5. Conclusions

5.1. Summary of Findings
This paper is the first to investigate how Wuhan’s local government used official social media for citizen engagement during the COVID-19 crisis. We used crawler technology to collect all the Weibo posts released by Wuhan Release from 31 December 2019, to 19 April 2020. We came to the following conclusion.
First of all, in terms of content types released in Wuhan, we have identified eight main types. Among them, thanks to workers and news reports, are the two most important information types that Wuhan release would like to post. During the outbreak, medical workers from all over China came to Wuhan, helping to cure and eliminate the epidemic. The government was committed to adopting a positive attitude and making these selfless actions public. News reporting is a routine job of a new media account. This will be followed by government measures, scientific guidance, and the release of epidemic data, which will be made transparent by making daily updates available.
Secondly, in response to Hypothesis 1, we used the method of median statistics to explore the trend of the content of different content types in different stages of the epidemic and found some special phenomena. From the result, we find that various types of messages play a different important role in every stage of the crisis. However, the information on refuting a rumor was always essential for the crisis response. This suggests the authorities should release information depending on the varying needs of the public and the actual situation of the crisis. Stopping the spread would be a key step throughout the whole period of the crisis.
Thirdly, we use multivariate statistics to verify which influencing factors of Weibo can generate greater citizen participation (repost, reply, likes). Concerning Hypothesis 2, we find that the dialogic loop is partially associated with citizen participation. Because @other accounts provide users with a distinct form of dialogue, it positively influent repost and thump-up behavior of users. Because the analysis of the data shows that the form of forwarding is accompanied by short characters, sometimes even in the form of “forwarding Weibo”, indirect information will lead to the decline of citizen participation. The hashtag did not affect citizen engagement because 95.6% (n = 3412) of all posts contained one or more hashtags.
As for Hypothesis 3, the media richness of the Wuhan release is positively correlated with citizen participation. Among them, the addition of pictures and links to information has a significant positive effect, which is consistent with previous studies [1,59,79]. Photos and links convey information in a fast, intuitive way that produces more interaction (likes or reposts) than with other types of media. However, more videos will lead to fewer citizen commentary, combing with the concise and rapid characteristics of social media, people will not be to understand a specific case to open a video.
The findings show that for Hypothesis 4, the scale of content is significantly positively correlated with citizen participation. This is an exploratory variable, in previous studies, many people considered multiple variables, but in terms of length, there was little research on the impact of exploration content on citizen participation. We found that during the COVID-19 management process, the more words published by the government, the higher the users get engaged. This research result provides some references for future research.
Finally, as for Hypothesis 5, this research anecdotally pointed to a preference of the citizens for the Wuhan local government original information during the COVID-19. It could be explained that at first the COVID-19 crisis was a local health crisis, and the public prefers the information from local rather than other sources, because the former was more closely related with them.

5.2. Practical Implications
This study brings some evidence of the factors influencing citizen engagement on Weibo. Firstly, government departments should comprehensively take the crisis lifecycle into account when conducting crisis response, which including a clear understanding of the crisis situation and a full analysis of the public’s information demand on different stages of a crisis. Secondly, government agencies should selectively employ media richness theory and actively use the dialogic loop to enhance public engagement. Government accounts can take full advantage of social media functions, especially mentions @ and hashtags, the number of original posts, add pictures and text length appropriately, to increase interactions with the public, to improve the level of engagement. Therefore, the results can be used to optimize the government information release model during the COVID-19 crisis to explain the influence of antecedence on social media users. We also provide a new method to capture and analyze all Weibo data officially released by Wuhan local government during the crisis.
Additionally, our findings may have practical implications for community managers in the public sector who are responsible for online communication during crisis outbreaks. Thus, our research can help local governments improve their online communication strategies, revealing factors that lead to higher citizen responsiveness. It is also possible to improve the quality of the information release services provided by maintaining a smooth dialogue with citizens by meeting their needs.

5.3. Limitations and Future Research
This study has its limitations as well as offering some directions for future research. First of all, we exclusively focused on Weibo during a situation of COVID-19 in Wuhan. At the beginning stage of the crisis, the government may choose some communication strategies such as denial, evasion, and diminishment to maintain the image and reputation, which may be deleted as the main strategies changed, this phenomenon makes it difficult to have a complete data. Moreover, because of the heavy censorship, negative or sensitive comments may be screened or deleted, which decreased the engagement motivation of the citizens. However, the anonymous and networked nature of social media makes it impossible to have an overall control [102], and the citizens also came up with corresponding strategies under the limitations on their freedom. The government cannot limit the expressive power of the people [103]. In this research, we still could evaluate the engagement level through the citizen behavior, such as like, comment and repost. Secondly, another limitation that needs to be recognized is the influence factors with government information release and reposts, replies, and likes, all of which are considered proxies for citizen engagement, but combined with the actual context, they may have a different use. Such as the time of forwarding, the emotion of the reply [27], special expression, or symbols [104]. Further analysis is needed in future studies to reduce this bias. Thirdly, although this study was conducted at the local government level in Wuhan, its main contribution can be extrapolated to other regions and applied to comparisons with other countries and international regions, which will improve the generality and understanding of the results. Findings from this study can represent a starting point for further understanding these phenomena in a more systematic fashion.