This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Various online rumors have led to inappropriate behaviors among the public in response to the COVID-19 epidemic in China. These rumors adversely affect people’s physical and mental health. Therefore, a better understanding of the relationship between public emotions and rumors during the epidemic may help generate useful strategies for guiding public emotions and dispelling rumors.
This study aimed to explore whether public emotions are related to the dissemination of online rumors in the context of COVID-19.
We used the web-crawling tool Scrapy to gather data published by People’s Daily on Sina Weibo, a popular social media platform in China, after January 8, 2020. Netizens’ comments under each Weibo post were collected. Nearly 1 million comments thus collected were divided into 5 categories: happiness, sadness, anger, fear, and neutral, based on the underlying emotional information identified and extracted from the comments by using a manual identification process. Data on rumors spread online were collected through Tencent’s Jiaozhen platform. Time-lagged cross-correlation analyses were performed to examine the relationship between public emotions and rumors.
Our results indicated that the angrier the public felt, the more rumors there would likely be (r=0.48,
Our findings confirm that public emotions are related to the rumors spread online in the context of COVID-19 in China. Moreover, these findings provide several suggestions, such as the use of web-based monitoring methods, for relevant authorities and policy makers to guide public emotions and behavior during this public health emergency.
In December 2019, the earliest cases of COVID-19 were reported in Wuhan, Hubei Province, China. On January 23, 2020, several cities in Hubei Province were quarantined in an attempt to slow down community transmission of the disease. On January 30, 2020, the World Health Organization officially announced that the COVID-19 outbreak was a public health emergency of international concern [
Rumors have abounded since ancient times, but they have been the focus of researchers only since the Second World War [
In the context of COVID-19, various online rumors have led to inappropriate behaviors among people in response to the epidemic, which have adversely affected people’s physical and mental health [
Previous studies on public emotions during major public health emergencies have found that people usually experience negative emotions such as panic, anxiety, anger, grief, and disgust [
On January 20, 2020, China launched a monitoring and quarantine system to record daily information about COVID-19 cases. The epidemiological data, issued by the National Health Commission of People’s Republic of China (NHC), were collected at the grassroots level on a daily basis. For this study, we used the epidemiological data (specifically, the daily number of newly confirmed COVID-19 cases) from all provincial-level regions of China that were released by the NHC from January 20 to March 10, 2020.
Sina Weibo, the Chinese version of Twitter, is the largest social media platform in China. People’s Daily, with more than 116 million followers on Weibo, is one of the most influential and authoritative news media on Sina Weibo. In this study, we used the web crawler Scrapy to gather relevant data published by People’s Daily on Weibo between January 20 and March 8, 2020, as well as netizens’ comments under each Weibo post. Millions of real-time comment texts contain substantial emotional information; we identified and extracted these public emotions from the comments collected.
Although many sentiment analysis tools based on natural language processing or machine learning [
Tencent’s Jiaozhen [
(A) Daily confirmed cases of COVID-19 in China. Blue dots represent numbers for the whole country, red triangles represent numbers for Hubei Province, and black circles represent numbers for other provinces. (B) Calibration of the daily number of confirmed cases in Hubei Province. Red triangles represent the daily number of confirmed cases in the province before calibration, and black dots represent the daily number of confirmed cases in the province after calibration. The sharp peak disappeared after calibration. HB: Hubei Province.
According to the Hubei Health Committee, on February 12 and 13, 2020, the ratios of COVID-19 cases that were clinically diagnosed and those that were detected based on nucleic acid testing were 8.44 (13,332/1580) and 1.79 (3095/1728), respectively. Most of these cases were a result of long-term accumulation of suspected cases. The decline in the number of clinically diagnosed cases on February 13, 2020, suggested that the cumulative suspected cases had been processed within those 2 days. The corresponding ratios of COVID-19 cases that were clinically diagnosed and those detected via nucleic acid testing on February 14 and 15, 2020, were 0.89 (1138/1282) and 0.93 (888/955), which average at 0.91. Therefore, from January 20 to February 13, 2020, the calibrated daily number of newly confirmed cases (as shown in
As the data on daily epidemic situations and emotional comments published on Weibo showed an exponential distribution, the data were log-transformed to stabilize the variance of time-series before conducting further analysis.
We used time-lagged cross-correlation to examine the relationship between public emotions and the total number of rumors and that between public emotions and the rumors with different types of emotions. We also used Pearson correlation coefficient to examine these relationships, with the maximum time lag of 10 days and a
Among the 276 rumors collected, 176 (63.8%) were neutral rumors, 62 (22.5%) were fear-related rumors, 19 (6.9%) were happiness-related rumors, 12 (4.4%) were anger-related rumors, and 7 (2.5%) were sadness-related rumors.
(A) Cross-correlation between the daily number of newly confirmed cases outside Hubei Province and all rumors. (B) Cross-correlation between the daily number of newly confirmed cases in Hubei Province and all rumors. Blue dashed lines denote the 95% confidence intervals of the uncorrelated time series. HB: Hubei Province.
During the study period, anger (2,248,362/17,328,675, 12.97%) dominated public emotions on the internet, followed by fear (627,407/17,328,675, 3.62%), happiness (216,072/17,328,675, 1.25%), and sadness (195,708/17,328,675, 1.13%).
Trend of public emotions expressed on Weibo in the context of COVID-19 during the study period.
Next, we found that the peak correlation of anger and rumors was 0.48 (
Cross-correlation between emotions and all rumors. Cross-correlation between (A) anger and all rumors, (B) fear and all rumors, (C) happiness and all rumors, and (D) sadness and all rumors. Blue dashed lines denote the 95% confidence intervals of the uncorrelated time series.
Finally, we analyzed the relationship between the 4 types of emotions and different types of rumors.
Cross-correlation between fear and fearful rumors. Blue dashed lines denote the 95% confidence intervals of the uncorrelated time series.
To our knowledge, this is the first study to explore the relationship between public emotions and rumors during COVID-19 by using a web-based monitoring method. A large number of comments and rumors related to COVID-19 were obtained from the Weibo account of People’s Daily and Tencent Myth Busters by using the web crawler Scrapy. This web-based monitoring study overcomes the limitations of traditional survey methods and facilitates data-collection in a rapid and real-time manner during outbreak of a public health emergency. In particular, during a large-scale quarantine period, people are more likely to use the internet and social media to acquire and propagate the latest information about the epidemic [
In this study, statistically significant relationships were first observed among the daily number of newly confirmed cases and the total number of rumors both within and outside of Hubei Province, which is in line with a previous study that found that the uncertainty of sudden infectious diseases can lead to a large number of rumors [
Regarding the relationship between the 4 basic human emotions and 4 different types of emotional rumors, our results showed that fearful rumors were positively associated with fear. This result supports the argument presented by of Na et al [
This study, however, has several limitations. First, this study only focused on Sina Weibo for data collection. Other media platforms, such as WeChat in China and Twitter in other countries, were not included to acquire more extensive data. Second, batch-fetching data can only be obtained as a whole; hence, we cannot investigate the change in public emotions at an individual level. In future research, we may use other methods such as surveys or interviews to examine the trajectories of public emotions at an individual level. Third, this study analyzed the relationship between rumors and 4 basic human emotions, and further research could be extended to examine the correlations between rumors and more complex emotions such as anxiety.
Our findings provide several suggestions for relevant authorities and policy makers in guiding emotions of the public during public health emergencies. First, during a large-scale quarantine period, the authorities can use web-based monitoring methods to identify public emotions and behaviors in real time and provide timely guidance to channel public emotions and behaviors. Second, rumors are a catalyst for public emotions, and disproving them in a timely manner would be helpful to increase positive emotions of the public. Third, our findings showed that fearful rumors were associated with fear. Thus, media platforms should strengthen the monitoring of online rumors, identify and verify emotional rumors in a timely manner, and minimize the spread of fearful rumors to reduce fear among the public.
The National Health Commission of People’s Republic of China
We would like to thank all the research assistants who coded the online comments. Funding was provided by the National Social Science Foundation of China (No. 16CTQ029).
None declared.