How Media’s factor and governmental factor affected individual’s opinion on haze in China

Understanding the underlying mechanism of individual’s opinion on haze is important to govern public opinion on haze, and the major question of exiting research is that it has not yet explored what factors and how do they affect individual’s opinion on haze. This study aimed to uncover the underlying influencing mechanism of individual’s opinion on haze. This study conducted Partial Least Square Structural equation Modeling (PLS-SEM) to examine the impacts of media’s factor (i.e. agenda setting, media accountability, and information credibility), governmental factor (i.e. policy instrument) and psychological factor (i.e., social distance, spatial distance) on individual’s opinion on haze. Based on the Agenda Melding Theory and stakeholders (i.e., media, government, and resident), this study constructed a conceptual model, which is empirically examined by conducting questionnaire survey data of 973 residents in China. The results indicate that media’s factor (media accountability, information credibility), governmental factor (policy instrument), and psychological distance have significant positive direct impacts on individual’s opinion on haze. Furthermore, media’s factor (agenda setting) only has an indirect impact on individual’s opinion on haze through the mediation effects of information credibility. However, information credibility also had significant moderation effects on the relationship between psychological distance and individual’s opinion on haze. The moderation and mediation effects of information credibility indicated that it played a vital role in guiding individual’s opinion on haze. These findings suggest that further research is required on the interaction of stakeholders and that concerted efforts of different stakeholders to assist governing individual’s opinion on haze.


Introduction
Haze pollution has increasingly become a prominent environmental problem (Liu and Dong 2021), threatening public health, social stability, national security, and economic development across the global (Ali et al 2019, Raza et al 2021). It is well-known that, 'mass media do not tell us what to think, the mass media do have considerable power to tell us how to think about topics', which means that the traditional agenda setting of mass media (i.e., newspaper, television, broadcast, etc) has the function of the 'agenda' for the public (McCombs and Shaw 1972). The development of internet technology has promoted the emergency of various media, which symbolized the diversity of media, that is, media diversity. In the context of media diversity, the public can express their opinions and dissatisfaction freely and cheaply through different media, which led to audience fragmentation (King et al 2013). Meanwhile, agenda-setting of haze was also challenged by the media diversity and audience fragmentation (Feezell 2018), affecting residents' cognition of haze and might lead to negative public opinion on haze. Especially, the information disseminated by social media can reshape the public risk perception , and further influenced scientific consensus (Weber 2017). The psychological distance (i.e., the spatial and social distance between the individuals and haze polluted areas) also affect their willingness to take action on haze (Trope and Liberman 2010).
In order to improve environment and air quality, many countries around the world are committed to taking effective prevention and control measures, such as reducing pollutant emission, selling licenses to business setting a limit to the pollution, government intervention on vehicles, introducing renewable energy, and strictly abiding and implementing China's blue-sky Plan (Li and Peng 2016, He et al 2018, Zhang and Hao 2020. In addition, policymakers should pay more attention to attitudes and emotions of the public, and to fully understand the public's willingness to make efforts to alleviate the haze pollution (Sun et al 2016). Public opinion can influence the environmental protection behaviors of local governments, thereby improving air quality in China .
It is well known that agenda setting cannot be separated from the public and the government, is a matter of public interest, an urgent need to solve the contradiction between economic development and social sustainability (Shi et al 2016). Public's strong appeal for environmental issues required the government's efforts to control environmental pollution (Shen et al 2019), and the public opinion on haze has attracted governments' attentions , through media's agenda of governance of public opinion on haze, haze's causes and harms, and haze's control measures, and so on (Yang et al 2019). Media diversity and public awareness of environmental protection are the major factors of promoting public to engage in environmental issues, but they may also lead individuals to take risk behaviors, which bring challenges to the governance of public opinion on haze, if the ecological environment continues to deteriorate, it will cause unrest (Huang 2015), this ultimately lead to much more negative public opinion on haze.
Previous studies have pointed out haze's harms, prevention and control measures, and the impact of public concern on air quality (He et al 2018, Raza et al 2021, while information produced by many social media (network media, social media, etc) often lacked professional scrutiny and supervision, making it difficult to confirm the credibility of information and its source, which resulted in uncertainty on individual's opinion on haze. However, an important question remains: what factors and how do they affect individual's opinion on haze? This question becomes especially important to answer for stakeholders (residents, government environment agency, and media, etc), as the responsibility for judging content and accuracy shifted from information producers to information consumers (Wearden 2003), making the influencing factors of individual's opinion very complex and confusing. This study purposed to construct a comprehensive framework and empirically verified the impacts of media's factor, governmental factor, and psychological factor on individual's opinion on haze.
The structure of this study is as follows. Section 2 reviewed previous literatures, gave the hypothetical framework of this study, and constructed the hypotheses. Section 3 illustrated samples and measurement of variables. Section 4 showed the results of the impact of media's factor (agenda setting, information credibility, and media accountability), governmental factor (policy instrument), and psychological distance on individual's opinion on haze in China. Section 5 showed the discussion and policy implications. Section 6 provided the conclusions.
2. Literature review and theoretical framework 2.1. Theoretical framework The mass media's news reports and information transmission activities provided various 'issues' with different levels of significance, which influenced the public's judgment on the 'major events' and their importance. Its theoretical basis is one-way liner communication, which completely separated the communicator from the receiver, and the communicator decided the content and quantity of the communication and played a guiding role for the receiver. Meanwhile, it should be noted that elite media (i.e., New York Times, Washington Post) also showed a strong agenda-setting effect (Meraz 2011, Vargo et al 2018a. However, in the era of the rapid development of the internet and social media, the communicator and receiver gradually become one, realizing two-way communication and interaction. Therefore, it is necessary to introduce a new 'Agenda Melding Theory' to explain the influencing mechanism of individual's opinion on haze. 'Agenda Melding Theory (AMT)' is an extension of the Agenda Setting Theory (AST). The starting point of AMT is the public, and it pointed out how the public used different kinds of media and its social effects, and emphasized that media played a key role in the interaction between individuals and others, which helped audience to reduce cognitive dissonance between individual 'issues' and group 'issues' (Donald L. Sharp 1999), such as subgroup's brand community agenda was similar to the aggregate's (Ragas and Roberts 2009).
'Current public opinion' refereed to public opinion that occurred contemporaneously as air pollution developed, including complaints, condemnations, disputes, queries, and so on . 'Current public opinion' could influence the governmental actions, if a haze pollution incident aroused public awareness of improving air quality and quickly disseminated online, attracting the attention of higher authorities, once environmental agency took action to reduce haze pollution, the media reportcan help publicize the government's efforts to reduce haze pollution (Stockmann and Gallagher 2011). However, if the haze pollution is getting worse, leading to great public opinion pressure online, ultimately may cause social or political event (Esarey and Qiang 2011).
Public concern about environmental problem, that has been verified, is very important to developing environmental intervention policy instrument (Forsyth 2014). According to different levels of economic development, public concerns can improve air quality to varying degrees in the short term , and the public concern about health risk of haze pollution varies by different individual (Liu et al 2016), which can also affect government's actions . The aforementioned literature can indicate the media, the government, and the public are the important stakeholders in the formation process of individual's opinion on haze, and the conceptual model is presented in figure 1. This model combines media's factor (agenda setting, media accountability, and information credibility), governmental factor (policy instrument) and psychological distance as the determinants, and verifies their impacts on individual's opinion on haze. Moreover, information credibility is viewed as both the mediation factor and moderation factor in this study.

Hypotheses
2.2.1. Variables of stakeholders in the agenda melding theory Agenda setting for political issues is a dynamic interaction process between traditional media and mass audience (Neuman et al 2014). The multiple tensions between government authority, journalism transformation, and participatory democracy are formed, making the online public sphere very complex and turbulent (Shao and Wang 2017). Social media is an important channel for disseminating air quality information, promoting the environment protection actions (Hswen et al 2019). Public opinion on social media could affect traditional commercially oriented media, and the interaction between the public and policy agenda, that is, public participation has made journalism an ongoing process, ultimately affecting the content of traditional media and governments' behaviors (Luo and Harrison 2019). Public opinion online has become a competitive force in the process of agenda setting, which has a bidirectional impact on traditional agenda setting (Luo 2014). Similarly, in this study, we expect that the more salient agenda setting, the more positive individual's opinion on haze. Thus, we propose the following hypothesis: H1. Agenda setting may positively affect an individual's opinion on haze.
Whether environmental policy can be successfully implemented or not depends on public risk perception (Kim et al 2012). When controlling the influences of knowledge, attitudes, and expectations from government governance of air pollution, the policy instrument could significantly reduce the public risk perception (Wang and Watanabe 2019). Public awareness of air quality, government credibility in environmental governance, controllability, policy factor, personal income level, and occupation can affect the public's willingness to pay for addressing air pollution (Shi et al 2019. Some environmental policy and regulation were not well known by the public, and were not well implemented, causing undesirable effects, particularly in heavily polluted areas (Wang and Watanabe 2019). It can be speculated that the more known of policy instrument, the more positive individuals' opinion on haze. Thus, this study hypothesizes that: H2. Policy instrument may positively affect an individual's opinion on haze. The media stimulated actors to reflect on their actions, expanding formal accountability through media coverage that acted as an informal accountability and independent accountability (Jacobs and Schillemans 2016). Media coverage affected public accountability process in the public department (Jacobs et al 2021). Internal structure and market forces (i.e., information consumer) are key determinants of news organizations' power over public opinion (Bovitz et al 2002). From the above, we can speculate that individual's action, media's coverage, and government regulations are interacting with each other. In fact, some social or business media have neglected social responsibilities, that is, in pursuit of economic interests, and the fiercely competitive media market restrained the space for public media to correct deviations of public opinion online (Bardoel and Dhaenens 2004).
According to the governance of individual's opinion on haze, media accountability is an interactive process, involving government regulation, media autonomy and public participation. It focused on the interaction and game of haze information among media, government and the public, with emphasis on the institutional arrangement of media accountability. It can be seen that although local government recognized the importance of the public participation mechanism, in reality, the channel for the public speak out was often one-way, leading to the 'continuous spread of wrong public opinion', and ultimately leading to the occurrence of haze public opinion deviation. We similarly speculated that, if the higher intensity of media accountability, which can bind the media behavior, the more positive individual's opinion on haze. Hence, this study hypothesizes that: H3. Media accountability may positively affect an individual's opinion on haze.
There are more and more channels for the public to understand the haze pollution, including mobile apps and online media content, etc (Li and Tilt 2019). As time goes by, people are inclined to resist the media information that is inconsistent with the first-hand information from friends and neighbors, mass media mobilization can affect public opinion online and bring pressure on administrators (Stockmann and Gallagher 2011). In fact, fact-checkers have no power of determining the agenda of news media overall, indicating that fact-checkers face with difficulties in expressing their opinions (Vargo et al 2018b). Public perceptions of news credibility were different among newspaper, online news and television news (Kiousis 2001). To effectively guide public opinion on haze pollution, civil society (refers to Chinese environment grassroots NGOs, citizen activists, and the media) played an important part in China's environmental governance, which raised public awareness of environmental pollution, helping improve transparency and accountability of China's environmental information (Ho 2001, Zhang 2018, indicating that reaching a consensus among different stakeholders on environmental protection is the key to tackling haze pollution. Based on these studies, we speculate that information credibility has a significant impact on individual's opinion on haze pollution. Hence, this study hypothesizes that: H4. Information credibility may positively affect individual's opinion on haze. 'Psychological distance' is a concept, which refers to what extent an individual is removed from the geographical distance and social distance, when an individual is believed to be psychologically close to himself, their willingness to take action increases (Trope and Liberman 2010). Individual values, norms and beliefs are the optimization framework of psychological distance, which can help individuals reduce fear and regulate emotional reactions (McDonald et al 2015). The closer psychological distance to climate change, indicating the more participation in pro-environmental behaviors . Based on these studies, we speculated that the haze pollution is closer to an individual's living area and hometown, the stronger emotion reactions they have, which will affect their opinions on haze. Thus, this study hypothesizes that: H5. Psychological distance may significantly affect individual's opinion on haze.

Mediation and moderation effects of information credibility
For news information, information credibility is measured from the perspective of the public's evaluation of the accuracy, credibility and authenticity of news (Appelman and Sundar 2016). Information on social media affected people's perception of information credibility (Westerman et al 2014). The credibility of online information depended on the information-seeking context and the information seeker (Metzger and Flanagin 2013). Accordingly, we posit that, in the process of haze information dissemination, the agenda setting, the public awareness of policy instruments, and the public cognition of the media accountability have impacts on the haze information credibility, thus affecting an individual's opinion on haze. Thus, we propose the following hypotheses: H6.a-c: Information credibility mediates the relationship between agenda setting/policy instrument/media accountability and individual's opinion on haze.
H7.a-b: Information credibility moderates the relationship between spatial distance/social distance and individual's opinion on haze.

Methods and data
3.1. Data collection The data was collected by conducting a field and online together questionnaire survey. The questionnaire composed of four sections. The first section described the aims and background of the study. The second section was used to collect respondents' evaluation of media's factor, governmental factor, and psychological factor. The third section collected respondents' opinion on haze pollution. The last section aimed to collect the demographic information. The sample is a survey of 31 provinces in China during the period of January to March 2021 (not including Hong Kong, Taiwan and Macao), which is the most polluted period of the whole year (Huang et al 2014). Haze pollution caused respiratory diseases (He et al 2018), and the public hoped to alleviate haze pollution and attract the attention of relevant government departments.
We employed a few convenience field surveys in Xi'an, Shaanxi Province, including universities, parks, market and residential communities, which had large urban populations. Meanwhile, we also conducted an online anonymously survey through Tecent QQ and WeChat. All the participants were confidential and voluntary to fill out the questionnaire. This survey was conducted following the ethical standards of the Declaration of Helsinki (World Medical Association 2013) 1 . The questionnaire was shared on social platforms like Tecent QQ and WeChat, we first disseminate it to our friends and relatives, and they also requested to reforward the questionnaire link to their contacts and social networks. According to that the different levels of haze pollution have different impacts on public concern (Li et al 2021), we collected more samples in the areas with relative more serious haze pollution 2 , and for example, Shanxi Province (19.83%), Hebei Province (15.40%), Shandong Province (11.81%), and Hei Longjiang Province (10.23%), which were the most pollution areas. The proportion of samples from the other provinces was less than 10%.
A total of 1036 responses were obtained, and 63 invalid responses were eliminated because of missing values and logical errors. Finally, we obtained 973 valid responses. The values of recovery and effective rate of the questionnaire are 31% and 93.9% respectively. The demographic information was showed in table 1. The results show that the proportion of males is 56.6% and the proportion of females is 43.4%, and the male was a little more than females, indicating the gender distribution of the sample was closer to the national population of the Chinese mainland (51.24% male, 48.76% females) 3 . The majority of respondents were aged between 18 and 40 (78.7%), and over 40 years old (21.3%). All the recipients were over 18 years older, with different levels of education and ages, as shown in table 1. From the above, the spatial distribution of the recipients was consistent with the haze pollution level, and the gender distribution was also consistent with national population, that is, the samples were approximately representative and were acceptable for this study.

Measures
All constructions in this study were latent variables, and were measured using the multi-item 5-Likert scale. The measurement of an individual's opinion on haze was mainly cited and amended from the previous studies  policies and regulations in China (Liang and Langbein 2015). Psychological distance was adopted from the construal level of climate change (Wang et al 2019). Information/message credibility was adopted from the existing studies of message credibility (Appelman and Sundar 2016). All the items were rated using a 5-point Likert scale, with 5 indicated that respondents highly agreed with the items, and 1 indicated strongly disagreed. The constructs and items were described in appendix.

Data analysis method
We conduct a Partial Least Squares Structure equation Modeling (PLS-SEM) to examine the conceptual model and hypotheses, for two reasons. First, comparing with the Structural equation Modeling (SEM) (Zhu and Yao 2019), the PLS-SEM has the advantage, that it fully mines the information of data to minimize the error item, and has low requirements on the sample size of the data, model recognition and distribution state, that is, employing the PLS-SEM can handle the non-normal data, and it can effectively deal with the problem of collinearity between variables. Especially, it can be applied to testing a less developed or developing theory (Hair et al 2013), and it is also suitable to estimate a developing theory from a prediction perspective (Khan et al 2022).
Second, based on the Agenda Melding Theory, we developed a conceptual model and mentioned the important stakeholders (government, media, individual/residents), and introduced two variables (i.e., media accountability which involved all the stakeholders, and psychological distance which is a psychological construct) to uncover the influencing mechanism of individual's opinion on haze.

Measurement model analysis
Before examining the research hypotheses, we followed the suggestion of previous study to test the reliability and validity of the constructs . Confirmed factor analysis is used to evaluate the reliability and validity of the questionnaire before testing the hypothesis (Larcker 1981). Individual's opinion on haze is constructed as an index of five different emotions in daily life, and psychological distance is measured by two types of distance, which are not also required of reliability and convergent validity analysis . Reliability of all the variables was judged by composite reliability (CR) (Larcker 1981) and Cronbach's alpha (Larcker et al 2007). All the CR values ranger from 0.838 to 0.867 (shown in table 2), and were more than 0.70, indicating that these constructs passed the reliability test . Standardized factor loadings and Average Variance Extracted (AVE) were used to test the convergent validity of the latent constructs. All the factor loading values of each item (shown in figure 3) were more than threshold value 0.60 . These values indicated that our data exhibited satisfactory convergent validity. The squared roots of AVE values were shown in table 2, which were more than 0.50 and higher than the correlation coefficients with the other constructs, indicating the each item connected with its own construct more strongly than with the other constructs (shown in table 3). Therefore, each construct can be judged to have good discriminant validity. Furthermore, the results showed that all variance inflation factor (VIF) values ranged from 1.363 to 1.651 (shown in table 2) and were less than 10.0, indicating there was no multi-collinearity problem between independent variables (Wang et al 2017).

Partial least square structural equation modeling (PLS-SEM)
In order to compare the explanatory power of the Agenda Setting Theory (AST) with the Agenda Melding Theory, we conducted two PLS-SEM analyses based on the two theories. First, we conducted a PLS-SEM analysis based on the AST, and the results were shown in figure 2. The results showed that the AST model's fit indicators SRMR (= 0.054) was less than 0.08, NFI (= 0.830), d_ULS (= 0.676), and d_G (= 0.246) were less than 0.95. The R 2 values of the Information Credibility and Individual's Opinion were 0.693 and 0.694, respectively, indicating the PLS-SEM AST model had an adequate fit (Hair et al 2019). Second, we also conducted a PLS-SEM analysis based on the Agenda Melding Theory (AMT), and the results were shown in figure 3. The results showed that AMT model's fit indicators SRMR (= 0.056) was less than 0.08, NFI (= 0.798), d_ULS (= 0.946), and d_G (= 0.405) were less than 0.95. The R 2 values of the Information Credibility and Individual's Opinion were 0.715 and 0.771, respectively, indicating the PLS-SEM AMT model had an adequate fit (Hair et al 2019).
Third, we compare the results of AST Model with AMT Model, we find that the AMT Model's R 2 (= 0.771) were more than AST Model's R 2 (= 0.694), indicating the explanatory power of AMT Model is higher than the AST Model. Therefore, we used the AMT Model to examine the research hypotheses, the resultswere shown in table 4 and figure 3.
Finally, the PLS algorithm was conducted for 300 iterations to obtain the standard coefficient, and the Bootstrapping was performed for 5000 subsamples to test the significance of the path coefficient, the results were shown in table 4. Figure 3 and table 4 presented that the path coefficients and significance. The results showed that agenda setting (β = 0.038, p = 0.310 > 0.05) had no significant direct impact on individual's opinion on haze, thus not providing support for H1. The relationships between the stakeholders of agenda-melding and individual's opinion on haze were tested by H2, H3 and H5. Media accountability (β = 0.390, p < 0.001), policy instrument (β = 0.247, p < 0.001), and psychological distance (β = 0.152, p < 0.001) also had positive and significant direct impacts on individual's opinion on haze. Meanwhile, the results also showed that information credibility (β = 0.173, p < 0.001) had a positive and significant impact on individual's opinion on haze. Hence, H2, H3, H4 and H5 were validated. These findings showed that individual's opinion on haze was directly influenced by policy instrument, media accountability, psychological distance and information credibility. Control variables (gender, age, and education) had no significant impact on individual's opinion on haze.

Mediation effect analysis
Based on the significant correlations among agenda setting, policy instrument, media accountability, information credibility, psychological distance, and haze public opinion, we explored the mediating roles of the information credibility. This study conducted the bootstrap analysis to estimate the multiple mediation relationship (Wang et al 2020). Agenda setting, policy instrument, and media accountability were entered as the independent variables, individual's opinion on haze was entered as dependent variable, and information credibility was entered as proposed mediator. The bootstrapping method with 5000 subsamples was conducted to test the significance of the mediation effect of information credibility. Preacher and Hayes (2008) asserted that 'significance of the total mediation effect was not a necessary precondition for identifying significant specific mediation effects in multiple mediator models'. The results of mediation effects were shown in table 5, agenda setting, policy instrument, media accountability exerted significant mediation effects on haze public opinion via information credibility.
Furthermore, the results showed that information credibility had full specific mediation effect between agenda setting and individual's opinion on haze. It was indicated that, through the information credibility, the more agenda setting highlight 'consensus on environmental risks, call for participation in haze reduction, government policy instrument for haze control, harm of haze and air quality monitoring ranking', the more positive of haze public opinion will be. Information credibility also had partial mediation effects on the relationship of policy instrument and individual's opinion on haze, and on the relationship of media  accountability and individual's opinion on haze. These indicated that information credibility played an important role in the association between agenda setting, policy instrument, media accountability and individual's opinion on haze. Hence, H6a, H6b and H6c were validated.

Moderation effect analysis
For a better understanding of the associations between psychological distance (spatial distance, social distance) and individual's opinion at different levels, according to the variable value 2.5, we recoded the information credibility into two categories (0 = low level, 1 = high level). Similar, we recoded the spatial distance into two categories (0 = near, 1 = far) and the social distance into two categories (0 = near, 1 = far). Then a 2 (two levels of spatial distance)×2 (two levels of information credibility) two-way ANOVA was carried out with categorical credibility and spatial distance as independent variables or between-groups factor, and individual's opinion on haze as the dependent variable ( figure 4). The main effect of information credibility on haze public opinion is significant, F = 150.350, p < 0.000. So does the main effect of spatial distance on haze public opinion is significant, F = 58.737, p < 0.000. There is significant interaction between spatial distance and information credibility on haze public opinion was also observed, F = 6.525, p = 0.011 < 0.05. Similarly, a 2 (two levels of social distance)×2 (two levels of information credibility) two-way ANOVA was carried out with categorical social distance and information credibility as independent variables or betweengroups factor, and individual's opinion on haze as the dependent variable ( figure 5). The main effect of information credibility on haze public opinion is significant, F = 163.870, p < 0.000. So does the main effect of social distance, F = 90.463, p < 0.000. There is significant interaction between social distance and information credibility on haze public opinion was also observed, F = 8.588, p = 0.003 < 0.05.
Hence, the above ANOVA analyses confirmed a result that, at the same level of social distance and spatial distance, the group who reported higher level information credibility had the more positive individual's opinion on haze, which indicated the H6a, H6b and H6c were validated.

Partial least squares predict analysis
The PLS predict algorithm uses training and hold-out samples to generate and evaluate predictions estimated by the PLS-SEM model, meaning it combines aspects of out-of-sample predictive power and in-sample explanatory power (

Discussion
This study aimed to uncover the underlying influencing mechanism of individual's opinion on haze and to estimate the relationships between agenda setting, policy instrument, media accountability, information creditability, and individual's opinion on haze. The results showed that positive individual's opinion on haze experienced high levels of agenda setting. The mean value of individual's opinion on haze found in our sample were 3.830 (SD = 0.681), which is little higher than the medium 3, and it indicated that individual's opinion on haze in China urgent to be improved. As expected, the results showed that agenda setting, policy instrument, media accountability, information creditability, and psychological distance were indirectly or directly had significant and positive impacts on individual's opinion on haze. The results are consistent with existing studies about the relationship between  policy instrument, information creditability and public opinion (Luo 2014, Shi et al 2019, Wang and Watanabe 2019. In addition, the results indicated that, taking the mediation effect of information credibility into account, the more prominent the agenda setting, the more positive individual's opinion on haze, which offered some support to existing findings on the positive correlation between agenda setting and public opinion (Luo 2014). Furthermore, the finding that information creditability positively correlated with public opinion is consistent with the studies of (Metzger and Flanagin 2013, Westerman et al 2014, Curry and Stroud 2021. The contributions of this study are shown as follows. First, agenda setting has indirect impact on individual's opinion on haze through the mediation effects of information credibility, which was similar with that information credibility had a mediation effect on the relationship between agenda setting and engagement intentions or public opinion (Curry and Stroud 2021). Second, we apply the Agenda Melding Theory to the field of environmental communication, and introduce the variable (media accountability) to explain the influencing mechanism of individual's opinion on haze, and provide empirical evidence to test it, helping us to strengthen the theoretical understanding of the determinants of individual's opinion on haze. Third, this study also found media accountability has a significant positive impact on individual's opinion on haze, which had not been examined in previous studies. There were still some limitations to be addressed for future studies. First, all the scales and items in this study were self-reported questionnaires, there inevitably are subjective bias. Second, individual's opinion on haze is conceptualized as a composite emotion based on five kinds of perceptions. Thus, it is necessary to conduct detailed research regarding network sentiment analysis.
The results of this study also provide some policy implications for governing individual's opinion on haze. First, in terms of agenda setting, the media should continuously improve the information credibility and closely follow the haze issues of public concerned. Based on the results of the empirical research, when setting the haze agenda, the media should highlight air quality monitoring, reduction measures and haze harms, to enrich the public's knowledge of haze, which can dispel public doubts about haze pollution. It should be noted that the Agenda Melding Theory emphasizes interaction, including the interaction between the media, the public and the government, as well as the interaction between traditional media and online media, especially among different new media. To enable the public to actively and equally participate in the discussion of haze topics, and to have a dialogue with the government or the media in a timely or rapid manner, so that the government should satisfy different demands of public concerns and create a good atmosphere for public opinion. Publicity campaigns and environmental protection education towards haze pollution can be introduced to individuals, which can increase their positive attitude and confidence in reducing haze pollution . Thus, the media and government should take more effective information dissemination and mobile the public to actively participate in reducing haze pollution (Shi et al 2016). To do this, residents or netizens should disseminate scientific haze information through social media and traditional media, such as Weibo, WeChat, QQ, television and newspaper. More importantly, to foster a social atmosphere of haze reduction, information credibility should be taken seriously by all the media (Metzger and Flanagin 2013). In this regard, authoritative and credible social media accounts should be used to disseminate relevant information on haze intervention, monitoring or protective campaigns. Meanwhile, environmental protection agency should seek help by opinion leaders to enhance the persuasive effects of haze information.
Second, the results showed that media accountability have a significant direct impact on individual's opinion on haze, which was consistent with the previous study about symbolic rewards and sanctions by the media, the media can expect the tone of editorial comments (Maggetti 2012). Therefore, it is suggested that the media agencies should formulate their ethics norms and employees' responsibilities norms, government departments should do a good job in top-level design, establish and improve media regulations and policies, supervision mechanism and accountability mechanism, and strictly restrict media institutions' misconduct. The public supervised the haze information disclosure and response behaviors of media organizations through different channels, such as hearings and interactive platforms, and urged the media to report haze information in a more scientific way. In terms of media accountability, it is necessary to clarify the position and functions of the media in the governance of haze opinion communication through media identity identification, which is to make media work better, and coordinate and balance the society. For example, media management agency should stratify the various media, while giving more tolerance and understanding to the multiple roles of the media.
Finally, the previous studies found the necessity of specific, controllability and urgent policy instruments should be conducted to promote individuals' ability to execute the behavior of reducing haze pollution (Yang et al 2016, Shen et al 2018. The dominance of policy instruments is mainly reflected in the transparency with which actors and the media carry out their responsibilities, especially when compared to policy instruments that undermine self-regulation (Pérez-Díaz et al 2020). In this regard, environmental protection authorities should construct and improve the environmental policy instruments system, including mandatory administration policy instruments, economic incentive policy instruments, and voluntary participation policy instrument (Huang et al 2021), as well as supplement measures, subsidies, grants and tax credits etc Meanwhile, it can be found that psychological distance has a significant positive impact on the individual's opinion on haze.

Conclusions
From the perspectives of the agenda-melding theory and stakeholders, this paper has conducted a hypothetical model and presented the statistical examination of influencing mechanism of individual's opinion on haze. The results are shown as follows. First, for the perspective of media's factor, agenda setting indirectly influenced individual's opinion on haze through the mediation effect of information credibility, media accountability directly influenced individual's opinion on haze. Especially, it is worth noting that information credibility presents a significant mediation and direct effects on individual's opinion on haze. Second, for the perspective of governmental factor, policy instrument directly influenced individual's opinion on haze. Third, the information credibility had a significant moderation effect on the relationship between psychological distance and individual's opinion on haze. This study underscored the importance of media's factor, governmental factor, and psychological factor as drivers of individual's opinion on haze, and enriched our understanding of individual's opinion on haze governance. Our findings are also relevant to environment governance practices concerning public opinion management. It is suggested that local governments should publicize and interpret the connotation of the haze control policy instruments through different channels and regions, constantly enhance the public awareness of the policy instruments, improve the media accountability system, and guide the public to rationally understand haze pollution.