Responses to News Overload in a Non-Partisan Environment: News Avoidance in China

Studies conducted in Western democratic countries with privately owned press presenting news from different ideological perspectives have found the phenomenon of news overload causes stress in media users (affective load) and this leads to news avoidance behavior. This study, conducting a structural equation modeling using AMOS 27, investigates the relationships between news overload, affective load, and consequent news avoidance behavior in China. The findings suggest, consistent with studies in Western democratic jurisdictions, that news overload and affective load influence users’ news avoidance behavior. However, in contrast to studies outside of China, this study found demographic variables of age, gender, education, and occupation had no statistically significant influence on avoidance behavior by Chinese social media news consumers. The results suggest that either social attributes and responses by media users in China are significantly different from those in Western countries or ideology plays an important role in avoidance behavior and the impact of ideology is different across different demographic groups in Western jurisdictions.


The Unbearable Weight of Limitless Information
The internet is without doubt the most valuable instrument in the dissemination of information in global history, its reach and impact growing exponentially and the proliferation of devices that provide access to the internet greatly amplifying its reach.For some, access to limitless information is a joyful windfall: sports fans can feast on continuous sports broadcasting and news junkies can feed their fix without breaks.For most, however, relentless pummeling with new information is overwhelming and a self-defense response is inevitable.The most common response by those feeling overwhelmed by information overload is the creation of an information cocoon-finding ways to limit available information to sources and subjects inside the protective cocoon.A subset of information cocoons involves the news avoidance phenomenon-individuals limiting their access to news by limiting their access to news sources.
News has in the past played, and continues in the present moment to play, a central role in molding citizens' views of the world, particularly in respect of issues extending beyond their personal experience (Patterson, 2008;Zhu & Krever, 2017).The news environment for media users today is, however, fundamentally different from that enjoyed by their predecessors in past decades (Clark & Marchi, 2017): the expanded capacity of the internet, the sophistication of smartphones, the rise of social media platforms and constant news updates have multiplied many-fold the volume of news to which people may be exposed on a daily basis (Matthes et al., 2020;Schmitt et al., 2018).Social media platforms in particular have opened the door to usergenerated news sources that now account for a significant amount of the news stories obtained by many consumers (Huang & Wang, 2021).While outside news sources may largely be filtered out for Chinese media users by the country's ''Great Firewall,'' the amount available remains far greater than any person could possibly access in the time available (Hjelle, 2017).
The study of information avoidance and cocoons in broad terms, and news avoidance behavior in more specific terms, is a well-trodden path in the media studies field (Gossart, 2014;Guo et al., 2020;Sunstein, 2006).It remains a particulary challenging area of study, however, because of two competing explanations for avoidance behavior, that might be broadly labeled the reactive and proactive explanations for avoidance.The former envisages avoidance as a reaction to news volume-the virtually unlimited volume of news available, even if a user relies only on a single internet or social media source or platform, can be overwhelming and users react by adopting a news avoidance technique.The latter encompasses instances where users actively avoid news based on content rather than volume.
Thus, for example, some media users may find the inherently negative content of news generally-disasters, war, famine, suffering, and similar subjects being treated as newsworthy topics-depressing, leading them to establish broad cocoons filtering out almost all news (Kalogeropoulos, 2017;Schrøder, 2016).Others might have limited and parochial interests, prompting them to ignore or block all news outside their particular areas of interest such as sports, entertainment, or the environment.For others still, ideological shutters may prompt news consumers to reject news sources with a different political perspective from their own (Dubois & Blank, 2018;Marks et al., 2019), leaving only sources that provide news that is psychologically and ideologically comforting to them (Janssen & de Poot, 2006;Sunstein, 2006).And finally, there is a group who are simply not interested in the news and limit exposure not by adopting any blockage or cocoon behavior but instead by ignoring the available sources (Blekesaune et al., 2012).
The same user interests and disinterests that prompt proactive news avoidance-avoidance to limit consumption to subjects of interest or to avoid subjects of disinterest-might be used as filtering criteria for the media users who avoid news in a reactive fashion, persons overwhelmed with the volume of news they face and who adopt a news avoidance tactic as a reaction to negative feelings from the overwhelming news load they would otherwise face.This group of news avoiders is the subject of the current study and the data collection process was carefully designed to identify respondents in this group.
A useful framework for the analysis of news overload and responses is the stimulus-organism-response (S-O-R) model.This model maps avoidance behavior as a threestage process: a stimulus (a lot of news) triggers a negative reaction or affective load in a media user (the organism) such as anxiety and the person responds in a selfprotection manner by limiting news input.The study is complicated, however, by several factors.First, a study of S-O-R in the context of news avoidance needs to exclude those instances of proactive news avoidance where media users restrict their exposure to the news because of disinterest in the news or interest in only a limited number of particular news subjects.The framework is only relevant to persons adopting news avoidance tactics as a reaction to negative feelings from news overload.
The study distinguishes itself from previous studies of news overload and news avoidance by focusing on the phenomenon in China, a jurisdiction with a significant feature that distinguishes its news environment from that of Western democratic nations in which an unconstrained press adopts positions across the ideological spectrum.In the context of often highly politically polarized environments, the role of ideology as a filter is significant and is likely to be the primary filter for news avoidance behavior (Cargnino & Neubaum, 2021).While prior studies have sought through regression analysis to identify a range of factors that might impact on news avoidance behavior such as age, gender, education, or occupation, it is almost impossible to segregate from any findings the overarching impact of ideology in societies with significant political and ideological polarization.
In contrast, this article reports on the findings of a study that seeks to carve out the impact of ideology on news avoidance by measuring the S-O-R phenomenon in China, where all media is state owned and all news adopts a similar ideological stance.In particular, this study tests the same hypotheses that would be examined in a similar study in a Western democracy on the assumption that relationships between factors other than ideology might be apparent in the same way in the context of news avoidance in China.On the other hand, if the hypotheses do not hold up to expectations, the results would suggest that either the cultural and social environment in China is so fundamentally different from that in the West that conventional Western expectations of responses to media cannot be transplanted to China, or that ideological divides are playing a larger role than recognized in Western societies and their impact affects all other independent variables that might influence media behavior.

Sina Weibo and the Media Environment in China
This study investigates the phenomenon of news avoidance by users of an internet platform in China.Like its Western counterparts, China has a number of internet platforms.With different features, these do not correspond exactly with their Western cousins.For example, WeChat, the largest communication service otherwise similar to WhatsApp, also contains groups and postings, similar to the combined Messenger and Facebook system.Douyin, the Chinese counterpart to TikTok, is almost identical to its Western counterpart, developed by the same company and essentially renamed for the market outside China.Sina Weibo, the platform used in this study, is a microblogging site featuring postings by news sources and users, with room for comments.It is owned by a listed technology company and is the largest blog site in China with over half the country's 1 billion internet users.
The primary news sources in China, radio, television, and the press, are state owned and, as might be expected, all follow the state's ideological perspective.The media is by no means homogenous, however.In addition to media owned by the national government, there are services owned by provincial, municipal, and even county governments with significant differences in content across stations and publications (Qin et al., 2018).While the different media outlets tend to project a consistent and supportive voice where the government has made its unified position on an issue clear, where there is no single government position and different government ministries have different views on an issue, parallel differing views are reflected in media coverage of the issue.In one sense, the differences in views represent different ideological perspectives, just as different viewpoints are held by different persons in different parts of the government.
At the same time, however, there is no partisan ideological divide as is common in Western media and there are no partisan news sources strongly supporting or deeply critical of the government of the day, its actions, and its leaders.Even the most critical outlets take care not to use language critical of the government or government positions (Duckett & Langer, 2013).The left-right divide found in many Western jurisdictions is simply not a feature of Chinese politics, society or media and the common Western phenomenon of a divided media with half openly critical of current government policies and leaders and the other half supportive, with media switching roles as governments rotate between opposing parties, does not exist.The ideological filter, a basic feature of news avoidance in Western societies, is thus not a factor in China.

The S-O-R Theoretical Framework and Methodology
This study, as noted, examines the phenomenon of news avoidance in an environment where a primary filter used to avoid news elsewhere, ideology, is not a plausible basis for avoidance in China.The study presumes, however, that the S-O-R model noted earlier, which may explain why individuals engage in news avoidance behavior, would operate in the same manner in China as it would elsewhere.The methodology adopted in the article accordingly tests whether the S-O-R phenomenon is found in Chinese users of Sina Weibo and whether it leads to avoidance behavior.The S-O-R model applied in this study to news overload evolved from a longstanding S-R (stimulus-response) psychology model (K.Z. K. Zhang & Benyoucef, 2016) that explained how human behavior can be seen as a consequence of or response to external stimuli (Woodworth, 1929).While it implicitly acknowledged the intermediary stage of the unobservable inner working of the mind, the model only explicitly recognized the externally observable initial stimuli and final stage behavioral response (Watson, 1994).Building on this foundation, Mehrabian and Russell (1974) raised the intermediary element between stimuli and responses, labeled the organism stage, to a deliberately recognized (and thus tested) element of the model.Subsequent variations of the S-O-R model include the observation of Jacoby (2002, 53) that the three key components need not necessarily interact in a sequential manner but rather might be considered in terms of overlapping circles forming a Venn diagram, with a stimulus at a particular moment in time and an organism or a response in another.This enhanced interpretation has been endorsed by some later scholars (Grace et al., 2015).A more recent version recasts the terminology as stressor-strainoutcome (S-S-O; Guo et al., 2020).This is not the first digital information issue to be studied in the context of the S-O-R model.Other issues investigated using this framework include social media discontinuances (Luqman et al., 2017); website design (K€ uhn & Petzer, 2018), online shopping behavior (Peng & Kim, 2014), and online product recommendations in virtual communities (H.Zhang et al., 2019).In terms of news avoidance, the initial operative hypothesis suggested by the S-O-R model is relatively straightforward, with three elements that can be tested as three separate hypotheses (Figure 1 A refinement of the combined S-O-R phenomenon hypotheses would suggest a direct correlation between the intensity of the negative perception in O with the degree of avoidance in R: H4: Affective load mediates the effect of perceived news overload on avoidance behavior.Finally, a fifth hypothesis to test the relevance of findings in studies in Western democracies concerns the links between independent variables and news avoidance: H5: In the absence of a notable ideological driver, it will be possible to identify a correlation between news avoidance behavior and a demographic characteristic (age, gender, education, or occupation) of a media user.
The starting point for development of the study was a full understanding of the three elements in the S-O-R model, beginning with news overload.News informs people, enabling them to be better informed citizens able to contribute more to the social commons and make better personal economic and life decisions.There is, thus, a correlation between increased news and beneficial individual outcomes (e.g., Kenski & Stroud, 2006) and in an ideal world, media users would not take active steps to avoid potentially valuable news (Golman et al., 2017).
There is, however, a finite amount of news that media users can process (Chewning & Harrell, 1990) and, as a result of the digital revolution, that threshold has been reached in many jurisdictions (Schmitt et al., 2018).The key question is whether media users react to the volume of news in a passive manner, ceasing to process excess news (Eppler & Mengis, 2004), or are affected by the volume and take active steps to reduce it so they can process all the news that succeeds in entering into the news avoidance cocoon they create.It has long been recognized that persons receiving an ''infoglut'' (Shenk, 2009) or excessive information beyond their processing capacity suffer from responses variously labeled with terms such as information fatigue (Hallowell, 2005), ''individual psychological discomfort'' (Eppler & Mengis, 2004;York, 2013) or, the term adopted in this study, ''affective load.''The phrase is commonly attributed to Nahl (2004), who described how the time pressure and uncertainty of excessive information could lead to irritation, frustration, anxiety, or even rage, reactions that prompt strategies to reduce the flow of information.
A long line of studies outside China have established a link between news overload creating an affective load that has led to a news avoidance response (Blekesaune et al., 2012).However, research on the subject has tended to focus on more traditional news sources such as television news, news magazines, and news websites (Edgerly, 2015;Ksiazek et al., 2010), with the response of media users to news overload from social media sources less studied (Neubaum & Kra¨mer, 2017), particularly in China.

Sample and Data Collection
The hypotheses investigated in this study was tested by means of a questionnaire posted on the online survey platform ''Wenjuanxing'' (www.wjx.cn), the Chinese equivalent of Amazon Mechanical Turk (Choi et al., 2020;Hua et al., 2020).To prevent multiple answers from the same individual, the questionnaire could only be accessed through WeChat (the Chinese equivalent of WhatsApp), and each WeChat account could only be used once.Each participant would receive a monetary reward (roughly 0.8 USD, one-quarter the cost of a good value meal in Beijing) after successfully completing a questionnaire.Data were collected between October 26 and November 9, 2020.
An initial screen question allowed the investigators to confine the sample used to persons who consume their news primarily on Sina Weibo.A total of 1,000 questionnaires were completed and the responses were then subject to quota sampling, a research technique regarded as the optimal standard sampling method for internet-based studies (Im & Chee, 2011, 382) and one of the most commonly used non-probability sampling techniques for studies of social media platforms (Chen, 2020).Quota sampling uses a random set of responses (hence the ''non-probability'' label, as opposed to choosing respondents based on their profile) and then extracts items from the set responses that fit the different quotas or strata in numbers proportional to the population at large.This approach has the potential to produce results similar to those that would be obtained with probability sampling based on approaches to the different quota groups (Yang & Banamah, 2014).In this study, responses were extracted to create quotas proportional to the relative representation of each group in the entire population based on age and gender.This led to a data cull from the 1,000 questionnaires to 460 responses used in the study, with the drop in numbers attributable to the initial response rate over-representing some groups.
Following an initial confirmatory factor analysis (CFA) to confirm the validity of the measurement model, structural equation modeling (SEM) was conducted with the collected data via statistical software AMOS 27 by applying a maximum likelihood estimation method to measure the fit and path coefficients of the proposed model.Adopting a conventional methodology (Kim & Johnson, 2016), the reliability and validity of the measurements used for the proposed factor structure were examined via CFA, while the strength and direction of the hypothesized causal paths among the constructs were analyzed via SEM.All in all, the survey data was analyzed via the structural equation modeling to test four hypotheses.The results of the CFA and SEM are presented in the following findings sections.

Designing the Survey Questions
Survey questions sought to identify and measure three distinct themes-the extent to which respondents perceived news overload, the degree to which they felt an affective load as a consequence, and how they responded to the load.The questions were drafted following a review of previous validated queries used in news overload, affective load, and information avoidance studies, modified as appropriate for this news overload study.A total of three questions were used to test whether respondents perceived news content as an overload of information, four questions to determine whether they had an affective load as a consequence of news overload, and four questions to see if they took steps to avoid news as a consequence of affective load.Participants responded to each question on a 5-point Likert scale anchored by 1 (''never'') and 5 (''very often'').Four demographic variables were used: age, gender, education, and occupation.The S-O-R questions are set out in Table 1.
As the three elements of the S-O-R chain were connected with one another, there was some overlap between the questions.The use of overlapping questions was helpful to address the problem of selective and particular interpretation by respondents.For example, each person who constructs a news avoidance technique may perceive the nature of their actions differently though they yield the same outcome of reduced consumption of news.
In general, concepts of information overload used by media information scholars fall into two broad camps, with one understanding the construct as a unidimensional measurement, a volume of information over a particular time (Malhotra, 1982), and the other conceiving of information overload as a multidimensional construct, looking at quantity, time spent viewing and digesting the AL3: Reading too much news will make me feel weak or tired.AL4: Reading news that is different from my original idea makes me angry.(Nahl, 2004) Avoidance behavior NA1: I don't take the initiative to read news.NA2: I want to reduce the time I spend reading news.NA3: I will take measures to avoid reading news information.NA4: I don't think it's meaningful for me to read news.(Howell & Shepperd, 2016;Park, 2019) news, and the variety of sources (Huber & Daft, 1987;Schick et al., 1990).In this context, this article uses a multidimensional concept of news overload, drawing on four studies in particular: Bettis-Outland ( 2012), Holton and Chyi (2012), Karr-Wisniewski andLu (2010), andPark (2019), modified slightly to reflect features of China's social media environment.This construct comprises three components: (1) variety, measuring too many sources of news; (2) quantity, measuring news volume and availability, and (3) over-capacity, measuring the extent to which the quantity of news exceeds media users' ability to process the information.The Cronbach's alpha for these three items was .887,revealing sufficient reliability of the factors as indicators of the target phenomenon.
The indicators of the latent variable of affective load used in the article have been adapted from the four proxy variable measures postulated in the seminal introduction of the concept by Nahl (2004).The Cronbach's alpha for all four items was .900,revealing sufficient reliability of the factors as indicators of the target phenomenon.
Four avoidance cocoon identifiers were used in the study, all of which were derived (with slight modification) from a study by Park (2019, 6) on news avoidance by social media users and a study by Howell andShepperd (2016, 1697) on the propensity of individuals to avoid information more generally.The Cronbach's alpha for the four items was .855,revealing sufficient reliability of the factors as indicators of the target phenomenon.

Findings
As noted, the quota sampling process led to an initial set of 1,000 questionnaire responses, culled to yield a working set of 460 responses.The chi-square goodness of fit test, x 2 = 2.139, df = 3, p = .544(age) and x 2 = 0.849, df = 1, p = .357(gender), indicates that there is no statistically significant difference between our sample and the population of Sina Weibo users in China.That is, our collected sample by age and gender fits well with the real distribution of Chinese Weibo users.The sample consisted of 219 males and 241 females.The descriptive statistics are reported in Table 2.

Confirmatory Factor Analysis: Quality of the Measurement Model
Following an initial confirmatory factor analysis (CFA) to confirm the validity of the measurement model, structural equation modeling (SEM) was conducted with the collected data using statistical software AMOS 27, applying a maximum likelihood estimation method to measure the fit and path coefficients of the proposed model.Adopting a conventional methodology (Kim & Johnson, 2016), the reliability and validity of the measurements used for the proposed factor structure were examined via CFA, while the strength and direction of the hypothesized causal paths among the constructs were analyzed via SEM.
The study adopted the standard practice of selecting fit indices from previous research (Jackson et al., 2009;Zhou et al., 2012) and settled on the following fit indices to test the adjusted measurement model: x 2 , degrees of freedom (df), x 2 /df ratio, goodness-of-fit index (Keszei et al., 2014), adjusted goodness-of-fit index (AGFI), normed fit index (NFI), Tucker-Lewis index (TLI), comparative fit index (CFI), and root mean square error of approximation (RMSEA).The results, summarized in Table 3, showed all evaluation indices of the measurement model fell into acceptable ranges based on recommended criteria with the measurement model's fit indices well above the minimum values recommended by prior studies (Hair et al., 2019;Segars & Grover, 1993;Zhou et al., 2012).The measurement model was also found to have robust internal consistency reliability and convergent and discriminant validity.In our research, internal consistency reliability, the gauge of how well the test actually measures what we seek to measure (Glen, 2016) was tested using both Cronbach's alpha and a composite reliability measurement (Hair et al., 2019, p. 763).Table 4 shows that the values of the Cronbach's alpha and composite reliability (CR) of constructs in this model are all above .7,indicating an acceptable internal reliability (Hair et al., 2019).
The second criterion used to evaluate the measurement model, convergent validity, is a reflection of the degree to which the measurement scales are related to the corresponding construct.This is assessed by checking the average variance extracted (AVE), a common measure of convergent validity, of each construct from its indicators (Guo et al., 2020).Both the AVE and the standard factor loadings (the relationship between the collective queries and the primary factors they are intended to indicate) of the questionnaire items were over 0.50 and 0.60, respectively, indicating solid convergent validities (Fornell & Larcker, 1981).
A third check of the validity of the study data is the discriminant validity, a measure of the degree to which items distinguish between constructs.This study compares the square root of the AVE of each construct and its correlation with other constructs to test discriminant validity.Table 5 sets out the square roots of the AVEs for each construct (as the final bold text creating a diagonal pattern).All the correlations between variables were smaller than the square root of AVE, suggesting a good discriminant validity (Hair et al., 2019).

S-O-R Structural Equation Modeling and Hypotheses Tests
Figure 2 depicts the main path coefficients and explained variances of the endogenous variables (R 2 ).The analysis of the structural model suggests that the news overloadaffective load-news avoidance indicators can explain 22.8% of the variance in avoidance behavior, indicating that the research model has well enough developed explanatory power.
As Table 6 demonstrates, the structural model revealed that news overload is positively related to Chinese social media news avoidance behavior (H1, b = .213,p \ 0.001) and affective load (H2, b = .235,p \ 0.001).In addition, affective overload is positively related to users' avoidance behavior (H3, b = .367,p \ 0.001).Thus, the S-O-R hypotheses H1, H2, and H3 are supported by the data.
Hypothesis 4, looking at the relationship between the intensity of the affective load and behavior responses, was tested using bootstrap methods proposed by Shrout and Bolger (2002).As noted by David (2012), this method is appropriate for the estimation of mediated effects.
The commonly used bootstrap confidence interval was adopted to judge the existence of a mediating effect (Carpenter & Bithell, 2000): regardless of whether a  percentile-based bootstrap confidence interval or a biascorrected confidence interval is used, if zero is not between the lower and upper bound, the indirect effect can be regarded as significant, and vice versa.As shown in Table 7, for the news overload!affectiveload!news avoidance behavior link, total effects, direct effects, and indirect effects are all significant (zero is not between the lower and upper bound), indicating that affective load partly mediates the relationship between news overload and avoidance behavior.The findings support the validity of H4.

Independent Variables Test
The initial, and primary, issue investigated in this study was whether the S-O-R theory would apply to news overload in China in the same manner as has been found in Western democracies notwithstanding the absence of an ideology division in news sources.

Discussion and Implications
The findings from this study, supporting four proposed hypotheses, found that the S-O-R news avoidance phenomenon experienced in Western democracies is also present in China.Consistent with previous studies conducted in democratic countries (Park, 2019), this study presents new evidence of the direct impact of news overload on news avoidance behavior occurring in a voluntary social media information consumption context in China.Also, the study showed the direct influence of news overload on users' affective load, the direct influence of users' affective load on their avoidance behavior, and the mediating role of affective load between news overload and avoidance.These findings are also in line with the past studies that have argued that too much negative news causes emotional disruptions among news consumers and leads to their subsequent avoidance behavior.
In one key area, however, the findings of the study diverge from those in previous studies such as Holton and Chyi (2012), in that none of the independent variables (age, gender, education, or occupation) were found to have had an influence on users' avoidance behavior.In this respect, news avoidance behavior in China seems to be quite different from that in Western democracies.There are two possible paths that could be explored to explain this difference.First, despite the similarities in technology and media sources, there may be fundamentally different social influences in China that largely reduce or eliminate any differences in reactions to news overload by persons of different gender, age, education, or occupation.The second relates to the most significant difference between news in China and Western democracies, namely the relatively homogenous ideological message in state-owned press outlets in contrast to the sharp ideological divides in society, the news, and citizens' posts on news stories in Western countries.Almost all previous research has stressed the role of ideology in directing news consumers to news that is only in agreement with their ideological and political views, leading to an information narrowing effect (Sunstein, 2009).It could well be the case that the ideology factor is so large that it colors measurements of reactions based on the four independent variables studied.In other words, it could be that the differences between persons of different gender, age, education, or occupation reflect different levels of susceptibility to the impact of ideology on news overload and once that factor is removed, the differences in respect of each factor largely evaporate.This issue merits further study.

Caveats and Limitations
The media and news source investigated in this study has a finite reach in the media market as noted earlier-it is, as noted, used by about 20% of China's 1 billion internet users.It is possible that the findings do not reflect fully the responses of the broader group of internet users in China and instead only show outcomes from persons who have self-selected as users of this media platform.Nevertheless, a study of a representative sample of 200 million persons can provide insights into the behavior of a not insignificant segment of media users.
Two further limitations, that might be addressed in future studies, were identified.Media users who consumed news primarily via Sina Weibo also had access to a range of other types of news media including television, newspapers, and the internet.While the study included a screening question to ensure participants in the study consumed their news predominantly via a social media platform rather than via other media platforms, the study relied on respondents' subjective self-reports without adopting an objective means of testing the reliability of their responses.Also not measured was the impact on behavior of participants' previous experiences, peers, and family members.All three of these factors are known to impact individuals' news consumption behavior (Neuman et al., 1992).

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
): H1: Users of the Sina Weibo social media platform encounter what they perceive to be news overload (S) and engage in news avoidance (R), H2: The users (O) experience an affective load as a result of the news overload stimulus, and H3: The affective load prompts a reaction (R) by users in the form of news avoidance.

Figure 2 .
Figure 2. A theoretical model on the relationship between news overload and news avoidance.Note.ns = no significance.***p\.001.

Table 1 .
Constructs and Associated Items.

Table 3 .
Fit Indices of the Model.

Table 4 .
Internal Consistency Reliability and Convergent Validity of the Measurements.

Table 6 .
The Result of Path Analysis.

Table 5 .
Discriminant Validity of the Measurements: Correlation Analysis of Latent Variables and Square Root of the AVE.

Table 7 .
Specific Indirect Effect Test.Note.Standardized estimating of 1,000 bootstrap sample.NO = news avoidance; AL = affective load; AB = avoidance behavior.