Social media exposure’s effects on public support toward three-child policy in China: role of cognitive elaboration, perceived negative effects, and institutional trust

Abstract

The three-child policy constitutes a hotly debated socio-political issue in China. Upon its announcement, many Chinese citizens have ridiculed the move on social media. Adopting the cognitive mediation model and the influence of presumed influence theory, this study examines how social media exposure to three-child policy-related news and discussions could affect the Chinese public’s attitudes toward the policy. The online survey results show that social media exposure negatively predicts supportive opinion via cognitive elaboration and three types of perceived negative effects of the policy (i.e., perceived negative effects on self, on the public, and on females) in serial. It also finds that institutional trust moderates the relationship between cognitive elaboration and policy support. Only among people with high institutional trust, there is a positive effect of social media exposure on supportive opinion through cognitive elaboration.

Introduction

As there has been a rising number of studies on the effect of new media in the domain of political communication within Western contexts, especially in the USA (Bennett et al. 2013; Chadwick and May 2003), scholars are calling for special attention and a comparative approach to study the implications of new media on political change, particularly in regimes that practice Internet censorship, such as Arab countries and China (Eltantawy and Wiest 2011; Wu et al. 2011). In the context of China, the role of social media in political discussion is even more interesting and controversial. In recent years, new information communications technologies have flourished in China, with China’s nearly one billion Internet users constituting the world’s largest digital population. Despite Facebook, YouTube, and Twitter being blocked in the country, local social networking sites such as WeChat, Weibo, and Douyin have been attracting millions of users (Statista 2020). Although the Chinese government still effectively controls the online diffusion of information through content moderation censorship, the high degree of marketization of the Internet sector makes new media comparatively less of a government tool than traditional media (Wang 2017).

Social media acts as a double-edged sword for governmental issue management. On the one hand, it provides a platform for timely and less costly information delivery and facilitates a two-way conversation, enabling multi-level agencies to consistently monitor and respond to public inquiries (Eckert et al. 2018; Paek and Hove 2021). On the other hand, social media also serves as a channel for online firestorm propagation, emotional arousal, negative word-of-mouth, and the polarization of public opinion (Johnen et al. 2018; Pfeffer et al. 2014; Smith and Smith 2022). In terms of knowledge acquisition, propagation, and affirmation, social media mediates the cognitive reasoning processes. Thus, social media plays a significant role in mediating stakeholders’ opinions toward various socio-political issues, which may add external pressure to government issue management.

According to Chase (1984), an “issue” refers to an unsettled matter looking for resolutions. Issue management is important because an issue, if not managed properly, may produce a crisis. This study focuses on the issue of childbirth in China, a socio-political issue intertwined with the declining birth rate, fertility rights, gender equality, and policy inconsistency. On May 31, 2021, the Chinese government announced its support for couples who wish to have a third child through a list of measures. According to Xinhua News (2021), the purpose of implementing this new policy is to “improve China’s population structure, solve aging population problems, and preserve China’s advantage in human resources.”

Following the policy’s announcement, discussions related to the three-child policy saw a sudden popularity on China’s social media platforms with different opinions. Some individuals expressed support for the policy by saying that it renders families who want to have more children with flexible freedom and opportunity. However, others held skepticism and used ironic jokes and dark humor to question the policy as the high real estate price in cities, high cost of raising children, burden of supporting older parents, and women’s difficulties in job market remain as severe problems for the middle-aged generation. Such concerns pose a huge barrier to the effectiveness of the new three-child policy. As there is no clue yet when the policy will come into effect, this study specifically explores the underneath reason regarding to social media use. In other words, this study explores in the digital era how social media drives people’s understanding of the policy, and how the understanding could, in turn, affect their supportive attitudes toward the policy.

This study takes the announcement of the three-child policy as a timely opportunity to examine how social media facilitates or hinders the governmental management of childbirth issues. It is particularly interested in questions of how users process social media news and discussions, what are their perceptions about the policy’s impact, and how such factors drive their willingness to support or oppose the policy. By addressing these questions, it enhances our understandings of how social media affordance is forming and changing China’s public sphere, presenting both opportunities and challenges for governmental issue management. This study also provides practical implications on how government communication could be more effective in forming supportive public opinions, especially in the digital era.

Governmental issue management

Issue management should be understood as an integrated continuum with crisis management. Jaques (2007) proposed an issue and crisis management relational model, emphasizing that risk, issue, and crisis do not occur in a strictly chronological order. Effective issue management could prevent a crisis, but a new list of issues and risks threatening the organization’s reputation could also be created afterward. Thus, effective issue management is critical for safeguarding government reputation, garnering public support, and maintaining the legitimacy of governance.

In China, the declining birth rate has emerged as a pressing social issue. The country’s recent census revealed a decline in the number of births for a fourth consecutive year, with only 12 million births recorded in 2020 (National Bureau of Statistics of China 2021). Officials worry that there will not be enough taxpayers to sustain pensions for the retired and elderly. In 2016, the country scrapped its decades-old one-child policy and implemented a two-child limit, but this policy change failed to lead to a sustained increase in births. The high cost of raising children in cities has deterred many Chinese couples from expanding their families. Therefore, in 2021, the Chinese government further announced that it would allow couples to have up to three children, which has now been formally passed into law, along with a list of measures aimed at boosting the birth rate and “reducing the burden” of raising a child.

Will the three-child policy become an effective measure in facilitating low-birth-rate issue management? The effectiveness of a communicator’s strategy could be reflected through stakeholder opinions, attitude changes, or other persuasion effects. Previous studies have examined how to adjust response strategies to meet the requirements of different people based on factors such as risk perception, emotion, personal involvement, expectations, values, and desires (Heath et al. 2009). Regarding the three-child policy, the factors (e.g., personal involvement, expectations, values, priorities) are contested among the citizens. Rather than straightly following the policy to boost birth rate, citizens are more concerned about obstacles such as childcare costs, employment discrimination against women, and insufficient protection of children’s welfare across various industries in China. Consequently, the newly announced three-child policy for the purpose of managing low-birth-rate issue may give rise to new issues, such as economic burdens, gender inequality, and policy inconsistency (i.e., in conflict with the prior one-child policy).

This study typically focuses on social media use and negative perceptions of the three-child policy in different dimensions. While an increasing number of studies could be found focusing on political discussions in China’s cyberspace (Harwit 2017), seldom studies empirically examine how cognitive processing of social media content could impact people’s political beliefs and opinions in China, particularly in relation to supporting or opposing governmental policies. Most importantly, the underlying dynamics of the indirect paths have not been fully documented. Incorporating the cognitive mediation model and theory of the influence of presumed influence (IPI) (Gunther and Storey 2003), this study offers insights into how social media in China either drives or hinders citizens’ support toward three-child policy. It also contributes to understanding how an issue management strategy could possibly give rise to other issues due to negative evaluations among the stakeholders. In addition, the role of institutional trust was also examined by further exploring the boundary conditions of social media exposure’s effects on supportive opinions.

The cognitive mediation model

The tradition of media effects studies was based on a “Stimulus-Response” framework to examine the direct effect of media exposure on news consumers’ political attitudes or behaviors, such as the agenda-setting effects (McCombs and Reynolds 2009) and cultivation theory (Morgan et al. 2009). However, the underlying mechanisms of how media use impacts political beliefs remain to be further explored. To address the gap, studies have examined possible factors that can mediate the effect of media use on political outcomes (Eveland 2001; Shahin et al. 2020). To examine whether and how social media influences political beliefs (i.e., policy support), this study adopts the cognitive mediation model (Eveland 2001). Being different from the simple “Stimulus-Response” framework, the cognitive mediation model explicates that media use indirectly affects behaviors through prompting different reasoning processes. Eveland (2001) held that there should be an indirect effect on news learning through information processing behaviors. In other words, when people encounter new information, the way in which they process the information will affect how they learn from the news.

In the current study, we take cognitive news elaboration as the cognitive reasoning process. Social media exposure is proposed to be situated as the stimuli in the model, which combines news coverage and online discussions (e.g., user-generated content). This approach is similar to Gil de Zúñiga et al.’s (2014) study, which takes social media use for news as stimuli when examining its effects on online and offline political participation through social media news expression. In this study, to operationalize social media exposure, we focus on three-child policy-related news and discussions on Chinese Weibo. Weibo is one of the largest social media platforms in China with more than 200 million users, which constitutes China’s “public sphere” with open debates and limited freedom (Tong and Zuo 2014; Yang and Fang 2021).

News elaboration has been widely recognized as a key mental process after being exposed to news information (Eveland 2002). In Kim et al.’s (2020) study on political persuasion, elaboration on cross-cutting perspectives is identified as the key mediator. Similarly, in our study, the information processing of three-child policy-related news and discussions was taken as the post-exposure reasoning process, which requires a motivated individual to actively follow the news on social media and then to cognitively elaborate on it. Studies showed that the increase in news exposure on political topics motivates and empowers the elaboration process (Wei and Lo 2008; Chen 2018). Thus, it is reasonable to assume that the more users read about news related to three-child policy, the more they will be motivated to take cognitive resources in processing it. The first hypothesis was proposed as follows:

H1: Social media exposure on China’s three-child policy will be positively associated with cognitive elaboration on the topic.

In this study, the dependent variable to be examined is policy support, which is operationalized as individuals’ supportive opinions toward the three-child policy as responses to news exposure, elaboration, and reasoned perceptions. News elaboration is expected to lead to a scientific and rational reflection of the policy. In China, as mainstream news available to the public are mainly pro-governmental, news elaboration on the relatively homogenous content could possibly lead people to be persuaded by the government discourse, to rethink future plans, and to consider the possibility of supporting and adhering to the policy. Therefore, a positive relationship between news elaboration and support for the three-child policy was proposed:

H2: Cognitive elaboration on three-child policy-related news and discussions will be positively associated with supportive opinions toward the policy.

In line with the cognitive mediation model, the mediating role of cognitive elaboration was further proposed in bridging the relationship between social media exposure and public opinion. However, it is still uncertain the extent to which the indirect effects could positively increase supportive opinion. Therefore, a research question was raised:

RQ1: Could cognitive elaboration mediate the relationship between social media exposure and supportive opinion toward the three-child policy in China, and how?

Perceived negative effects

One of the main cognitive elaboration outcomes is perception that plays significant roles in further promoting political actions or intentions. In this study, perceived negative effects of the policy will be examined as the cognitive outcome of elaboration, given that the potential negative effects of raising more children hinder the fertility willingness of China’s current middle-aged population, most of whom are the only child in their families. They are not only trying to survive themselves in cities with high living cost but also bear the responsibility of supporting their aging parents.

A focus on negative effects

The concept of perceived effects has been studied as the core element in the third-person effects theory (Davison 1983), stating that individuals tend to believe others are more likely to be affected by media messages than themselves. In this study, attention is paid to the perceived effects of the three-child policy rather than media messages, given that the extent to which people will support the three-child policy largely depends on how they evaluate the policy itself. In addition, as most studies take “perceived effects” as a general concept without distinguishing the characteristics and directions of the effects (Lo et al. 2017), few studies could be found to explicate the effects into different dimensions. As an exception, Hald and Malamuth (2008)’s survey on self-perceived effects of pornography consumption distinguished perceived positive and negative effects. They found that women participants reported more perceived negative effects than men did, while perceived positive effects were equally reported by both men and women.

Acknowledging that the effects of three-child policy could be both positive and negative to individuals and society at the same time, it is believed that the perceived negative effects may be a stronger predictor of people’s attitudes toward the policy. This assumption is in line with previous studies (e.g., Jisu et al. 2006). For instance, Jisu and colleagues (2006) found that people’s support for a ban on advertising is associated with their perceptions of negative effects on themselves and others. Lo et al.’s (2010) study on exposure to Internet pornography proved that female respondents’ greater perceived negative effects of Internet pornography on others drive them to support restrictions of Internet pornography. Upon the announcement of the three-child policy, public discourses on Chinese social media platforms largely concern its negative effects on economic pressure, family burdens, and female discrimination in job markets. Therefore, in this study, perceived negative effects of the policy will be considered as the main perceptual outcomes.

Differentiating perceived negative effects

In addition to a focus on perceived negative effects, this study proposes that the perceived effects on self (e.g., reducing one’s own quality of life), on the public (e.g., creating more burden on middle-aged Chinese), and on females (e.g., enlarging discrimination against females at workplaces) should be analyzed in separate paths as the three perceptions may prompt policy support with varying levels of effect power. According to Lo and Paddon (2000), perceived harm on others actually was a better predictor of support for pornography restrictions than the magnitude of the perceptual bias. Therefore, in this study, different types of negative effects will be examined separately in parallel.

Empirical evidence has showed inconsistent relationships between media use and perceived media effects, which could be either negative or positive (Gunther and Storey 2003; Lo et al. 2010; Park 2005). However, as social media provides a high-choice information environment with diversified opinions, it is easy for individuals to be exposed to discussions that emphasize the Chinese adults’ hardship in raising children as well as job discrimination against women. In addition, cognitive elaboration, as the post-exposure news processing process, could facilitate individuals in relating the three-child policy to their own experiences, families, and potential social consequences (Petty and Cacioppo 1986). It is reasonable to assume that the more cognitive resources people take in cognitive elaboration, the more they will understand the positive and negative impacts of such policy on individuals, society, and mothers, as a main news learning outcome (Eveland 2002; Wei and Lo 2008). Therefore, a positive relationship between news elaboration and perceived negative effects was proposed:

H3: Cognitive elaboration will be positively associated with perceived negative effects of the policy on self (H3a), on the public (H3b), and on females (H3c).

IPI theory

To examine the relationship between perceived negative effects and supportive opinion, this study adopts the IPI model as the theoretical framework, which posits that people tend to adjust their behavior based on their perceptions of how others are influenced by the message (Gunther and Storey 2003). The model incorporates two processes. First, individuals assume that other groups will be influenced by media messages after exposure (Eveland et al. 1999). Second, upon such perceived effects on others, they will adjust their intentions or behaviors as a response (Liao et al. 2016). For instance, a study on fake news effects found that people are more likely to support regulation against fake news if they presume that fake news influences both “others” and “me” (Baek et al. 2019). In another study, Park (2005) found that reading beauty and fashion magazines increased the desire for thinness among females through the presumed influence of the thin ideal on themselves and others.

Therefore, based on the well-tested IPI model, it is appropriate to assume that the perceived negative effects of the three-child policy on the public and females (who are more likely to be influenced by the policy) will decrease people’s support for the policy. It is also straightforward to expect a negative association between perceived negative effects on self and the intention to support the policy. Thus, the hypotheses on the negative relationship between the perceived negative effects and supportive opinion were proposed:

H4: Perceived negative effect of the three-child policy on self (H4a), on the public (H4b), and on females (H4c) will be negatively associated with supportive opinion toward the policy.

Taken together, a mediation model was proposed to examine the indirect effects of social media exposure on supportive opinion toward the three-child policy. Regarding the indirect effects of the proposed model, the second research question was raised:

RQ2: To what extent does social media exposure indirectly influence Chinese people’s support for the three-child policy through cognitive elaboration and perceived negative effects of the policy on self, the public, and females?

The moderating role of institutional trust

Trust is a multi-level concept with various dimensions and forms (Dietz and Den Hartog 2006). Generally, it is defined as the “willingness to make oneself vulnerable to another based on a judgment of similarity of intentions or values” (Siegrist et al. 2005, 147). In the realm of political science, trust typically refers to the confidence that citizens hold in a political sector’s capability and willingness to represent and act in their best interests. Citizens’ trust in government institutions is a critical element of political support and government legitimacy (Easton 1975), making it a primary driver of citizens’ positive evaluation of government policies (Liu and Mehta 2021). In this study, institutional trust is considered as an exogenous variable because it is long-term oriented and cultivated by an integration of historical, political, cultural factors, as well as media effects.

Previous studies demonstrated the moderating role of trust in cognitive and behavioral effects. For example, the effect of motivation on organic food purchasing intention is moderated by trust such that with a higher level of trust, people will be more likely to buy organic food as a result of self-regulated motivation (Tandon et al. 2020). Trust is also a well-tested boundary factor in moderating people’s risk perceptions such that with a higher level of trust, people will be less likely to think that they would suffer from the negative events (Viklund 2003). Heath et al. (1998) proposed in their study that information openness and accessibility are closely linked with trust and could effectively reduce uncertainties for communities under risk. Similarly, it is assumed that institutional trust could strengthen the indirect positive effects of social media exposure on supportive opinion while attenuating the indirect negative effects at the same time. A hypothesis was proposed:

H5: Institutional trust strengthens the positive effects of social media exposure on supportive opinion via cognitive elaboration (H5a), whereas attenuating the negative effects of social media exposure on supportive opinion via perceived negative effects on self (H5b), on the public (H5c), and on females (H5d).

The proposed moderated mediation model is visualized in Figure 1.

Figure 1. Proposed hypothetical model.

Method

Sampling

The data for this study were collected by Wenjuanxing (https://www.wjx.cn/), a professional Chinese survey platform, through an opt-in online panel of adults in mainland China who have read news and discussions related to the three-child policy via social media. The survey was conducted from June 7th to June 9th, 2021, 1 week after China’s announcement of the three-child policy. After quality check, 842 valid cases were obtained. Since the three-child policy is more relevant to married people who are still able to give birth and raise babies, this study typically focuses on adults aged between 20 (the minimum legal marriage age in China) and 50. After filtering the dataset based on the target age, 802 cases were finalized.

Of the sample, 49.8% are young people aged 30 or below (N = 399), 43.5% are between the ages of 31 and 40 (N = 349), and 6.7% are above 40 years old (N = 54). In terms of gender ratio, 46.1% are male (N = 370) and 53.9% are female (N = 432). Although the gender was not evenly distributed, the dataset is closer to reality as females are more interested and relevant to the three-child policy. Other descriptive statistics show that the education level of the sample is relatively high, with 85.8% having a university degree or above (N = 689), 10.7% having an associate degree (N = 86), and 3.3% being middle school or primary school graduates (N = 27). Most of the respondents (41.6%) have a monthly income of 5,001–10,000 yuan (N = 334), 16.1% between 3,001 and 5,000 yuan (N = 129), and 30.7% between 10,000 and 20,000 yuan (N = 246). Only 7.4% earn no more than 3,000 yuan per month (N = 59). A known value of 34.9% of the respondents live in first-tier cities (i.e., Beijing, Shanghai, Guangzhou, and Shenzhen). Details of sample profile are listed in Table 1.

Table 1 Sample profile

Although the sample does not represent the general Chinese population, it is believed that the better-educated urban people with higher income are the primary target group of the country’s three-child policy to boost fertility willingness. Therefore, the sample of this study is appropriate and meaningful for investigating policy support.

Measurement

Social media exposure

Respondents were originally asked during the past week, how often they read three-child policy-related 1) news and 2) discussions via Sina Weibo, WeChat, and Douyin (1 = “never;” 5 = “often”). Confirmative factor analysis (CFA) showed low factor loadings of the items on WeChat and Douyin. As Sina Weibo is the largest social media platform in China, the measurement on social media exposure was further adjusted to focus only on exposure to three-child policy-related news and discussions on Sina Weibo during a short period after the policy announcement. Scores on the two items were significantly correlated (r = 0.76, p < 0.001). Thus, we averaged the two items to indicate the level of social media exposure on the three-child policy (M = 3.78, SD = 1.12).

However, it is possible that exposure to news and discussions may trigger different cognitive processes and attitudinal outcomes given that discussions are likely to be more critical of the policy than news (Rauchfleisch and Schäfer 2015). Therefore, we also ran separate analyses with news exposure on Weibo (M = 3.85, SD = 1.17) and discussion exposure on Weibo (M = 3.71, SD = 1.23), which are included in Appendix A. To enhance the robustness of our results and explore the potential differences across platforms, we also ran separate analyses for WeChat exposure (M = 3.84, SD = 0.88) and Douyin exposure (M = 3.82, SD = 1.00), which are included in Appendix B.

Cognitive elaboration

Items on news elaboration were adapted from previous studies (Eveland 2001; Wei and Lo 2008). Respondents were given a 5-point scale ranging from 1 (totally disagree) to 5 (completely agree) to rate their agreement with statements like “after reading news about the three-child policy, I thought about the impact of having more children on my family” and “after reading discussions on the three-child policy, I linked the policy with my own situation.” Scores on the four items were averaged to indicate the level of cognitive elaboration on the three-child policy (M = 3.97, SD = 0.73).

Perceived negative effects

To measure perceived negative effects, respondents were given a 7-point scale ranging from 1 (totally disagree) to 7 (completely agree) to rate their agreement with a list of statements. Originally, there were three items for perceived negative effects on self, on the public, and on others. After checking the factor loading results of CFA, only two items were included in the final data analysis to ensure better measurement validity and reliability.

For perceived negative effects on self, the statements included “following the three-child policy will reduce the quality of my life” and “following the policy will put me under financial pressure.” Scores on these two items were averaged to form an index of perceived negative effects on self (M = 5.45, SD = 1.48). For perceived negative effects on the public, the statements included “following the policy will reduce Chinese people’s quality of life” and “the policy will bring economic pressure to Chinese families.” Scores on the two items were then averaged to form an index of perceived negative effects on the public (M = 4.89, SD = 1.34). For perceived negative effects on females, the statements included “the policy will negatively affect women’s life quality” and “the policy will negatively affect women’s situation in job markets.” Scores on the two items were averaged to form an index of perceived negative effects on females (M = 5.29, SD = 1.44).

Institutional trust

According to Bakker et al.’s (2018) measurement on trust in government, trust comprises six dimensions, including competence, openness, honesty, expertise, concern, and care. In this study, the focus is on competence in supporting families who are willing to follow the policy and raise three children. Following previous scholarship (Huang et al. 2017; Mayer et al. 1995), respondents were asked to what extent they characterize the government as efficient, skillful, and credible. Therefore, they were asked to indicate their level of agreement (1 = “strongly disagree;” 5 = “strongly agree”) on the five items such as “I have great confidence in the capabilities of the government sectors.” The scores were averaged to form a measure of trust in government (M = 3.20, SD = 0.99).

Supportive opinion

To measure individual support, respondents were given a 7-point scale ranging from 1 (totally disagree) to 7 (completely agree) to rate their agreement with two supportive statements like “I think this new policy benefits Chinese people” and three non-supportive statements like “I am afraid that the government cannot afford the cost of raising more children.” Scores on the three non-supportive statements were reversely coded and then averaged together with scores on two supportive statements to form an index of supportive opinion toward the policy (M = 3.46, SD = 1.42).

Statistical analysis

To test the proposed hypotheses on mediation, Hayes (2015)’s PROCESS macro model template 81 has been adopted. Ten thousand bias-corrected bootstrap samples and 95% confidence intervals (CIs) were employed. Statistical significance (p < 0.05) is achieved when lower bound (L.L.) and upper bound (U.L.) CI do not include zero. To avoid potential confounding effects and provide a more robust analysis, demographic factors including gender, age, education level, income, website news exposure, and issue importance were included in the analyses as control variables. To measure website news exposure (M = 4.20, SD = 0.81), respondents were asked to report their frequency of reading three-child policy-related news from news websites during the past week on a 5-point scale. To measure issue importance (M = 3.75, SD = 0.10), respondents were asked to rate their agreement with the statement “giving birth and raising children is an important issue for me” on a 5-point scale.

To further test the whole moderated mediation model, the syntax was further customized for model template 81 to first add institutional trust as the moderator on the path from cognitive elaboration to supportive opinion and then add institutional trust as the moderator on the path from perceived negative effects (on self, on the public, and on female) to supportive opinion.

Results

Survey reliability and validity testing

Before testing the hypothesized model, reliability and construct validity for each construct were examined. CFA with Mplus showed an acceptable model fit (χ²/df = 3.08, p < 0.000, TLI = 0.94, CFI = 0.95, RMSEA = 0.05). Although the χ2/df was significant, it is highly sensitive to the sample size. Other fit indices were all within the ranges of a good fit (Byrne, 2013). The standardized factor loadings of each latent construct were greater than 0.5 (p < 0.001), suggesting high convergent validity (Hair et al. 2010). According to Fornell and Larcker (1981), average variance extracted (AVE) measures the level of variance captured by a construct versus the level due to measurement error, the level of 0.5 is good. Composite reliability (CR) is a less biased estimate of reliability than Cronbach’s alpha, the acceptable value of CR is 0.7 and above. Discriminant validity could be achieved if the AVE for the latent construct was greater than the squared intercorrelation of any two variables (Fornell and Larcker 1981). Table 2 reports the AVEs, CR, Cronbach’s alpha, and factor loadings while Table 3 shows AVEs, correlations, and squared correlation coefficients of the constructs. Based on the above criteria, the measurements, despite comprising self-developed items, demonstrate acceptable reliability and validity.

Table 2. Reliabilities and confirmatory factor analysis properties

Note: Standardized coefficients reported.

***p < 0.001; AVE = average variance extracted; CR = composite reliability.

Table 3. AVE, correlations, and squared correlation coefficients

Note: **p < 0.01; the diagonal elements (bold) represent the AVE values; upper diagonal represents squared correlations of each construct.

Hypotheses testing

Hayes (2015)’s PROCESS macro model template 81 was adopted to test the mediated effects. The results (Model 5A in Table 4) show that there is no significant relationship between social media exposure and supportive opinion (B = 0.02, SE = 0.04, p > 0.05), which implies that social media exposure will not directly impact the extent to which people support the policy. However, the indirect effect was significant through various paths. The results further demonstrate that social media exposure is positively related to cognitive elaboration (B = 0.06, SE = 0.02, p < 0.01) while cognitive elaboration is positively associated with supportive opinion (B = 0.19, SE = 0.06, p < 0.01). Additional analyses were further conducted to test if Weibo news exposure and discussion exposure present different main effects on news elaboration. We ran separated models by only entering “Weibo news exposure” and “Weibo discussion exposure,” the results were consistent with the combined index as our independent variable (Weibo news: B = 0.06, SE = 0.02, p < 0.01; Weibo discussions: B = 0.05, SE = 0.02, p < 0.05), providing additional support to our findings on the relationship between social media exposure and news elaboration (see Appendix A). Thus, H1 and H2 were supported.

Table 4. The regression coefficients in the mediation model and moderated mediation model

Note: Cell entries are unstandardized coefficient with standard errors in parentheses.

*p < 0.05; **p < 0.01; ***p < 0.001.

In addition, there are three parallel paths bridging the negative effects of social media exposure on supportive opinion. Regression analysis shows that cognitive elaboration is positively related to perceived negative effects on self (B = 0.37, SE = 0.07, p < 0.001), on the public (B = 0.25, SE = 0.07, p < 0.001), and on females (B = 0.25, SE = 0.07, p < 0.001). Meanwhile, perceived negative effects on self (B = −0.24, SE = 0.04, p < 0.001), on the public (B = −0.17, SE = 0.04, p < 0.001), and on females (B = −0.25, SE = 0.04, p < 0.001) are all negatively associated with supportive opinion. H3 and H4 were supported.

Entering institutional trust as the moderator on the path between cognitive elaboration and supportive opinion, results (Table 4 Model 5B) show that the relationship between social media exposure on supportive opinion through cognitive elaboration is conditionally affected by institutional trust (B = 0.14, SE = 0.05, p < 0.01). Further entering institutional trust as the moderator on the paths between cognitive elaboration and perceived negative effects, results (Table 4 Model 5C) show that the newly added moderating effects are all non-significant. Specifically, the interaction of institutional trust and perceived negative effects (i.e., self, public, females) is not significantly associated with supportive opinion (self: B = −0.05, SE = 0.03, p > 0.05; public: B = −0.03, SE = 0.03, p > 0.05; females: B = 0.04, SE = 0.03, p > 0.05). Thus, H5 was only partially supported.

Comparing Model 5B and 5C, a very minor additional proportion of the variance for the dependent variable could be explained (ΔR² = 0.01). Therefore, the model was finalized as Model 5B as it is with good fit and more parsimonious (see Figure 2). Examining the indirect effects of Model 5B in great detail (see Table 5), social media exposure shows non-significant direct effects on supportive opinion (B = −0.0517, SE = 0.0315, CI = −0.1134 to 0.0101). The indirect positive effects of social media exposure on supportive opinion through cognitive elaboration are only significant among people with high institutional trust (trust > 2.87). RQ1 was answered.

Figure 2. Finalized moderated mediation model.

Table 5. Indirect effect of social media exposure on supportive opinion through cognitive elaboration, perceived negative effects on self, public, and female, moderated by institutional trust

Note: Bootstraps resample = 10,000. Estimates were calculated using the PROCESS macro (Model 81). Control variables are included in the analysis.

In the hypothesized model, three serial mediation effects were proposed between social media exposure and supportive opinion first through cognitive elaboration and then through perceived negative effects. Results of Model 5B show that social media exposure negatively affects supportive opinion through cognitive elaboration and perceived negative effects on self (B = −0.0032, SE = 0.0017, CI = −0.0071 to −0.0006), perceived negative effects on the public (B = −0.0019, SE = 0.0012, CI = −0.0047 to −0.0003), and perceived negative effects on females (B = −0.0026, SE = 0.0014, CI = −0.0059 to −0.0004). RQ2 was answered.

We also ran additional analyses to test the finalized model (Model 5B) with different independent variables. First, we examined if Weibo news exposure and Weibo discussion exposure played similar roles in the model. Results showed that both news exposure and discussion exposure could not directly affect supportive opinion but could exert indirect effects through different paths. The results of the separate models were consistent (see Appendix A). Second, we tested if the model stayed consistent across platforms. Results were also consistent with that shown in Figure 2. However, if institutional trust was not included as a moderator in the model (i.e., Model 4A), Douyin exposure showed direct positive effects on supportive opinion (B = 0.1121, SE = 0.0393, CI = 0.0350 to 0.1892) while Weibo (B = 0.0170, SE = 0.0368, CI = −0.0553 to 0.0892) and WeChat exposure (B = 0.0831, SE = 0.0458, CI = −0.0068 to 0.1730) did not have such direct effects (see Appendix B). The difference may be explained by the algorithm curation of Douyin and widespread Chinese nationalism on that platform (Chen et al. 2021). This result directs to future research to further examine how platforms mobilize deliberative and critical political discussions differently.

Discussions

As news and discussions regarding the three-child policy with diversified viewpoints flooded Chinese social media upon its announcement, this study explores how social media exposure impacts people’s attitudes toward the policy, through what channels, and in what directions.

Empirical results from this study suggest that social media exposure indirectly impacts support for three-child policy through different cognitive outcomes. Primarily, by examining the proposed cognitive mediation model with IPI theory, we found that social media exposure negatively predicts supportive opinion toward three-child policy first through cognitive elaboration and then through three separate mediators in parallel (i.e., perceive negative effects on self, on the public, and on females). In other words, social media exposure positively predicts news elaboration, which further motivates users to think about the negative effects of the three-child policy on different social groups. The more people elaborate on the three-child policy content, the more likely they will find the negative effects of the policy on economic burdens and women’s situation. The indirect paths through news elaboration and perceived negative effects make people less supportive toward the three-child policy.

However, if the path from social media exposure to policy support was not through perceived negative effects but only through cognitive elaboration, the effects could be positive. In addition, among people with high level of institutional trust, such positive indirect effects become significant and grow stronger. In other words, only if when people believe that the government is capable of supporting families who raise more children, elaboration on social media news and discussions could lead to higher policy support. However, the results show that institutional support fails in moderating the paths between perceived negative effects and supportive opinions.

It is also interesting to compare the three parallel indirect paths. As shown in Table 5, the indirect effect is strongest via cognitive elaboration and perceived negative effects on self, moderate strong via cognitive elaboration and perceived negative effects on females, and weakest via cognitive elaboration and perceived negative effects on the public. Although there are no significant pairwise comparison differences in terms of effect sizes (p > 0.05), comparing the absolute value of indirect effects still reflects a hierarchical order of cognitive elaboration outcomes. The highest order should be self-relevant while the lowest order should be the most general concerns. Thus, individual needs and gender equality are the most important issues to handle in terms of boosting fertility rate at present.

Apart from the empirical findings discussed above, this study has several implications for our understanding of the intersection between social media effects and public policy issue management, as well as for governmental policy making. Theoretically, this study presents a step in combining the cognitive mediation model with IPI theory to test the cognitive processes, which affect political opinion. In particular, it addresses the role of social media in affecting people’s understanding of the policy and their perceived effects of the policy, which further make a difference in their attitudes. A twofold mechanism has been identified. First, the mechanism of social media exposure’s positive effects on supportive opinions could be through cognitive elaboration with the boundary of high institutional trust. Second, the mechanism of social media exposure’s negative effects on supportive opinion could be through perceived negative effects on different levels and targets. Bearing the efficacy of encouraging interaction and participation rather than passive consumption, social media plays a significant role in facilitating individuals’ elaboration with news, which in turn makes the audience have a more comprehensive and profound understanding of social and political issues.

Moreover, the findings seem to reveal a complex scenario of the media environment in China, echoing the call for attention to authoritarian regimes that practice Internet censorship and media control. On the one hand, although the mainstream media discourse has helped the government in promoting the policy and persuading the younger generation to give birth to more children for the benefit of the country and of their own, elaborating on diversified social media content provides an opportunity to reflect on the negative effects of the policy on various levels and social sectors. Despite strong Internet censorship, Chinese netizens may not always conform as “good” citizens who unquestionably adhere to government policies. The more they conceive about the negative effects of the policy, the more likely that new social issues, such as concerns about the lack of gender equality and child welfare in China, are brewing in due course. On the other hand, as the whole media environment of China is pro-governmental, interactions with news could lead people to further believe in what the government proposed and why they should be encouraged to give birth to more children. Although Chinese netizens may question the lack of governmental support in raising more children, the news elaboration could lead them to rationalize the three-child policy as their reproductive freedom, which has been curtailed for decades under the one-child policy. Further research could be done to unpack the complex relationship among social media, censorship, and public opinion in non-western contexts.

This study also points to some practical directions for governmental issue management. First, social media has become a critical channel for the government to assess public opinion and the effectiveness of issue management in order to further adjust the strategies. An effective governmental communication should take into account netizens’ voices and discussions on social media. This finding also aligns with previous literature that controlling media scrutiny may be beneficial in effective governmental communication (Lee 2009).

Second, the study highlights the role of institutional trust. It implies that while institutional trust works well in promoting a supportive opinion toward public policy via processing the information in favor of the governmental discourses, it may lose the magic power when people have already generated negative perceptions of the policy through reflective thinking. Essentially, the government needs to pay more attention to managing the issues related to citizens’ perceived negative effects simultaneously when promoting the policy. In sum, this study suggests that to facilitate effective government communication in promoting a new policy, it is important to focus on limiting and attenuating the negative effects of the policy on individual citizens, specific social groups, and the public. Building, maintaining, and repairing institutional trust should be considered as a long-term goal and should be continuously assessed and reiterated throughout the administration and management processes.

Several limitations and possibilities for future studies should be highlighted. First, the data analyzed in this study were based on an opt-in online survey. The sample profile is comparatively better educated and wealthy, which may limit the generalizability of the model. Second, by analyzing only cross-sectional samples, definitive claims of causal relationships could not be made. Future researchers are suggested to adopt a multi-wave panel survey design to overcome such limitations or use experimental design to test the model with various socio-political topics. Last but not least, this study focuses on a controversial socio-political issue in China that is not politically sensitive. It is good since diversified opinions are permitted to exist online. However, the model tested in this study may not be applicable to other more stringent and mandatory policies (e.g., Covid-19 patient tracing, mass testing, lock down, and travel restrictions from 2020 to 2022). Further endeavors are needed in more specific but also generalized investigations on how social media use could form and change public opinion toward government policies.

Despite such limitations, based on the cognitive mediation model, the study tries to offer a more comprehensive framework for understanding social media effects on issue management effectiveness via cognitive elaboration and perceived negative effects on different dimensions. More importantly, trust, which could be affected by the effectiveness of government issue management, also plays a vital role in shaping public reactions and the outcomes of government communication.