled sources in the American media. Hussel's (1996) content analysis discovered that for two years around the Watergate scandal, the instances of attribution to unnamed sources in The Washington Post stories increased significantly. In another study, Culbertson (1978) found that unnamed sources appeared in more than 70% of the stories in newsmagazines Time and Newsweek in 1976. When compared to the 1950s and the 1960s, the decades that preceded the scandal, the use of anonymous sources in the 1970s raised by 17.7% and 14.5%, in The New York Times and The Washington Post, respectively. By 1978, nearly 50% of front-page articles of the two dailies used the technique (Duffy and Williams, 2011). Duffy and Williams (2011) explained that the success of The Washington Post's reporting about the Watergate affair based on accurate statements provided by Deep Throat, encouraged journalists to use the method.

Although isolated studies show that the usage of anonymous sources further increased in the early 1980s (e.g. Wulfemeyer, 1985), researchers mainly agree that the practice dropped after Janet Cooke's fabrication of a heroin addict and the heavily criticized coverage of the O.J. Simpson trail (e.g. Duffy and Williams, 2011; Smith, 2007). In 1988, front-page articles of The New York Times and The Washington Post used veiled sources in 29.5% of cases (Duffy and Williams, 2011). In the following decades, the practice was far from being eradicated but stayed on at significantly lower levels than in the ‘Golden Age’ of journalism. Blankenburg (1992) found that 30% of stories from the LA Times, The New York Times, and The Washington Post, from February 1990 to February 1991 contained words such as ‘source said’ or ‘official said.’ Duffy and Williams's (2011) longitudinal analysis showed that in 2008, the drop in the usage of anonymous sources in the front pages of The New York Times and The Washington Post, hit 26.4%, a level similar to the one measured in decades that preceded the Watergate scandal.

Literature on Chinese media lacks longitudinal examinations of the practice. However, recent research notes significant usage of anonymous sources in news articles in China. The analysis of 100 stories that turned out to be false and that appeared in Chinese media (both traditional and online) between 2001 and 2010 showed that 50% of them did not have named sources (Zhang, 2011). Another study analyzed 40 news stories from a dozen Chinese daily newspapers, as well as from several foreign news agencies, and found that among 164 sources only 41.5% were identified (Guo and Tian, 2007). Practitioners and researchers agree that in China, the technique of unnamed sourcing probably comes from a lack of professional policies and standards. China Youth Daily editor, Shaohua Ma, commented that some media, rushing to get the exclusive news, are willing to publish unverified information and attribute it to anonymous sources (Ma, 2002). Chen (2004) writes that Chinese media do not set specific standards of news reporting in the policies of their news organizations. The author notes that this leads to a lack of information review on the part of journalists and editors. They do not double check unnamed sources and their claims, which cultivates a common culture of frequent and easy anonymous sourcing. In addition, many journalists in China lack professional ethics.

When they encounter sources that are not willing to be identified, they do not try to convince them or attempt to find other sources to replace the anonymous one. On the contrary, they would use the unnamed source to get the news published quickly, but not necessarily correctly. (Chen, 2004: 1).

On the other hand, in the United States, most news organizations have strict standards for the use of anonymous sources. For example, the Associated Press allows the use of anonymous sources only if the information is not elsewhere available, is independently verified, and is authorized by reporter's superiors, to whom the source identity has been revealed. The news agency demands that its journalists explain in their articles why the source requested anonymity and provide attribution that establishes source's credibility (Associated Press, 2013). Stricter sourcing policies have been introduced in the American news organizations after the 1980s scandals and strengthened after 2003, when Jayson Blair, a New York Times reporter, resigned after fabricating quotes, some of which were attributed to unnamed sources. After the later incident, the practice of unnamed sourcing in 16 large circulation newspapers in the United States dropped. In 2003, 35% of articles used at least one unnamed source, whereas in 2004 that was the case with 9% of news articles (Martin-Kratzer and Thorson, 2007).

News credibility

Generally, the concept of news credibility in the United States has been researched in three principle domains: source, message, and medium credibility. Defined as ‘the amount of believability’ attributed to a source of information (Bracken, 2006: 724), source credibility refers to either an individual or an organization sending the message (Sternadori and Thorson, 2009). Receivers are found to be persuaded more by sources they find more credible than by those they view less positively (McCroskey et al., 1974; Pornpitakpan, 2004). Hovland et al. (1953) discovered that in order to be credible the source has to have a certain level of expertise and trustworthiness. The communicator has to be perceived as a source of valid assertions and should have the genuine intent to communicate them. Later, scholars proposed other dimensions of source credibility such as dynamism, composure, and sociability (e.g. Berlo et al., 1970; Markham, 1968; Whitehead, 1968).

Article credibility focuses on the message itself, as in some cases message factors may be more important than source factors (Metzger et al., 2003). For example, recipients turn to message cues when issue involvement, knowledge, and personal relevance are high (Petty and Cacioppo, 1981) or in situations in which little information is available about the source of a message (Petty and Cacioppo, 1988). In media research, characteristics of news stories are found to influence the level of credibility that the audience attributes to them. For instance, Fico et al. (2004) found that balanced or imbalanced story structure influenced the perceived bias, which further influenced perceptions of newspapers credibility.

Scholars have further studied the medium through which the message is delivered and have often, in this regard, compared different media—for example, newspapers to television or new media to traditional media (Golan, 2010). Several analyses indicate that television news was typically deemed more accurate than printed news (e.g. Gaziano and McGrath, 1986; Jacobson, 1969; Westely and Severin, 1964). Newer research, however, shows that newspapers have closed the credibility gap (e.g. Flanagin and Metzger, 2000) and even surpassed television as well as online news in perceptions of credibility (Kiousis, 2001).

Scholars that have studied news credibility in China, have mainly done so by revolving around the concept of the credibility of media or a medium as a source of information. In Western literature, this concept dates back to Gaziano and McGrath's (1986) and Meyer's (1988) studies, from which most definitions and measurements of the concept of news credibility used in contemporary literature originate. Studies dealing with the credibility of Chinese media (e.g. Shen et al., 2011) conceptualize it as the extent to which news media are accurate and trustworthy in their reporting, and the extent to which they serve the public interest. This definition corresponds with Meyer's (1988) two-factor conceptualization of credibility, which includes believability, the extent to which news audiences see a material produced by news media as factual and accurate, and community concern, the extent to which news audiences see news media as being concerned about the community interest.

Several studies have found that, in general, the Chinese attribute high levels of credibility to their media. Liu and Bates (2009) write that two surveys conducted by Ke in 2003 found that media credibility in China reached 85.3% and 91.2%. Even during the SARS outbreak in 2003, when the most negative perceptions of media in China are likely to have occurred, due to the lack of media coverage of the issue, surveys have shown that Chinese have confidence in their media outlets. In a survey conducted by the Research Institute of Public Opinion at Renmin University of China on a sample of citizens of Beijing, two-thirds of the sample had ‘a great deal’ of credibility in media. Another survey, conducted by Tongji University of China during the late period of SARS reported a level of credibility of 83.3% (Liu and Bates, 2009). A more recent study analyzed the data gathered by the China General Social Survey and found that Chinese place a great deal of trust in state media, affording them more trust than to government experts and word-of-mouth communication (Xu, 2012). However, the accuracy of survey data from China has been frequently challenged. High ratings of media credibility largely results from inadequate sampling and question wording (Zhu, 1997) and, more troubling, potential falsification of data for political purposes (He, 2006; Link, 2005). Other studies, however, suggest that credibility levels in China are not that high. A study conducted by Fuzhou GRT Market Information Analysis Company, a professional online survey company in China, found that less than 5% of participants consider Chinese media to be highly credible. One-third of the participants in this study rated Chinese media very low on the credibility scale (Fuzhou GRT Market Information Analysis, 2011). According to Hazelbarth (1997), the Chinese public generally does not trust Chinese media, because they know what they view or hear is controlled by the government. The author states, ‘Chinese public attitudes toward the media historically have been negative because of the media's traditional close identification with the State’ (p. 15).

Researchers agree that the Chinese attribute different levels of credibility to different types of media. According to Chang's (2009) study, Chinese people generally perceive traditional media to be more credible than the Internet. The citizens of three major cities, Beijing, Nanjing, and Wuhan, rated television as the most credible, followed by newspapers, radio, Internet, and magazines (Zhang, 2009a; 2009b; 2009c). Surveys from the city of Shanghai had similar findings. In Liao et al.'s (2007) study, 800 residents of Shanghai rated television news as the most credible. As Liu (2011) explains, Chinese suspicion towards the Internet as the source of news, might come from their uncertainty about website independence, website background, its professional and ethical standards, and policies. Finally, scholars have studied the possible effects of trust in government (Xie, 2004), television news use, Internet use (Shen et al., 2011), age and educational level (Liao et al., 2007; Xu, 2012), on media credibility perceptions of Chinese people.

This study specifically tests the credibility of a news article, and defines it as the public perception of news story quality. This definition corresponds to Sternadori and Thorson's (2009) news credibility, which in turn is similar to Meyer's (1998) concept of believability and likeability, and Chinese concepts of believability and accuracy (Shen et al., 2011).

Anonymous sources and news credibility

Western literature distinguishes between different levels of anonymity when studying the influence of anonymous sources on news credibility. Adams (1962) discovered that unnamed sources that referred to government or to officials tended to be more acceptable than the ones that gave no clues as to who the real source was. Thus, the participants of his study rated information that has been attributed to ‘U.S. government,’ ‘official reports,’ ‘a high governmental official’ higher than those attributed to phrases such as ‘it was learned’ or ‘trustworthy indications show.’

Other studies have found that source attribution is less important to readers than the type of news stories when it comes to perceptions of credibility. Hale (1984) determined that, in factual news stories, the level of anonymity did not have an impact on the perceptions of credibility. Whether the story had no attribution at all, had general attribution, or had specific attribution, did not make a substantial difference in readers' perceptions of credibility of factual news stories. On the other hand, in stories with controversial issues, the change of source attribution has been found to have an effect on readers' evaluations of news stories.

Research also suggests that McCroskey's (1992) third component of credibility, good will that dates back to Aristotle, could also be a more important factor in predicting news credibility. McCroskey (1992) argued that Aristotle recognized the importance of caring as a component of credibility in his discussion of good will of the information source. In this regard, it would be of utmost importance for the source to be viewed as caring if the information coming from it is to be considered as credible. Participants in Smith's (2007) study accepted anonymous sources in a governmental whistle blowing story, but not in the stories that mounted a personal attack on a judge. In the personal-attack story, the sources' accounts attacked judge to be a womanizer, a sexist, and a judge who frequently cancelled court session to attend his private social affairs, whereas in the whistle-blowing story the sources unveiled an alleged cover-up of overcrowding at a county jail. People's credibility judgments may also depend on their desire to learn the identity of the anonymous source (Rains and Scott, 2007), as well as on their previous knowledge of public affairs, and their political orientation (Culbertson and Somerick, 1977).

In Chinese, as in Western literature, the term ‘anonymous source’ refers to a source without attribution (full anonymity) or with only general attribution (partial anonymity). To denote the practice, Chinese articles use ‘’, which in literal translation means ‘unnamed source’ (Zhang, 2011). However, the analyses of the effects of the practice on the perceived credibility among the Chinese population are almost absent. One study used a within-subject experiment to expose 48 participants to two news stories, one with anonymous sources and the other with identified sources. It was found that participants better recalled information from the story that used identified sources than fully anonymous sources (Zhao, 2010). Another study applied the concept to Internet news. Liu (2011) conducted an online survey with representatives of the Chinese population and found that Internet news that used identified public figures as sources of information was seen as more credible than news that cited sources with no names.

Although the above literature shows that news credibility perceptions might differ based on the level of anonymity of sources used in the story, and also on the type of story, good will, as well as some individual characteristics of readers, the overall conclusions suggest that the stories that use any type of anonymous sources would generally be perceived as less credible than stories that use identified sources (e.g. Fedler and Counts, 1981; Liu, 2011; Sternadori and Thorson, 2009; Zhao, 2010). Based on these findings, this study hypothesizes the following:

Hypothesis 1: News stories with only anonymous sources will be perceived as less credible than stories that contain identified sources in both America and China.

While the literature in the United States notes that other story characteristics can, simultaneously, with the use of anonymous sources (Fedler and Counts, 1981; Smith, 2007), influence the perceptions of news credibility, research in China did not explore these possibilities. Thus, it remains unclear what other cues Chinese audiences might use when assessing news credibility. Trying to make a first step in bridging this gap in the literature, and verifying whether these potential cues differ between the audiences in two countries, this study asks the following questions:

RQ1: Is there a significant difference in assessing the credibility of news stories with only anonymous sources between America and China?

RQ2: Is there a significant difference in assessing the credibility of news stories with identified sources between America and China?

Method

To address the above hypothesis and research questions (RQs), a within-subjects experiment, embedded in a paper and pencil survey, was conducted. The independent variable, use of anonymous sources, had two levels: high (stories with only anonymous sources) and low (stories with identified sources). The dependent variable was perceived news story credibility.

Participants

The participants were 620 undergraduate students: 322 from a large public Southeastern university in the United States and 298 from a large urban public university in China. The students were recruited from undergraduate classes from schools of communication and information in both countries and were not offered any incentives for the participation in the study. In both countries, the participants reported to classrooms to take the paper and pencil survey. On average, participants were 19.68 (SD = 1.59) years old and in their first or second year of university attendance (M = 1.86, SD = 0.99). United States' participants were 19.83 (SD = 1.96) years old on average and in their second year at school (M = 2.10; SD = 1.20), whereas Chinese participants were 19.52 (SD = 0.98) years old on average and in their first year (M = 1.58; SD = 0.57) at the university. Females comprised 64.8% of the sample with males making up 33.2% of the sample (the rest were missing data). In the United States, females comprised 63.4% of the sample with males making up 36.0% of the sample. In China, females comprised 66.4% of the sample and 30.2% of the sample.

Stimuli

The stimuli consisted of two news stories adapted from the Associated Press news stories from 12 January 2011 to 27 January 2011. The stories presented to American students were written in English, whereas the stories presented to Chinese students were translated into Chinese. On the stimuli itself, news organization or the journalist were not indicated in order to avoid possible perception biases. The first news story described anti-government protests in Tunisia, whereas the second story described anti-government protests in Egypt. The stories were very similar in order to yield adequate comparison. They both were the same length (244 and 240 words), dealt with the same topic (Arab Spring), and had the same structure (first paragraph described where and when the protests occurred, who participated, and against whom they were directed; the next two paragraphs described the consequences of protests and introduced a direct quote; the last two paragraphs provided more information about developments of the protests). The difference was that the first story relied only on anonymous sources as sources of information (high use of anonymous sources treatment), whereas the second story used named sources (low use of anonymous sources treatment). In the first story, both levels of anonymity, full and partial, were used. Anonymous sources were referred to as ‘a protestor who wanted to stay anonymous,’ ‘unnamed protestor,’ ‘the protestor who wanted to keep his identity secret,’ and ‘our anonymous source.’ In contrast, the information in the second story was attributed to named sources such as ‘our correspondent from Cairo, John Hendawi,’ ‘businessman Said Abdel-Motalib,’ or ‘the Interior Ministry.’ Factual news stories were chosen in order to make the experiment more conservative as research has shown that the readers would be less likely to make distinctions in credibility assessments when anonymous sources are used in factual news stories than in investigative news stories (Fedler and Counts, 1981; Hale, 1984). Possible motivation bias was addressed by choosing topics outside of participants' countries.

Procedure

Each participant read both stories, one with anonymous sources and the other one with identified sources, and completed a 13-item measure, after each reading, to assess their credibility. The participants also completed a battery of demographic items and three other measures that were not included in this study. The questionnaires in the United States were administered by the authors of this study whereas the questionnaires in China were administered by a communication professor. In the United States, the data were collected in January 2012 whereas in China, the data were collected in February 2012. Once the questionnaires in China were completed, they were shipped to the authors in the United States where they were, together with the United States questionnaires, entered into SPSS for further analysis.

A total of 637 questionnaires were collected from both countries (328 from the United States and 309 from China). Cases with more than 10% missing data—six from the United States and six from China—were deleted (Hair et al., 2010). Five more cases from China were eliminated due to extreme response bias (respondents who answered all 7 s or all 1 s on the questionnaires). There were remaining 31 missing values in the Chinese surveys. The Little's MCAR test obtained for this study resulted in a Chi Square = 427,762 (df = 438, p = .63), which indicated that the data were missing completely at random (i.e. no identifiable pattern existed to the missing data). The range of missing data per variable was from 0.3% to 5.4%. Thus, all missing data percentages fell below the 20% exclusionary score established. As the missing data were established to be MCAR, the imputation technique of the mean substitution was employed to replace the missing values for all variables except for the demographics—which, not being pertinent to answering the RQs, were ignored. The final sample consisted of 620 participants, 322 from the United States and 298 from China.

Dependent variable

News story credibility

News story credibility was measured by a 13-item scale developed by Sternadori and Thorson (2009). Three items were dropped due to lower factor loadings in both treatments (λ = .430, λ = .488, and λ = .552 for treatment with high anonymous sources and λ = .624, λ = .645, and λ = .647 for treatment with low anonymous sources). The remaining 10 items were: (1) The story is fair; (2) The story is complete; (3) The story is accurate; (4) I trust the information in this story; (5) I trust the sources quoted in this story; (6) The story is believable; (7) The story is credible; (8) The story is informative; (9) I liked this story; (10) The story is well written. For this 10-item scale, we measured α = .886 for treatment with high anonymous sources and α = .929 for treatment with low anonymous sources. Reponses were given on a 7-point Likert scale ranging from one (strongly disagree) to seven (strongly agree). When testing hypothesis and answering RQs, the differences between groups were calculated on the basis of the means of the items constituting the dependent variable news story credibility.

Results

Establishing cross-cultural equivalence

Before testing the hypothesis and answering the RQs, the issue of cross-cultural equivalences was addressed. Comparative studies across different cultures can only be legitimate if equivalent bases of comparison were first established (Lonner, 1979). Following the suggestions from the cross-cultural literature (e.g. Berry, 1979; Gudykunst, 2002; Harkness, 2010; Mastsumoto and van de Vijver, 2011), we addressed functional, conceptual, linguistic, sample, response category, and measurement equivalence. Functional equivalence refers to the similarity of goals between two behaviors. For example, two acts of aggression are functionally equivalent if people of both cultures emit such behaviors in certain situations (Hui and Triandis, 1985). In this study, the literature review showed that people of both cultures, Chinese and American, when being exposed to media, attribute a certain level of credibility in them, their journalists and their stories (e.g. Fico, Richardson, and Edwards, 2004; Flanagin and Metzger, 2000; Gaziano and McGrath, 1986; Hazelbarth, 1997; Liu and Bates, 2009; Xie, 2004).

Conceptual equivalence focuses on the presence (or absence) of meanings that individuals attach to specific concepts: the meaning of the concepts under study should be the same across cultures under study (Gudykunst, 2002). In other words, the conceptual definitions of the constructs under study need to be constant in each of the cultures being compared (Levine et al., 2007).The news credibility concept as well as the concept of anonymous sourcing were reviewed carefully in the literature review section to establish conceptual equivalence in two culturally different societies. It was established that, in both cultures, news credibility refers to the amount of believability attached to a source, message or a medium (e.g. Meyer, 1988; Shen et al., 2011; Sternadori and Thorson, 2009), and that anonymous sources refer to the sources that appear in news articles without the attribution or with only general attribution (e.g. Blankenburg, 1992; Smith, 2007; Zhang, 2011). In addition, equivalence of meaning was tested using factor analytical procedure. It was assumed that if two constructs mean the same thing across different cultures, they will have a similar factorial structure. The 10 items of the News Story Credibility scale were subjected to principal components analysis (PCA) using SPSS 20. Prior to performing PCA, the suitability of data for factor analysis was assessed. Inspection of the correlation matrix revealed that all coefficients were of .341 and above in the Chinese sample and of .402 and above in the American sample. The Kaiser–Meyer–Oklin value was .928 for the American sample and .926 for the Chinese sample, exceeding the recommended value of .6 (Kaiser, 1970, 1974); Bartelett's Test of Sphericity (Bartlett, 1954) reached statistical significance in both samples, supporting the factorability of the correlation matrix. PCA revealed the presence of one component with eigenvalues exceeding 1 in both samples, explaining 60.272% of the variance in the American sample and 58.051% of the variance in the Chinese sample. Inspection of the screen plots revealed a clean break after the first component in both samples. The inspection of the component matrices shows that all the items load quite strongly on one factor (American sample: from λ = .670 to λ = .849; Chinese sample: from λ = 632 to λ = 823).

Linguistic equivalence was established by the technique of back-translation (Gudykunst, 2002). Two bilinguals fluent in both Chinese and English, assisted. One of them translated the English stimulus and questionnaire to Chinese and the other back-translated them. The variations in original wording were reconciled. Following Gudykunst's (2002) suggestions, extreme scores wording, such as strongly agree or strongly disagree, were avoided in the Chinese version of questionnaires and replaced with moderate scores wording in order to establish response category equivalence. Research suggests that Asians do not use extreme scores (e.g. strongly agree or strongly disagree) as much as Americans and can therefore score lower because of response tendencies and not because of cultural differences (e.g. Chen et al., 1995; Peng et al., 1997). The sample equivalence was addressed by choosing samples with similar demographic characteristics. By conducting the survey among undergraduate students at public universities in China and the United States, the similarities in participants' age and educational level were established as the data show.

Finally, to test the cross-cultural measurement equivalence of the News Story Credibility scale, multigroup metric invariance was tested using AMOS 20. The groups were formed based on participants ethnicity (American, N = 320, and Chinese, N = 298). Analysis was first conducted on the scale used in the high anonymous sources treatment. Computation of the Chi Square difference between unconstrained model and measurement weights model reveals $Δ χ_{(9)}^{2} = 28.75$ , significant at p < .001. Taking into account that this χ² difference is statistically significant at the probability of less than .05 we could conclude that one or more factor loadings is not operating equivalently across groups. However, researchers have argued that this χ² difference value is an impractical and unrealistic criterion upon which to base evidence of equivalence, and suggested moving towards basing comparative models on the difference between the CFI values (Byrne, 2008). Computation of the CFI difference between unconstrained model and measurement weights model shows ΔCFI = .007 (CFI for unconstrained model = .871; CFI for measurement weights model = .864). Taking into account Cheung and Rensvold's (2002) criterion, who arbitrarily suggested a CFI difference value not exceeding .01, we can conclude that factor loadings for the News Story Credibility scale measured in the high anonymous sources treatment are operating equivalently across groups (American and Chinese). Second, we conducted the analysis on the scale used in the low anonymous sources treatment. Computation of the Chi Square difference between the unconstrained model and the measurement weights model reveals $Δ χ_{(9)}^{2} = 50.748$ , significant at p < .001, whereas CFI difference was ΔCFI = .01 (CFI for unconstrained model = .925; CFI for measurement weights model = .913). It can be concluded that factor loadings for the News Story Credibility scale measured in the low anonymous sources treatment are operating equivalently across groups (American and Chinese). Once those cross-cultural equivalences were established, we proceeded to testing the hypothesis and answering the RQs.

Testing hypothesis and answering RQs

The first hypothesis stated that stories with only anonymous sources will be perceived as less credible than stories with identified sources in both America and China. This hypothesis was supported. Paired samples T-test showed that there was a significant difference in News Story Credibility scores between stories with anonymous sources (M = 4.02, SD = .960) and stories with identified sources (M = 5.04, SD = .938) in the American sample, t (321) = –14.752, p < .0001 (two-tailed). The effect size was large, η² = .40, following the criteria proposed by Cohen (1988). This indicates that participants from the United States perceived stories with only anonymous sources as less credible than stories with identified sources. In the Chinese sample, paired samples T-test also showed that there was a significant difference in News Story Credibility scores between stories with anonymous sources (M = 3.73, SD = 1.00) and stories with identified sources (M = 4.37, SD = .992), t (297) = –9.056, p < .0001 (two tailed). The effect size was large, η² = .22. This indicates that participants from China also perceived stories with only anonymous sources as less credible than stories with identified sources (Table 1).

Table 1.

Difference in news story credibility scores between stories with anonymous sources and stories with identified sources.

Country of origin		M (SD)	ΔM	95% CI	t	df	p	η ²
United States	Anonymous	4.02 (.96)	−1.03	(−1.16, −.89)	−14.752	321	.000	.40
United States	Identified	5.04 (.936)	−1.03	(−1.16, −.89)	−14.752	321	.000	.40
China	Anonymous	3.73 (1.00)	−.64	(−.78, −.50)	−9.056	297	.000	.22
China	Identified	4.37 (.992)	−.64	(−.78, −.50)	−9.056	297	.000	.22

CI: confidence interval.

The first RQ asked if there was a significant difference in assessing the credibility of news articles with only anonymous sources between America and China. An independent T-test was conducted to compare News Story Credibility scores for two countries. There was a significant difference in scores for America (M = 4.02, SD = .960) and China (M = 3.73, SD = 1.00); t (618) = 3.571, p < .0001 (two-tailed), although the effect size was small η² = .02. This indicates that American students assess stories that use only anonymous sources as more credible than Chinese students (Table 2).

Table 2.

Differences in perceived news story credibility between America and China.

Type of story		M (SD)	ΔM	95% CI	t	df	p	η ²
Anonymous sources	America	4.02 (.96)	.28	(.13, .44)	3.571	618	.000	.02
Anonymous sources	China	3.73 (1.00)	.28	(.13, .44)	3.571	618	.000	.02
Identified sources	America	5.04 (.938)	.67	(.52, .82)	8.660	618	.000	.11
Identified sources	China	4.37 (.992)	.67	(.52, .82)	8.660	618	.000	.11

CI: confidence interval.

The second RQ asked if there was a significant difference in assessing the credibility of news articles with identified sources between America and China. An independent T-test showed that there was a significant difference in News Story Credibility scores for America (M = 5.04, SD = .938) and China (M = 4.37, SD = .992); t (618) = 8.660, p < .0001 (two-tailed). The effect size was reaching the large margin η² = .11. This indicates that American students assess stories that use identified sources as more credible than Chinese students (Table 2).

Discussion

This study tested the impact of the use of anonymous sources on perceived news story credibility in two countries with assumed different journalistic standards—America and China. The study exposed audiences in both countries to a within-subjects experimental design, in which each participant read two similar news stories—one that only used anonymous sources and the other that used identified sources—and filled out the questionnaire, which tested the perceived news story credibility.

The major finding indicates that both Americans and Chinese rated news stories that used only anonymous sources as less credible than stories that used identified sources. This finding supports the previous literature, which found that the use of anonymous sources has a negative impact on perceived news story credibility with American audiences (e.g. Sternadori and Thorson, 2009) and with Chinese Internet audiences (Liu, 2011). Our study furthers the credibility research in several areas. In general, credibility research on Chinese audiences is scarce. Several studies that have appeared sporadically have either compared the credibility of different media (e.g. Chang, 2009; Liao et al., 2007; Liu, 2011; Zhang, 2009a) or explored demographic and attitude variables that could influence credibility perceptions (e.g. Liao et al., 2007; Shen et al., 2011; Xie, 2004). Those that dealt with the influence of message characteristics on perceived credibility, have tested the impact of the use of anonymous sources on information recall (Zhao, 2010) and on perceived credibility of Internet news stories (Liu, 2011). By exposing the participants of this experiment to a written news story that could be easily used on multiple media platforms, this study expands the research on the impact of anonymous sources on Chinese audiences beyond the Internet area.

Furthermore, our findings show that some message characteristics, such as the use of anonymous sources, represent a credibility cue across cultures. Although previous literature has demonstrated that the use of anonymous sources has a negative effect on credibility perceptions in both America and China (e.g. Liu, 2011; Sternadori and Thorson, 2009), these findings were discovered in separate studies, using different samples, different time frames, and different stimuli. This study represents a comparative research that tests the influence of different types of sourcing on credibility perceptions using the same stimulus and questionnaire with carefully established functional, conceptual, linguistic, response category, and measurement equivalences on equivalent samples in America and China. Using methods to establish cross-cultural validity, this study demonstrated that, with both American and Chinese audiences, the use of anonymous sources trigger suspicion about accuracy, fairness, trustworthiness, believability, and the overall quality of a news story. Cross-cultural psychology has noted that in order to understand social behavior, we need to account for both individual and cultural variability (e.g. Smith et al., 2013). If both American and Chinese audiences consider that unnamed sources diminish the credibility of news stories, this might indicate that attitudes towards anonymous sources could be widespread, and might not depend on the immediate social context, such as journalistic standards and practices. Lacking persuasive value, anonymous sources may be perceived, around the globe, as less competent or credible than they would be if they were identified. Receivers may feel that anonymity releases senders from accountability for their contributions. The attitudes towards anonymous sources might then be more fundamental to the human nature and less malleable and likely to emerge in a different form, depending on the ways in which particular individuals are socialized in nations around the world. Future studies should expand this comparison to other nations and cultures, taking into account both individual and cultural differences, and use larger samples, with the ultimate goal of testing whether the use of anonymous sources is indeed a universal cue for news credibility.

This study also found that the general attitude of American students towards news stories is more positive than the attitude of Chinese students. Although recent polls show that trust and credibility perceptions of news media in the United States have substantially fallen in the last decades (Pew Research Center, 2011, 2012), the fact is that the First Amendment guarantees the right to publish freely and that American media are generally independent from the government and their journalistic professionalism is strongly developed (Hallin and Mancini, 2004). On the other hand, the totalitarian political regime in China has made the news media a tool of propaganda and censored their content (Hazelbarth, 1997; Xie, 2004). The findings that show that Americans rated both news stories that use anonymous sources and stories that use identified sources higher in terms of credibility than Chinese did, might go in line with the research that suggests that people who live in countries with high censorship are likely to be more critical of news than people who believe their countries have a free and objective press. Testing the information processing strategies between students from America and students from Indonesia, Gunther and Snyder (1992) found that Indonesians were more critical readers; they paid more attention to the actual source of the news and varied accordingly the strategies of attributing the story to the writer, and judging the story to be biased and distrusting it.

The main limitation of this study is its sample. Taking into account that the study used communication students from both countries, the results are not necessarily generalizable to entire populations. Another one is that, although using similar samples in terms of age and level of education, this study did not take into account other individual differences that might appear as moderating variables in assessing perceptions of news story credibility. Future studies should use larger samples in a bigger number of countries and multilevel modeling to account for cultural and individual differences in predicting news credibility. However, as this study represents the first attempt to test credibility cues across cultures using experimental and cross-cultural methods, it sets a solid ground for future studies, which could further the testing of the impact of the use of anonymous sources on credibility perceptions.

References

Adams

(1962) The relative credibility of 20 unnamed sources. Journalism Quarterly 39: 79–82.

Associated Press (2013) AP news values and practices. Available at: www.ap.org/company/news-values (accessed 20 January 2013).

Bartlett

(1954) A note on the multiplying factors for various chi square approximations. Journal of Royal Statistical Society 16(Series B): 296–298.

Berlo

Lemert

Mertz

(1970) Dimensions for evaluating the acceptability of message sources. Public Opinion Quarterly 33: 563–576.

Berry

(1979) On cross-cultural comparability. International Journal of Psychology 4: 119–128.

Blankenburg

(1992) The utility of anonymous attribution. Newspaper Research Journal 13(1/2): 10–23.

Bracken

(2006) Perceived credibility of local television news: The impact of television form and presence. Journal of Broadcasting and Electronic Media 50: 723–741.

Byrne

(2008) Testing for multigroup equivalence of a measuring instrument: A walk through the process. Psicothema 20: 872–882.

Chan

Lee

FLF

Pan

(2006) Online news meets established journalism: How China's journalists evaluate the credibility of news websites. New Media and Society 8: 925–947. .

10.

Chang

(2009)

(Analysis of internet credibility in China). Today's Mass Media 11: 95–96. .

11.

Chen

Lee

Stevenson

(1995) Response style and cross-cultural comparisons of rating scales among East Asians and North American students. Psychological Science 6: 170–175.

12.

Chen

(2004)

(Analysis of abuse of anonymous sources by media). The Journalist Monthly. Available at: http://xwjz.eastday.com/eastday/xwjz/node27685/node27687/userobject1ai408177.html.

13.

Cheung

Rensvold

(2002) Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling 9: 233–255.

14.

Cohen

(1988) Statistical Power Analysis for the Behavioral Sciences, Hillsdale, NJ: Lawrence Erlbaum Associates.

15.

Culbertson

(1975) Veiled News Source – Who and What are They?, Washington, DC: ANPA News Research Center Study.

16.

Culbertson

(1978) Veiled attribution – An element of style? Journalism Quarterly 55(3): 456–465.

17.

Culbertson

Somerick

(1977) Variables affect how persons view unnamed news sources. Journalism Quarterly 54(1): 58–69.

18.

Duffy

Freeman

(2011) Unnamed sources: A utilitarian exploration of their justification and guidelines for limited use. Journal of Mass Media Ethics 26(4): 297–315.

19.

Duffy

Williams

(2011) Use of unnamed sources drops from peak in 1960s and 1970s. Newspaper Research Journal 32(4): 6–21.

20.

Fedler

Counts

(1981) Variations in attribution affect readers' evaluations of stories. Newspaper Research Journal 2(3): 25–34.

21.

Fico

Richardson

Edwards

(2004) Influence of story structure on perceived story bias and news organization credibility. Mass Communication and Society 7: 301–318.

22.

Flanagin

Metzger

(2000) Perceptions of Internet information credibility. Journalism and Mass Communication Quarterly 77: 515–540.

23.

Fuzhou GRT Market Information Analysis (2011) Analysis of media credibility. Available at: www.blog.cntv.cn/13665226-1574859.html (accessed 20 January 2013).

24.

Gaziano

McGrath

(1986) Measuring the concept of credibility. Journalism Quarterly 63: 451–462.

25.

Golan

(2010) New perspectives on media credibility research. American Behavioral Scientist 54(1): 3–7.

26.

Gudykunst

(2002) Issues in cross-cultural communication research. In: Gudykunst

Mody

(eds) International and Intercultural Communication, Thousand Oaks, CA: SAGE, pp. 165–182.

27.

Gunther

Snyder

(1992) Reading international news in a censored press environment. Journalism and Mass Communication Quarterly 69: 591–599.

28.

Guo

Tian

(2007) Analysis of sources and quoting verbs in news stories. Contemporary Communications 6: 142–144.

29.

Haiman RJ (1998) Hey, editors: Just stop the nonsense. Newspaper Research Journal 19: 2–9.

30.

Hair JF, Black WC, Babin BJ and Anderson RE (2010) Multivariate data analysis: A global perspective. Upper Saddle River, NJ: Pearson Education, Inc.

31.

Hale

(1984) Unnamed sources: Their impact on the perceptions of stories. Newspaper Research Journal 5(2): 49.

32.

Hallin

Mancini

(2004) Comparing Media Systems: Three Models of Media and Politics, New York, NY: Cambridge University Press.

33.

Harkness

(2010) Survey Methods in Multinational, Multiregional, and Multicultural Contexts, Hoboken, NJ: Wiley.

34.

Hazelbarth

(1997) The Chinese Media: More Autonomous and Diverse-within Limits: An Intelligence Monograph. Washington, DC: Center for the Study of Intelligence, Central Intelligence Agency. .

35.

(2006) A survey study on voting behavior and political participation in Zhejiang. Japanese Journal of Political Science 7: 225–250.

36.

Hovland

Janis

Kelley

(1953) Communication and Persuasion: Psychological Studies of Opinion Change, New Haven, CT: Yale University Press.

37.

Hui

Triandis

(1985) Measurements in cross-cultural psychology. Journal of Cross-Cultural Psychology 16(2): 131–152.

38.

Hussel

(1996) Before Watergate, beyond Jimmy's world. In: The annual convention of the association for education in journalism and mass communication, Anaheim, CA17 December 1996.

39.

IREX (2008) Media sustainability index Iran. Available at: www.irex.org/system/files/MSIMENA09_Iran.pdf (accessed 24 January 2013).

40.

IREX (2010) Media sustainability index Africa. Available at: www.irex.org/project/media-sustainability-index-msi-africa (accessed 24 January 2013).

41.

IREX (2011) Media sustainability index Middle East and North Africa. Available at: www.irex.org/project/media-sustainability-index-msi-middle-east-north-africa (accessed 24 January 2013).

42.

Jacobson

(1969) Mass media believability: A study of receiver judgments. Journalism Quarterly 46: 20–28.

43.

Kaiser

(1970) A second generation Little Jiffy. Psychometrika 35: 401–415.

44.

Kaiser

(1974) An index of factorial simplicity. Psychometrika 39: 31–36.

45.

Kiousis

(2001) Public trust or mistrust? Perceptions of media credibility in the information age. Mass Communication and Society 4: 381–403.

46.

Levine

Park

Kim

(2007) Some conceptual and theoretical challenges for cross-cultural communication research in the 21^st century. Journal of Intercultural Communication Research 36: 205–221.

47.

Liao

Zhang

(2007) Analysis of news credibility in China. Chinese Journalism Quarterly 4: 66–73. .

48.

Link

(2005) China wiping out the truth. New York Review of Books 52(3): 36–39. .

49.

Liu

(2011) Credibility Analysis of Internet News in China, Wuhan: Huazhong University of Science and Technology.

50.

Liu

Bates

(2009) What's behind public trust in news media: A comparative study of America and China. Chinese Journal of Communication 2(3): 307–329.

51.

Lonner

(1979) Issues in cross-cultural psychology. In: Marsella

Tharp

Cibrowski

(eds) Perspectives in Cross-Cultural Psychology, New York, NY: Academic Press, pp. 17–45.

52.

Ma S (2002) Anonymous sources harm media credibility. Available at: zqb.cyol.com/content/2002-11/26/content_569793.htm (accessed 25 January 2013).

53.

Markham

(1968) The dimensions of source credibility of television newscasters. Journal of Communication 18: 57–64.

54.

Martin-Kratzer

Thorson

(2007) Use of anonymous sources declines in U.S. newspapers. Newspaper Research Journal 28(2): 56–70.

55.

Marx

(1999) What is in a name? Some reflections on the sociology of anonymity. The Information Society 15: 99–112.

56.

Mastsumoto

van de Vijver

FJR

(2011) Cross-Cultural Research Methods in Psychology, New York: Cambridge University Press.

57.

McCroskey

(1992) An Introduction to Communication in the Classroom, Edina, MN: Burgess International.

58.

McCroskey

Hamilton

Weiner

(1974) The effect of interaction behavior on source credibility, homophily, and interpersonal attraction. Human Communication Research 1: 42–52.

59.

Metzger

Flanagin

Eyal

(2003) Credibility for the 21^st century: Integrating perspectives on source, message, and media credibility in the contemporary media environment. In: Kalbfleisch

(ed.) Communication Yearbook 27, Mahwah, NJ: Lawrence Erlbaum Associates, pp. 293–335.

60.

Meyer

(1988) Defining and measuring credibility of newspapers: Developing an index. Journalism Quarterly 65: 567–588.

61.

Peng

Nisbett

Wong

(1997) Validity problems comparing values across cultures and possible solutions. Psychological Methods 2: 329–344.

62.

Petty

Cacioppo

(1981) Attitudes and Persuasion: Classic and Contemporary Approaches, Dubuque, IA: William C Brown.

63.

Petty RE and Cacioppo JT (1988) The elaboration likelihood model of persuasion. Advances in Experimental Social Psychology 19: 123–203.

64.

Pew Research Center (2011) Press widely criticized, but trusted more than other information sources. Available at: people-press.org (accessed 20 January 2013).

65.

Pew Research Center (2012) Further decline in credibility ratings for most news organizations. Available at: www.people-press.org/2012/08/16/further-decline-in-credibility-ratings-for-most-news-organizations/ (accessed 20 January 2013).

66.

Pornpitakpan

(2004) The persuasiveness of source credibility: A critical review of five decades' evidence. Journal of Applied Social Psychology 32(2): 243–281.

67.

Rains SA and Scott CR (2007) To identify or not to identify: A theoretical model of receiver responses to anonymous communication. Communication Theory 17: 61–91.

68.

Shen

Guo

(2011) News media use, credibility, and efficacy: An analysis of media participation intention in China. Chinese Journal of Communication 4(4): 475–495.

69.

Shepard

(1994) Anonymous sources. American Journalism Review, 19–25. .

70.

Smith

(2007) Impact of unnamed sources on credibility not certain. Newspaper Research Journal 28(3): 8–19.

71.

Smith

Fischer

Vignoles

(2013) Understanding Social Psychology across Cultures: Engaging with other in a Changing World, Thousand Oaks, CA: SAGE.

72.

Sternadori

Thorson

(2009) Anonymous sources harm credibility of all stories. Newspaper Research Journal 30(4): 54–66.

73.

Strupp

(2005) Everyone favors more careful use of anonymous sources-but is the current crackdown going too far? Losing confidence. Editor and Publisher 138(7): 32–39.

74.

Westely

Severin

(1964) Some correlates of media credibility. Journalism Quarterly 41: 325–335.

75.

Whitehead

(1968) Factors of source credibility. Quarterly Journal of Speech 54: 59–63.

76.

Wulfemeyer

(1985) How and why anonymous attribution is used by Time and Newsweek. Journalism Quarterly 62(1): 81–86.

77.

Xie

(2004) Belief consistency: Exploring a balance among trust in government, perceived news media credibility and media use. Master's Thesis, Washington State University, USA.

78.

(2012) Trust in Chinese state media: The influence of education, Internet, and government. Journal of International Communication 18: 1–16.

79.

Zhang H (2009a) Media credibility in China: Survey in Wuhan. Available at: media.ifeng.com/news/200910/1009_4009_1380647.shtml (accessed 15 January 2013).

80.

Zhang H (2009b) Media credibility survey in Nanjing: Xinhua.net highest among web sites. Available at: news.xinhuanet.com/newmedia/2009-10/21/content_12288499_1.htm (accessed 15 January 2013).

81.

Zhang H (2009c) Media credibility in China: Survey in Beijing. Available at: finance.sina.com.cn/leadership/myxcl/20090716/17416490235.shtml (accessed 15 January 2013).

82.

Zhang TF (2011) «Content analysis of Shanghai journalism review 2001 to 2010 Top 10 fake news. Available at: news.xinhuanet.com/zgjx/2011-04/06/c_13815607.htm (accessed 15 January 2013).

83.

Zhao

(2010) An experiment and thoughts about anonymous sources. News and Writing, 6.

84.

Zhu

(1997) Antecendens, covariates, and outcomes of media crediblity in China. Mass Communication Review 24(3/4): 4–17.