2. Literature Review

2.1. Government Information Release and Crisis Lifecycle
SM has emerged as an important medium for governments and citizens to capture and explain crises, and take action accordingly [30]. On the one hand, SM platforms allow citizens to search and share reliable information. On the other hand, SM can enhance citizens’ abilities in understanding present situations and solving problems collaboratively due to its powerful capacity for inter-connectivity. Therefore, a two-way, responsive, cyclical communication process is obtained. The government can eliminate the middlemen in dialogue communication (such as journalists) [31]. Currently, public agencies are increasingly integrating SM into local administration communication strategies and interacting directly with citizens [32]. Especially when a crisis happened, individuals who access news would like to follow the principle of proximity. A national survey indicated that during the COVID-19, the Wuhan citizens, as the most suffered group, trend to pay close attention to the local news release [33]. This not only bring challenges to the local government, but also gives them opportunities to better handle the crisis response.
SM could capture the change of mood, values and attitudes of the public, which is particularly important for managers to understand the importance of social media platform [34]. During emergency management, with the help of SM, teasing out the timing of the crisis and understanding that kind of information is useful to the public would be helpful for the crisis management job. In a public crisis, the identification of what kind of information the public need is an essential aspect of the emergency response. It will also be important to define the different phases based on the behavior of the public and the characteristics of the event. According to Fink [35], crisis management focuses on a comprehensive process, which also is described as the theory of crisis lifecycle. Crisis moves through several stages with disparate characteristics and requirements for governments and first responders. There are various theories of the communication and crisis development of, for example, the three-stage model [36], the four-stage lifecycle [35], or the five stages of a crisis lifecycle [37]. Coombs provides an overview of his own three-stage crisis development model: pre-crisis (detection, prevention, and preparation), crisis (recognition and containment), and post-crisis. Some scholars differentiated four stages in a crisis lifecycle: two pre-crisis phases (mitigation and preparedness) and two post-crisis phases (response and recovery) [38].
In this research, we refer to the three-stage model due to its simplicity and the specific-stage features. The pre-crisis occurs when an impending or potential crisis is detected, the actors start to prepare and prevent the event. The second stage is the extreme of the crisis, crisis response occurs, and the authorities implement crisis reduction plans and seek to control the damage. The third and final phase emphasizes recovery, inventory, and learning. The purpose to divide crisis into several stages is focusing on how the authority and the public behave on the social media platform during the different stages of COVID-19 might help to understand the crisis itself. In the first stage, people are in an incubation state, in the second stage the greatest damage is inflicted and the authorities take more extreme mitigation and containment responses. The final stage starts the progress of recovery and returns to the normal.
The public needs of information shift constantly in response to the spread of the epidemic. For example, in the early stages of SARS, because of the contagious and insidious nature of the virus, the public wanted to know the truth about the outbreak and preventive measures related to their own safety. During the outbreak period, the public was concerned about the latest data on the outbreak and the government’s response. While at the end, the public turned to the attribution of responsibility and the development of vaccine drugs [39]. During the Zika epidemic, the public mainly focus on the access to the truth, effective government decision-making, ways to participate, resource allocation and maintaining a normal life [40]. Therefore, we come up with the first hypothesis:
Hypothesis 1.    The type of content that the public is most interested in has changed over time.

2.2. Dialogic Loop and Citizen Engagement on Social Media
If a government aims to achieve more citizen engagement, it must go beyond more publication of information [41]. Similarly, Zavattaro and Sementelli [42] claim that one-way communication cannot boost interaction, nor lead to an engaged society. Nevertheless, Mergel [41] has shown that public administration is still inclined towards a unidirectional communication model on SM.
Dialogue is regarded as “negotiated exchange of ideas and opinions” [43]. The dialogue can be incorporated into daily public relations practices, which could lead to positive outcomes, such as trust, mutuality, and empathy [44]. According to the theory, dialogic loop is regarded as a key dialogic principle for a positive two-way dialogic relationship [43], which refers that the content released by organization could stimulate public dialogue, and the organization could raise questions to promote engagement, adding providing the dialogue channels and responding to public feedback timely [27]. Dialogic loop could be embodied by many terms, e.g., with the use of hashtags and the @function [27,45], it is visible in the commenting and responding functions of the social media [46].
Few studies have indicated how organizations maintain dialogic relationships and promote public engagement through social media such as Facebook [47] and Twitter [48]. There is little empirical evidence of the actual impact achieved by the dialogic loop on levels of engagement in the Chinese context. Citizen engagement has been defined as the individual or collective behavior aimed at resolving social problems in the community [49,50]. The essence of it is in the interaction between citizens and the government [51]. By using this term, academics or professionals might refer to stakeholder involvement, co-creation, political participation, citizen engagement, participatory democracy, or activism [50,52,53,54,55]. Given the important role it plays in the social system, citizen engagement is a complex phenomenon that deserves further studies [56]. SM empowers the capacity of engaging people and organizations [42]. It has enabled wide engagement due to its ubiquitous and real-time access to information. People can share knowledge and situational awareness clearly and effectively through the government release on online platforms (uploading photos and records), thus promoting collaboration between citizens and the government.
Nowadays, the adoption of SM by governments has attracted the attention of academics and much research on this subject has been conducted [57,58,59,60] following diverse methodologies [32]. Several authors [61,62,63] have pointed out the importance of SM as a key tool to boost communication and dialogue with citizens. According to several authors [50,53,60,64], the use of SM positively influents citizens’ engagement. Nevertheless, Bonsón et al. [61] argue that only SM is not enough, and local governments must learn online communication strategy (like dialogic loop) and encourage two-way communication between citizens. Similarly, Skoric et al. [65] note that although the relationship between the use of SM and participation is crucial, citizen engagement is more complex than thought. Therefore, we come up with the second hypothesis:
Hypothesis 2.    Dialogic loop (the use of hashtag and @function) would generate a higher engagement rate.

2.3. Media Richness and Information Influencing Factors
Daft and Lenge [66] firstly come up with media richness theory, which gradually became popular with the emerge of electronic media. Media Richness theory regards communication channels as possessing many characteristics to carry information. The rich information is more able to reduce equivocality than lean information [66]. As a theory which refers to the relative ability of a communication channel to deliver messages [67], it includes four dimensions: (a) immediate feedback, (b) multiple cues, (c) language variety, and (d) personal focus [68,69].
Media richness theory has been used in multiple fields. In computer science fields, integrated media richness theory is used combined with flow theory, and the technology acceptance model to investigate how they can explain the acceptance of e-learning technology [70]. It is also an important factor influencing the customer experience in e-commerce [71]. This theory represents the structure of richness (e.g., multiple cues and responsive feedback) to determine the level of richness. Moreover, it also has been used to indicate the effect of different types of communication media (see, e.g., [72,73]).
However, on social media platforms, media richness theory research generally focuses on communicators’ choices of channels (e.g., a face-to-face meeting vs. a teleconference), few of them study whether rich media influent the effect of information transfer [66,74,75]. For example, visual images are more powerful tools for communicating messages regarding all aspects of organizations [44]. In particular, if a website uses multimedia functions and frames, it would have a higher level of media richness than text alone.
The multi-perspective medium could affect the effectiveness of the information disclosure. In other words, media richness theory assumes that media vary in degree of “richness”, or the “ability of information to change understanding within a time interval” ([66], p. 561). Thus, “richer” media are characterized by the use of responsive feedback, multiple cues, appropriate use of language, and a “tailored frame of reference” [76].
As discussed above, SM allows diverse multimedia content to be published, including images, videos, or hyperlinks. The importance of media for conveying information has been highlighted [77,78]. They argue that plain text is no longer the best type of medium to articulate information, and that multimedia is the most effective way to enhance the potency of a given message. Media analyses of Facebook also conclude that the use of pictures encourages citizen interaction, with more likes and comments than text-only publications [1,59]. Similarly, a media analysis study on Twitter [79] also shows that messages with a high level of richness, such as images and photos, tend to have a greater response from citizens. Therefore, we come up with the following hypothesis:
Hypothesis 3.    Higher media richness (picture, video, web-links) would generate a higher engagement rate.
In addition, we also assume that other information influencing factors potentially affect public engagement, which are as follows:
Text length: While some scholars found the relationship between text length and engagement, some research indicates that longer text could get more attention [80]. For example, an organization’s lengthy tweet could significantly reduce the negative sentiment of customers’ tweets [81]. It could be explained that longer sentence requires more time and resources, so it might positively influent users’ engagement and develop a strong online community [82]. Li et al. [83] found that, during the COVID-19 pandemic, the content length of most types of situational information has a positive correlation with the propagation scale. Another researcher assumed that a longer length content may include enough detail and have been recognized to positively affect the feedback [84]. Therefore, we come up with the following hypothesis:
Hypothesis 4.    Longer posts would generate more engagement than shorter posts.
Information source: In addition to releasing original information, the official account could also be operated as a news aggregator. Because of multiple source layers in online information transmission, the official account frequently cites some important information from other sources. Compared to the original source (the account that is actually responsible for creation of the core content), there are another type of posts with visible source, which can be defined as “the source seen by the receiver to be delivering the message or content” [85]. Until now, it is unclear that whether a post is original or not would influence the public engagement. Some similar research found that people are more likely to comply with the direct message from official source [86]. When facing a crisis or disaster, local departments were more important sources than routine times, and people regarded local government sources as references [87,88]. Therefore, we come up with the last hypothesis:
Hypothesis 5.    The original post will lead to more engagement than posts from the other source.