In this section, we present our results based on the research questions in the study design section.
4.2. Most Relevant Sources, Countries, and Publications
In this subsection, results are presented to answer the RQ2: What are the most relevant and influential sources, countries, and publications on social media research? Table 3
presents the top 10 productive sources based on the number of publications. These sources cover 23% of the total number of publications in our dataset. The top three journals that cover articles on social media research for COVID-19 issues are the Journal of Medical Internet Research
, the International Journal of Environmental Research, and Public Health and Sustainability
. We also can see that the first nine journals were open access except for Computers in Human Behavior,
and they also had a high impact as shown by the journal impact factors. This shows that most authors preferred to publish papers in high-impact and open-access journals during the COVID pandemic to be accessible to the research community quickly. Table 4
shows the top 10 most local cited sources. The local cited sources measure the number of local citations received by each cited reference within the reference lists of publications in the dataset. From these results, we can see that the Journal of Medical Internet Research
is the most cited source among researchers in the social media research field for COVID. This source has been cited 1492 times. The second highest cited source is Computers in Human Behavior
(1081 times), followed by PLOS One
(1072 times). This shows that these top journals are the main references in this research domain.
presents the top 10 most influential sources based on total citations. The total citations of a source is the number of citations received by published papers within the source in the dataset. We also note that Table 5
presents those sources that have at least one citation. From these results, we can see that the top three journals, the Journal of Medical Internet Research
, PLOS One
and the International Journal of Environmental Research and Public Health
, have published many citable articles and received a high number of total citations with a high h-index. However, there are also journals, such as Phychological Science
, Human Vaccines and Immunotherapeutics
and Phychological Medicine,
that have a high number of citations with a limited number of published articles on social media research for COVID.
demonstrates the country-specific scientific production of publications on social media use for COVID-19. Table 6
shows the details of the top 20 productive countries; SCP is Single Country Publication, MCP is Multiple Country Publication, and MCP ratio is the proportion of the total number of publications. These results show that the USA and China have published more than 150 papers and 67 of these involved international collaborations. The third highest production of publications is in the UK and 31 of these involved international collaborations. However, some observations can be made regarding the MCP ratio. Countries such as Japan, Malaysia, and Germany have higher degrees of international collaboration than other countries.
shows the most cited countries in the social media research field for COVID, and Figure 3
demonstrates the normalizing of the number of publications per country to their respective citations. Figure 2
shows that China is the leader in this research field, followed by the USA, UK, and Canada. The first four countries have a high impact because they have contributed a lot of research publications in this field. However, there are also countries such as Lebanon, Iraq, Switzerland, Vietnam, Bahrain, Bolivia, and Canada which seem to receive a high number of citations with a limited number of published articles (such as 3, 2, 6, 4, 1, 1, and 60 publications, respectively), as is shown in Figure 3
We divided the articles into research articles and review articles. The top 10 cited research and review papers based on total citations are presented in Table 7
and Table 8
, respectively. As we can observe from the results reported in Table 7
, the study by Gao, J. et al. (2020) [15
] has received the highest number of citations followed by the ones written by Pennycook, G. et al. (2020) [16
] and by Elmer, T. et al. (2020) [17
]. Based on content analysis of the top research articles, three themes are identified: 3 of the 10 papers focused on accessing mental health, 6 of the 10 focused on identifying infodemics, and 1 of the 10 focused on the evaluation of health information quality. Below is the analysis of these research papers.
The studies by Gao, J. et al. (2020) [15
], Elmer, T. et al. (2020) [16
], and Ni, M.Y. et al. (2020) [18
] refer to the impacts on mental health of social media exposure during the COVID pandemic. More specifically, the work by Gao, J. et al. (2020) conducted an online survey of 4872 Chinese citizens and showed that frequent social media exposure was associated with increased anxiety [15
]. The work by Elmer, T. et al. (2020) used two cohorts of Swiss undergraduate students, one that experienced the COVID crisis and one that did not, and they found that exposure to social networks, lack of interaction, and physical isolation were associated with negative mental health problems such as anxiety, stress, and loneliness [16
]. Finally, the work by Ni, M.Y. et al. (2020) conducted a similar online survey via social media platforms in China and they also reported mental health problems [18
The papers by Pennycook, G. et al. (2020) [16
], Cinelli, M. et al. (2020) [19
], Kouzy, R. et al. (2020) [20
], Allington, D. et al. (2021) [21
], Ahmed, W. et al. (2020) [22
], and Islam, M. et al. (2020) [23
] refer to infodemics. The term infodemics is a combination of accurate and inaccurate information about an epidemic, such as the COVID-19 pandemic [1
]. The inaccurate information may include misinformation, fake news, or rumors. More specifically, the work by Pennycook, G. et al. (2020) performed two studies to examine why people share and believe COVID-19 fake news [16
]. One study showed that people share fake news because they do not think about the accuracy of the content before deciding to disseminate it, whereas the second study showed that people with scientific knowledge and thinking can identify false information about COVID [16
]. The paper by Cinelli, M. et al. (2020) examined the spreading of fake news about the COVID pandemic on specific social media platforms and by users who have dealt with the topic [19
]. The study by Kouzy, R. et al. (2020) quantified the misinformation about the COVID-19 on Twitter by analyzing 673 English tweets. They also showed that 24.8% of tweets were misinformation and 12.3% of tweets from public health accounts were unverifiable information [20
]. The paper by Allington, D. et al. (2021) conducted three surveys of social media use in the UK and found a negative link between COVID-19 conspiracy beliefs and health-protective behaviors, as well as a positive link between COVID-19 conspiracy beliefs and the use of social media platforms [21
]. The work by Ahmed, W. et al. (2020) investigated 5G and COVID-19 conspiracy theories on Twitter by analyzing the content of 233 tweets and found that 34.8% linked 5G with COVID-19 and 32.2% denounced the conspiracy theory [22
]. Finally, the work by Islam, M. et al. (2020) analyzed 2311 reports about COVID in 25 languages from 87 countries and they found that misinformation was mainly driven by rumors, stigma, and conspiracy theories that were discussed on social media platforms [23
Finally, the paper by Puri, N. et al. (2020) refers to information quality against vaccine misinformation for COVID-19 [24
]. More specifically, this study examined the role of the propagation of vaccine hesitancy and proposed digital health strategies to overcome vaccine misinformation on social media platforms [24
Top 10 research articles by total citations.
Top 10 research articles by total citations.
|Authors||Article Title||Source Title||TC|
|Gao, J. et al. (2020) ||Mental health problems and social media exposure during COVID-19 outbreak||PLOS One||636|
|Pennycook, G. et al. (2020) ||Fighting COVID-19 Misinformation on social media: |
Experimental Evidence for a Scalable Accuracy-Nudge Intervention
|Elmer, T. et al. (2020) ||Students under lockdown: Comparisons of students’ social networks and mental health before and during the COVID-19 crisis in Switzerland||PLOS One||332|
|Cinelli, M. et al. (2020) ||The COVID-19 social media infodemic||Scientific Reports||295|
|Kouzy, R. et al. (2020) ||Coronavirus Goes Viral: Quantifying the COVID-19 Misinformation Epidemic on Twitter||Cureus||287|
|Puri, N. et al. (2020) ||Social media and vaccine hesitancy: new updates for the era of COVID-19 and globalized infectious diseases||Human Vaccines &|
|Allington, D. et al. (2021) ||Health-protective behavior, social media usage and conspiracy belief during the COVID-19 public health emergency||Psychological Medicine||251|
|Ahmed, W. et al. (2020) ||COVID-19 and the 5G Conspiracy Theory: Social Network Analysis of Twitter Data||Journal of Medical Internet Research||211|
|Islam, M. et al. (2020) ||COVID-19–Related Infodemic and Its Impact on Public Health: A Global Social Media Analysis||The American Journal of Tropical Medicine and Hygiene||211|
|Ni, M.Y. et al. (2020) ||Mental Health, Risk Factors, and Social Media Use During the COVID-19 Epidemic and Cordon Sanitaire Among the Community and Health Professionals in Wuhan, China: Cross-Sectional Survey||JMIR Mental Health||190|
We can also observe from the results reported in Table 8
that the three review studies that received the highest number of citations are Tsao, S.F. (2021) [1
], Shani, H.; Sharma, H. (2020) [25
], and Gabarron, E. et al. (2021) [26
]. Table 8
presents those review articles that have at least 10 citations. Based on content analysis of the top review articles, we identified that one of the five papers focused on an overview of the role of social media in the COVID era, two of the five focused on the impact of social media during the pandemic, one of the five focused on an overview of the misinformation on social media and one of the five papers focused on a review related to distance learning with social media.
The review paper by Tsao, S.F. (2021) conducted a systematic review of 81 papers from three databases (PubMed, Scopus, and PsycINFO) and identified six topics related to the role of social media in COVID-19 [1
]. The six topics were infodemics, public attitudes, mental health, detecting or predicting COVID-19 cases, government responses, and quality of health information in prevention education videos [1
]. The papers by Shani, H.; Sharma, H. (2020), and Verner Venegas-Vera, A.V. et al. (2020) outlined the positive and negative impact of social media platforms during the COVID pandemic on healthcare professionals and the general public [25
]. The work by Gabarron, E. et al. (2021) performed a systematic review of empirical publications on a specific topic of infodemics related to misinformation about COVID-19 on social media platforms [26
]. Finally, the paper by Cavus, N. et al. (2021) conducted a literature review of papers related to eLearning education with social media platforms during the COVID-19 era and highlighted the eLearning challenges and strategies for the sustainable educational use of social media by both institutions, teachers, and students [28
Top 5 review articles by total citations.
Top 5 review articles by total citations.
|Authors||Article Title||Source Title||TC|
|Tsao, S.F. (2021) ||What social media told us in the time of COVID-19: a scoping review||Lancet||82|
|Shani, H.; Sharma, H. (2020) ||Role of social media during the COVID-19 pandemic: Beneficial, destructive, or reconstructive?||International Journal of Academic Medicine||32|
|Gabarron, E. et al. (2021) ||COVID-19-related misinformation on social media: a systematic review||Bull World Health Organization||19|
|Venegas-Vera, A.V. et al. (2020) ||Positive and negative impact of social media in the COVID-19 era||Reviews in Cardiovascular Medicine||18|
|Cavus, N. et al. (2021) ||Efficacy of Social Networking Sites for Sustainable Education in the Era of COVID-19: A Systematic Review||Sustainability||12|
4.3. Research Topics and Keywords Trends
In this subsection, results are presented to answer the RQ3: What are the most common research topics and keyword trends on social media research for COVID-19? We present a thematic analysis to detect the main research topics in the field using a word cloud and a thematic map. To avoid deviant results, we removed the keywords inserted in the search query (such as terms related to social media and COVID-19). Figure 4
shows the word cloud for the 50 most common author keywords in the publications collection. The size of the keyword in the figure indicates the frequency of the keyword in the dataset. As we can see from the figure, the most common words determine the content of most studies in the collection. More specifically, the frequent keyword “pandemic” is the main topic since the papers in the collection address several issues about the COVID-19 pandemic. The keywords “sentiment analysis”, “machine learning”, “natural language processing” “topic modeling”, “content analysis”, and “text mining” show their importance and represent the main methodologies based on their conceptual meaning. These methodologies were used to analyze the public attitudes of social media users about the COVID-19 pandemic and vaccines. “Misinformation”, “public health”, “infodemiology”, “mental health”, “anxiety”, “infoveillance”, “vaccination”, “social networks”, “fake news”, “health communication”, “vaccine hesitancy” and “education” also show their importance. These keywords address several specific issues about the pandemic such as the dissemination of inaccurate information on social media platforms, the impacts of social media on users’ mental health during the pandemic, and the evaluation of the spreading of government messages on social media for the protection of public health.
To achieve further understanding, Figure 5
shows a thematic map based on author keywords, as proposed by Cobo et al. [29
]. For the thematic map, some parameters were fixed, such as the number of words (=500) and minimum cluster frequency (=3), while keywords inserted in the search query were removed. The thematic map consists of four quadrants according to their centrality and density values along two axes. The centrality measures the importance of a theme compared with other themes on the map. The density measures the development of internal links within a cluster represented by a theme. The size of the cluster indicates the number of occurrences of the keywords that it contains, and the position of the cluster is set according to the cluster centrality and density. The label of the cluster chosen by the Biblioshiny software corresponds to the most frequent keywords.
The upper-right quadrant involves motor themes that are important and well-developed for the field. In this quadrant, there are four clusters. Cluster 1 includes “sentiment analysis”, “analysis”, “sentiment”, and “topic modeling”, cluster 2 includes “infodemiology”, “infodemic”, “machine learning”, and “natural language processing”, cluster 3 includes “mental health”, “covid-pandemic”, “anxiety”, and “depression”, and cluster 4 includes “misinformation”, “fake news”, “information”, and “internet”. Therefore, these clusters focus on social media for surveillance and monitoring of public attitudes and perceptions, mental health, and the dissemination of inaccurate information and conspiracy theories on social media as in [20
]. The attitudes and perceptions of social media users were analyzed using machine learning methods such as topic modeling and sentiment analysis. The lower-right quadrant includes basic themes that are important for the field but not well-developed. In this quadrant, there is a cluster that is marginal with the lower-left quadrant. This cluster involves the keywords “pandemic”, “public health”, “sars-cov”, and “vaccination” and it concerns the sharing of health-related information by social media users to influence their decision-making about vaccination, i.e., vaccine hesitancy [24
]. The lower-left quadrant includes emerging or declining themes with low centrality and density. In this quadrant, there are three clusters. Cluster 1 includes “communication”, “content analysis”, “health communication”, and “crisis communication” and concerns the evaluation of government messages and other health information, and how announcements were consumed on social media platforms [35
]. Cluster 2 includes “social”, “media”, “fear”, and “physical activity” and it concerns studies addressing the links between the fear of COVID and social network use as well as the impacts of sharing physical activity experiences on social media platforms during COVID-19 lockdown [38
]. Finally, cluster 3 includes “health”, “social networks”, “public”, and “higher education” and it concerns eLearning education with social media during the pandemic for sustaining usage by students and faculties [28
4.4. Co-Citation Network
In this subsection, results are presented to answer RQ4: What are the main clusters of co-citations related to social media research for COVID-19? Figure 6
shows the co-citation network. This network was performed with a minimum degree of co-citation equal to three and a threshold of 50 network nodes. The node of a network was labeled by the first author and publication year of the paper whereas the edge of the network is the co-citation between two documents. The size of the node indicates the number of local citations received by the documents and the thickness of the edge represents the strength of co-citation ties. The color of the node shows the cluster with which the paper is associated.
From the co-citation network in Figure 6
, we can observe that there are three clusters of co-citations. The clusters were named based on the majority of the references belonging to them. The first cluster with a red color was named infoveillance and public attitudes
, focusing on empirical studies monitoring the attitudes and sentiments of social media users for the COVID-19 pandemic and similar viruses (i.e., H1N1). This cluster is the largest and it includes 23 works. The top three cited references of this cluster are represented by Chew et al. (2010) [41
], Abd-Alrazaq et al. (2020) [42
], and Chen et al. (2020) [43
]. The second cluster with a blue color referred to infodemics
, focusing on works addressing specific issues i.e., misinformation, fake news, and rumors on social media platforms. This cluster consists of 14 articles and the top three cited references are represented by Kouzy et al. (2020) [20
], Ahmed et al. (2020) [22
], and Cinelli et al. (2020) [19
]. The third cluster with a green color was named impacts on mental health
, focusing on mental health problems caused by exposure to social media. This cluster contains 12 articles and the three top cited references are by Depoux et al. (2020) [44
], Gao et al. (2020) [15
], and Ahmad et al. (2020) [45
]. We can also see that works within each cluster are interconnected. However, Figure 6
shows that the works of the red cluster linked more with the works of the blue cluster than the works of the green cluster. Therefore, there is a close relationship between the research areas of infodemics and infoveillance.