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 https://www.jmir.org/, as well as this copyright and license information must be included.
The onset of the COVID-19 pandemic and the consequent “infodemic” increased concerns about Twitter’s role in advancing antivaccination messages, even before a vaccine became available to the public. New computational methods allow for analysis of cross-platform use by tracking links to websites shared over Twitter, which, in turn, can uncover some of the content and dynamics of information sources and agenda-setting processes. Such understanding can advance theory and efforts to reduce misinformation.
Informed by agenda-setting theory, this study aimed to identify the content and temporal patterns of websites shared in vaccine-related tweets posted to COVID-19 conversations on Twitter between February and June 2020.
We used triangulation of data analysis methods. Data mining consisted of the screening of around 5 million tweets posted to COVID-19 conversations to identify tweets that related to vaccination and including links to websites shared within these tweets. We further analyzed the content the 20 most-shared external websites using a mixed methods approach.
Of 841,896 vaccination-related tweets identified, 185,994 (22.1%) contained links to specific websites. A wide range of websites were shared, with the 20 most-tweeted websites constituting 14.5% (27,060/185,994) of the shared websites and typically being shared for only 2 to 3 days. Traditional media constituted the majority of these 20 websites, along with other social media and governmental sources. We identified markers of inauthentic propagation for some of these links.
The topic of vaccination was prevalent in tweets about COVID-19 early in the pandemic. Sharing websites was a common communication strategy, and its “bursty” pattern and inauthentic propagation strategies pose challenges for health promotion efforts. Future studies should consider cross-platform use in dissemination of health information and in counteracting misinformation.
Misinformation over social media contributes to the global growth in vaccine hesitancy. The World Health Organization (WHO) defined vaccine hesitancy as the “delay in acceptance or refusal of vaccines despite availability of vaccine services” [
Twitter has documented importance in setting the public and political players’ agendas [
Contrary to predictions about the death of traditional media in the age of the internet, and particularly following the rise of social media, studies indicate that they remain important. Traditional media have documented synergy with social media that amplify their mutual impact on the agenda in intermediate agenda-setting processes [
Studies on Twitter’s role in political agenda setting revealed an intermediate effect, in which the agendas of traditional media and Twitter were dissimilar, but exerted mutual influences [
Studies that used surveys to examine individual beliefs and intentions yielded important information on how audiences make sense of novel vaccines in the face of emerging pandemics, including their use of mental frameworks from previously known vaccines [
Several studies documented the role of Twitter in disseminating vaccine-related misinformation prior to the current pandemic [
An additional lacuna in research on vaccine-related communication on social media involves its focus on single social media platforms. Research documented that most social media users use multiple media sources and social media platforms [
These previous studies underscored the potential importance of cross-platform information sharing on Twitter. Analyzing both the URLs and the domains shared as external content on Twitter can provide insights into the type of specific content and information sources included in social media messages about vaccination. Despite this importance, vaccine-related cross-platform use over Twitter received limited scholarly attention. Examination of links to websites shared within vaccination-related tweets early in the COVID-19 pandemic can enrich knowledge by gaining a broader understanding of this communication. Empirical implications of this knowledge include informing strategies for evidence-based vaccine-related message dissemination over social media. Moreover, it can shed light on the role of traditional media in the era of social media, as well as on how social media are used in cahoots with vaccine-related communication and their dynamics over time. Finally, in view of the documented role of inauthentic propagation of vaccine-related content on Twitter [
The goal in this study was to examine website sharing in vaccine-related tweets posted to COVID-19 conversations in the 20 weeks following the declaration of the pandemic. Specifically, we analyzed tweets that were posted from February 1 (two days after the WHO declared the outbreak of COVID-19 to be a Public Health Emergency of International Concern) through June 23, 2020, and sought to examine the magnitude, temporal patterns, and content of websites shared within these tweets. This study will provide unique contributions to theory and practice. Our examined time frame took place prior to the development of the vaccine and the implementation of COVID-19 vaccination campaigns. Therefore, tweets posted during that period can indicate the degree to which vaccinations were included in COVID-19–related discourse from its inception. It will also reveal the information sources that were promoted through cross-platform link sharing. These findings have the potential to indicate the effectiveness of official health sources in leading the agenda as health information providers and the prominence of vaccine-opposing sources. It can also uncover some of the tactics of the vaccination-opposing movement over time and in response to this new, unexpected, global health threat. This understanding is important for advancing theories about the role of social media in public health crises, as well as for informing future policies, interventions, and dissemination of health information to address audiences’ informational and emotional needs.
Given the importance of vaccination-related discourse and website sharing within COVID-19 Twitter conversations, including an understanding of the sources of vaccination-related information and their spread, we posed the following research questions.
First, as this is, to our knowledge, the first study to examine external content sharing in the context of vaccination in early COVID-19 conversations, we were interested in understanding the magnitude of external content, the degree to which vaccinations were featured in conversations about COVID-19 early in the pandemic, and the prevalence and dynamics of website sharing within these tweets.
Therefore, we posed the first research question: What are the prevalence and dynamics of vaccination-related tweets, including website sharing, posted between February 1 and June 23, 2020, as part of COVID-19 conversations, as evident in the number of these tweets over time?
We were further interested in learning about agenda-setting processes that are demonstrated in this relatively new social media strategy of information source promotion. As websites’ domains, such as television networks, vlogs, or individual social media accounts, represent specific information sources, we aimed to learn about these sources and their characteristics. Specifically, in view of the importance of information sources in public health communication, we sought to identify the sources of the websites that were shared most prominently in the early months of the pandemic in tweets about vaccination that were posted to COVID-19 Twitter conversations.
Therefore, the second research question was posed: What are the characteristics of the 20 most-shared website domains?
In addition to the information sources, we were interested in exploring the content of the most-shared information, as evident in the 20 most-tweeted websites in our data set. The prominence of these websites can stem from users being activated by the content and their desire to share it. However, specific spread strategies and coordinated efforts might also drive this prominence. Therefore, we wanted to examine both the content of the websites and the specific information that was shared in this cross-platform modality, as well as their propagation.
Therefore, the third research question was posed: What characterizes the content and spread of the 20 most-shared websites?
The analysis encompassed two different data sets of COVID-19–related tweets. The first data set was based on a collection of tweet IDs gathered using general COVID-19–related keywords, such as “coronavirus” and “Wuhancoronavirus” [
Both data sets were then filtered to include only the dates of overlap (February 1 to June 23, 2020) and to remove any duplicated tweets across the data sets. We then filtered the data to include only English-language tweets. Given our interest in tweets about vaccination, each data set was filtered using the substrings “vax” and “vaccin.” This process ensured that the tweets included in our analysis referred to vaccinations. The resulting data set contained 841,896 English-language tweets. Since our focus was on analyzing content available to users rather than the identity of users, we did not attempt to distinguish between human users and machine accounts (ie, bots) [
Data set combination, filtering, and exclusion process.
We first extracted the website URLs from the JSON “entities” object of each tweet in order to get the original URL rather than the version automatically shortened by Twitter. URLs that were still shortened were unshortened to their original form using an API [
In addition to computational methods, mixed methods content analysis was conducted in analyzing the 20 most-tweeted websites to identify the source of the websites and their content. First, the sources of the 20 most-tweeted URLs and the dates they were posted were recorded. The content of the 20 most-tweeted URLs was coded using inductive qualitative methods using the constant comparative method [
The qualitative analysis followed a multistep iterative process. A coauthor with expertise in mixed methods research created initial codes and recorded memos. The initial coding involved line-by-line detailed reading of the data, aimed at understanding the different views and actions described in the different URLs and approaching coding in an inductive manner while remaining open to different potential theoretical directions emerging from the data [
In the third stage of the qualitative analysis, the overall theme of “politicizing vaccination” emerged. A graduate student with training in qualitative research followed this process independently by coding each of the 20 URLs using inductive coding first and then coding by the previous categories. They then provided a quotation from each URL to support the coding. As a final check on consistency of results, we evaluated the intercoder agreement for all of the URLs between the initial coding and the third stage of qualitative analysis. The use of constant comparative analysis [
To examine the particular Twitter spread strategies that were used to propagate the most-tweeted websites, we have applied social cybersecurity methods to identify coordinated link sharing and flooding (or spamming) of the websites by tweeters [
Our aims were to understand the magnitude, dynamics over time, content, sources, and spread of websites shared within vaccine-related tweets as part of COVID-19 Twitter conversations. The first research question centered on the prevalence and dynamics of vaccination-related tweets, as well as website sharing posted as part of COVID-19 conversations over time. The analysis revealed that these conversations demonstrated an overall growth in tweets that related to vaccination. It also showed that website sharing had distinct patterns compared to overall tweets about vaccination. As mentioned above, our data set contained a total of 841,896 tweets. As seen in
Our analysis revealed that 1 in 5 of the 841,896 tweets (n=185,994, 22.1%) contained at least one website. A total of 1 in 4 of the 524,998 users (n=128,408, 24.5%) tweeted at least one website. In comparison, only 19.4% (n=163,743) of all tweets contained at least one hashtag, and 23.0% (n=120,699) of users tweeted at least one hashtag. Additionally, 85.2% (n=717,150) of all tweets contained a mention of another user’s account (ie, using the “@” symbol to refer to another Twitter account), and 87.4% (n=459,038) of users tweeted at least one mention in a tweet. Of the mentions, only 12.1% (n=87,097) were replies (ie, when a user was directly replying back to the tweet of another user).
A total of 1 out of 5 of the tweets that mentioned vaccination (185,994/841,896, 22.1%) included links to websites. These websites included 11,311 unique website domains. Most domains (n=6962, 61.6%) were tweeted with only one unique website tweeted.
Counts of unique tweets, users, and website domains for all of the vaccination tweets.
Rates of retweets, hashtags, mentions, and website sharing in all of the vaccination tweets.
Research question 2 explored the 20 most-shared website domains, with a focus on the role of legacy media, social media, and public health sources. As displayed in
Websites associated with these 20 domains constituted 8.25% (n=6244) of all of the 75,642 websites tweeted. Two of the most-tweeted website domains—Raw Story and The Jerusalem Post—had most of their tweets from just one news story each. Notably, the top two most–individually tweeted websites were in this category, indicating that “viral” tweets can increase a domain’s popularity.
On average, each domain had 4.26 (SD 24.4) websites per domain, with a mode of 1 website, indicating that many of the domains only ever had one website associated with them. In total, 62.7% (n=6963) of all domains in the data had only one unique website associated with that domain. An exception was YouTube, which had the greatest number of unique websites of any domain, followed by the CDC website.
Top 20 most-tweeted website domains.
Website domain | Tweets, n | Percentage of all tweets with website URLs that originate from website domain, % | Unique websites per domaina, n | Type of domain and country of origin |
Raw Story [ |
13,261 | 7.1 | 101 | US online tabloid |
Reuters [ |
4347 | 2.3 | 577 | International news organization |
YouTube [ |
4106 | 2.2 | 2013 | International, US-based social media platform |
The Guardian [ |
3167 | 1.7 | 364 | UK newspaper |
The Jerusalem Post [ |
3140 | 1.7 | 92 | Israeli newspaper |
Bloomberg [ |
2374 | 1.3 | 173 | International, US-based news agency |
CNBC [ |
2161 | 1.2 | 282 | US television channel |
The Daily Mail [ |
2102 | 1.1 | 291 | UK newspaper |
CNN [ |
1888 | 1.0 | 302 | Multinational, US-based television channel |
The New York Times [ |
1716 | 0.9 | 332 | US newspaper |
The Cable [ |
1642 | 0.9 | 19 | Nigerian digital newspaper |
STAT [ |
1552 | 0.8 | 122 | Health-oriented US news website |
Business Insider [ |
1447 | 0.8 | 203 | US financial news website |
The Washington Post [ |
1348 | 0.7 | 209 | US newspaper |
BBC [ |
1320 | 0.7 | 106 | UK public service broadcast organization |
Sky News [ |
1301 | 0.7 | 147 | UK television news channel |
The Independent [ |
1280 | 0.7 | 172 | UK newspaper |
The Hill [ |
1270 | 0.7 | 152 | US newspaper |
New York Post [ |
1148 | 0.6 | 152 | US conservative-leaning tabloid |
Centers for Disease Control and Prevention [ |
1139 | 0.6 | 435 | US government health organization |
aThis is the number of unique websites that originate from the higher-level domain. For example, a news website can have several unique websites representing different news stories that all come from the same single news website domain.
The temporal analysis of the patterns of the 20 most-tweeted domains documented three distinct peaks of domain usage, as presented in
The 20 most-used domains overall by the number of tweets featuring each domain. Counts are normalized within each time period (week).
In contrast to the majority of the domains that were characterized by “bursty” activity, some domains, such as YouTube and The Guardian, had consistent usage over time. The 20 most–persistently tweeted domains are summarized in
As
The 20 most–persistently tweeted website domains.
Home domain | Type of domain | Days that a website from the domain was tweeted at least once, % |
YouTube [ |
Social media platform | 77 |
The Guardian [ |
UK newspaper | 76 |
The Daily Mail [ |
UK newspaper | 72 |
Centers for Disease Control and Prevention [ |
US federal health organization | 71 |
CNBC [ |
US television news channel | 71 |
The New York Times [ |
US newspaper | 70 |
Reuters [ |
International news agency | 70 |
Bloomberg [ |
US business newspaper | 67 |
STAT [ |
Health-oriented US news website | 66 |
Business Insider [ |
US financial news website | 65 |
Instagram [ |
Social media platform | 65 |
The Independent [ |
UK newspaper | 65 |
CNN [ |
Multinational, US-based television channel | 64 |
Google [ |
Multinational technology company | 64 |
New York Post [ |
US tabloid | 63 |
BBC [ |
UK public service broadcast organization | 61 |
MSN [ |
Web portal by Microsoft | 61 |
The Washington Post [ |
US newspaper | 61 |
BBC [ |
UK public service broadcast organization | 60 |
NCBI (National Center for Biotechnology Information), NIH (National Institutes of Health) [ |
US federal health organization | 59 |
The third research question explored the content and dynamics of spread of the 20 most-tweeted websites.
The 20 most-tweeted websites comprised around 13% of all the tweets containing websites (n=185,994) in the data set. The most-tweeted website, which was by Raw Story, accounted for almost half of these, with 6.6% (n=12,201) of the overall number of tweets containing websites. As
As an overarching theme, the qualitative analysis indicated that the content of these websites demonstrated politicization of vaccination. The single most-tweeted website was by the digital tabloid Raw Story, which described Republicans blocking the COVID-19 bill to avoid posing limits on pharmaceutical companies’ charges for the vaccine. This politicization of vaccination was evident in other websites. As seen in
The 20 most-tweeted websites.
Rank | Title of webpage | Source; type | Topic | Coding | Tweets, n | Date in 2020 |
1 | GOP blocking coronavirus bill — because it limits how much drugmakers can charge for a vaccine: Report [ |
Raw Story; US progressive-leaning tabloid | Republicans block coronavirus bill because it limits how much pharmaceutical companies can charge | 4. Political focus: |
12,201 | March 3 |
2 | Israeli scientists: 'In a few weeks, we will have coronavirus vaccine' [ |
The Jerusalem Post; Israeli newspaper | Israeli scientists are close to developing a COVID-19 vaccine | 3b. News on vaccine development that proved unfounded | 2780 | April 13 |
3 | Israeli researchers announce breakthrough on coronavirus vaccine [ |
The Cable; Nigerian online newspaper | Israeli scientists are close to developing a COVID-19 vaccine | 3b. News on vaccine development that proved unfounded | 1507 | February 9 |
4 | EU sets out plans for advance orders of coronavirus vaccines [ |
The Irish Times; Irish newspaper | Britain will not be included in European COVID-19 vaccine supplies | 4. Political focus: |
865 | June 11 |
5 | UK will not participate in EU’s coronavirus fast track vaccine scheme [ |
The New European; UK pro-Europe newspaper | Britain will not be included in European COVID-19 vaccine supplies | 4. Political focus: |
789 | June 12 |
6 | James Clapper refuses to testify to Congress in person ‘until there’s a COVID vaccine’ [ |
True Pundit; US fake news website | Former Director of National Intelligence refused to testify until there is a vaccine | 4. Political focus: |
530 | May 15 |
7 | AstraZeneca agrees to supply Europe with 400 million doses of COVID-19 vaccine [ |
Reuters; international news organization | AstraZeneca signed a contract to supply 400 million doses of COVID-19 vaccine to Europe | 3a. News on vaccine development that proved founded | 521 | June 13 |
8 | Read Pedagogy of the Oppressed by Paulo Freire (tweet by Joshua Potash, liberal leaning with 146,000 followers) [ |
Twitter; US-based social media platform | “Trump thinks we should use the flu vaccine to defend against coronavirus. We could not be in worse hands.” | 4. Political focus: |
512 | March 2 |
9 | Trump’s baffling coronavirus vaccine event [ |
Washington Post; US newspaper | Commentary on Trump’s interactions with vaccine makers | 4. Political focus: |
454 | March 3 |
10 | PETITION: No to mandatory vaccination for the coronavirus [ |
Life Petitions; petitions website | Petition to prevent mandatory COVID-19 vaccination | 1. Content that overtly advances doubts regarding vaccines’ efficacy and safety and the motives of those who fund, develop, and/or test them | 434 | January 23 |
11 | Meet the all-female team working to create a COVID-19 vaccine in Maryland [ |
WJLA: ABC News; local DC news affiliate | Team working on developing COVID-19 vaccine | 3b. News on vaccine development that proved unfounded | 428 | February 28 |
12 | US scientists have completed a coronavirus vaccine, Texas-based genetic engineering company claims [ |
The Daily Mail; UK newspaper | Texas-based scientists reported completion of COVID-19 vaccine development | 3b. News on vaccine development that proved unfounded | 391 | February 20 |
13 | Israeli scientists: 'In a few weeks, we will have coronavirus vaccine' [ |
The Jerusalem Post; Israeli newspaper (mobile version) | Development of COVID-19 vaccine in Israel | 3b. News on vaccine development that proved unfounded | 387 | April 13 |
14 | California lab says it discovered coronavirus vaccine in 3 hours [ |
Fox Business; US television channel | Development of COVID-19 in California in 3 hours, funded by Gates, with China | 3b. News on vaccine development that proved unfounded | 381 | February 13 |
15 | Experts baffled as Trump asks why they can't just use flu vaccines to prevent coronavirus [ |
Indy100; UK online newspaper | Experts are baffled at Trump suggestion to use flu vaccine | 4. Political focus: |
377 | March 3 |
16 | COVID-19 vaccine shipped, and drug trials start [ |
TIME; US news magazine | Moderna vaccine shipped, and trials start | 3a. News on vaccine development that proved founded | 363 | February 25 |
17 | Trump removes US from global initiative to develop coronavirus treatments and vaccines [ |
Raw Story; US tabloid | Trump removes United States from global initiative to develop COVID-19 treatment | 4. Political focus: |
348 | February 25 |
18 | Coronavirus: Africa will not be testing ground for vaccine, says WHOa [ |
BBC News; UK broadcast organization | WHO says Africa will not be a testing ground for COVID-19 vaccine | 3b. News on vaccine development that proved unfounded | 325 | April 6 |
19 | Here's why Obamacare would likely make any coronavirus vaccine free for patients — and prove critical in fighting the disease [ |
Business Insider; US online media company | COVID-19 vaccine will be available for free thanks to the Affordable Care Act | 2. Provaccination: the content focuses on the efficacy and/or safety of COVID-19 vaccines, as a way to end the pandemic | 319 | February 29 |
20 | Director of US agency key to vaccine development leaves role suddenly amid coronavirus pandemic [ |
STAT; US health news website | Rick Bright steps down | 4. Political focus: |
300 | April 21 |
aWHO: World Health Organization.
Of the 9 (45%) websites that covered news about advances in vaccine development, only 2 of the developments that were covered proved founded (#7 and #16 covered AstraZeneca and Moderna, respectively). TIME reported on the rollout of Moderna vaccine clinical trials (#16), stood out as the only website that provided medical framing of the content, and included explanations regarding the vaccine’s mechanism of using messenger RNA (mRNA). In contrast, 7 (35%) stories covered advancements that, at the time of analysis, were unfounded or did not come to fruition (#2, #3, #11-14, and #17). Notably, 3 of these included news on development of an Israeli vaccine that allegedly was 3 days away from the finish line and 90 days from approval. This information came from the Israeli Science and Technology Minister, and the sources were an Israeli newspaper in English and a Nigerian digital newspaper. Of the coverage of vaccine development that did not reach the market, only the report about the Novavax vaccine (#11) included information about the phase of the trial. The report on Greffex (#12) included some scientific information on the technology of the vaccine and a timeline that emphasized the lengthy process.
Only 1 (5%) website expressed explicit opposition to vaccination (#10). It consisted of a petition to block mandatory vaccination, which was included in a website domain that promoted petitions for conservative, right-wing causes. Similarly, only 1 (5%) website (#19), by Business Insider, a US online media company, provided provaccination framing by positioning vaccines as the way to end the pandemic.
Tweets spreading the 20 most-tweeted websites.
Title of webpage | Website | Unique tweets with URL, n | Unique texts of tweets, n | Unique users that tweeted URL, n | Mentions in tweets with URL, mean (SD) | Retweets of tweets with URL, n | Days between first and last tweet of URL, n | Tweets within first hour of first tweet, n |
GOP blocking coronavirus bill — because it limits how much drugmakers can charge for a vaccine: Report [ |
Raw Story | 838 | 437 | 786 | 0.31 (1.61) | 11,363 | 12 | 129 |
Israeli scientists: 'In a few weeks, we will have coronavirus vaccine' [ |
The Jerusalem Post | 710 | 352 | 612 | 0.33 (1.14) | 2070 | 78 | 20 |
Israeli researchers announce breakthrough on coronavirus vaccine [ |
The Cable | 31 | 20 | 26 | 0.71 (1.54) | 1476 | 3 | 4 |
EU sets out plans for advance orders of coronavirus vaccines [ |
The Irish Times | 78 | 41 | 77 | 0.21 (0.43) | 787 | 2 | 1 |
UK will not participate in EU’s coronavirus fast track vaccine scheme [ |
The New European | 50 | 34 | 48 | 0.08 (0.3) | 739 | 2 | 2 |
James Clapper refuses to testify to Congress in person ‘until there’s a COVID vaccine’ [ |
True Pundit | 20 | 13 | 20 | 0 (0) | 510 | 21 | 1 |
AstraZeneca agrees to supply Europe with 400 million doses of COVID-19 vaccine [ |
Reuters | 62 | 38 | 57 | 0.21 (0.57) | 459 | 1 | 5 |
Read Pedagogy of the Oppressed by Paulo Freire [ |
112 | 111 | 110 | 0.16 (0.49) | 400 | 2 | 7 | |
Trump’s baffling coronavirus vaccine event [ |
The Washington Post | 351 | 197 | 339 | 0.19 (0.75) | 103 | 66 | 19 |
PETITION: No to mandatory vaccination for the coronavirus [ |
Life Petitions | 241 | 90 | 156 | 0.50 (2.32) | 193 | 37 | 3 |
Meet the all-female team working to create a COVID-19 vaccine in Maryland [ |
WJLA: ABC News | 56 | 33 | 56 | 0.16 (0.49) | 372 | 18 | 2 |
US scientists have completed a coronavirus vaccine, Texas-based genetic engineering company claims [ |
The Daily Mail | 135 | 64 | 133 | 0.67 (0.75) | 256 | 26 | 22 |
Israeli scientists: 'In a few weeks, we will have coronavirus vaccine' [ |
The Jerusalem Post (mobile version) | 110 | 55 | 104 | 0.79 (3.99) | 277 | 51 | 1 |
California lab says it discovered coronavirus vaccine in 3 hours [ |
Fox Business | 225 | 123 | 215 | 0.24 (0.69) | 156 | 19 | 8 |
Experts baffled as Trump asks why they can't just use flu vaccines to prevent coronavirus [ |
Indy100 | 5 | 2 | 4 | 0.2 (0.4) | 372 | 0 | 3 |
COVID-19 vaccine shipped, and drug trials start [ |
TIME | 111 | 81 | 109 | 0.43 (1.16) | 252 | 63 | 1 |
Trump removes US from global initiative to develop coronavirus treatments and vaccines [ |
Raw Story | 17 | 13 | 17 | 0.52 (1.24) | 331 | 1 | 3 |
Coronavirus: Africa will not be testing ground for vaccine, says WHO [ |
BBC | 18 | 14 | 17 | 0.38 (1.16) | 307 | 1 | 2 |
Here’s why Obamacare would likely make any coronavirus vaccine free for patients — and prove critical in fighting the disease [ |
Business Insider | 230 | 127 | 222 | 0.64 (1.05) | 89 | 6 | 4 |
Director of US agency key to vaccine development leaves role suddenly amid coronavirus pandemic [ |
STAT | 24 | 11 | 24 | 0.58 (0.57) | 276 | 6 | 6 |
The Twitter propagation statistics of some of the websites show attempts at spreading the websites through inauthentic means. Specifically, a high percentage of the tweets that included links to The Cable website (#3) and to the STAT website (#20) had a high percentage of retweets only, an indicator of spreading by flooding or spamming of a website on a social media platform in an effort to get the website artificially trending and, consequently, to give more exposure to other social media users. In addition, the text of over half of the tweets that included the 20th most-tweeted website, STAT, was identical, which is indicative of coordinated, inauthentic link sharing. Evidence of such coordinated inauthentic link sharing was also present in tweets that included links to the Life Petitions website (#10), which also had relatively few unique tweet texts compared to the number of tweets of the website and the number of unique users that tweeted that website.
This study is the first to examine the prevalence, dynamics, and content of websites shared in vaccination-related tweets. We focused on tweets that were part of COVID-19 conversations over 20 weeks, following the WHO’s announcement of COVID-19 as a pandemic until June 23, 2020. The main finding of this study is the use of a cross-platform strategy for promoting politicization of COVID-19 vaccination well before the rollout of these vaccines. This politicization of content, promotion of unfounded “advancements” in vaccine development, and coverage of unsettling political plots that left much unexplained are likely, in turn, to contribute to a decrease both in the public’s knowledge of the science behind vaccine development and effectiveness and its trust in vaccination. Future studies should investigate the impact of exposure to this coverage.
Whereas previous research on the topic typically focused on the degree to which Twitter discussions reflected specific vaccine sentiments, our study indicates the politicization of the topic, which was shared by both progressive-leaning sources and content (eg, the Raw Story online tabloid) and legacy media (eg, The Washington Post) as well as by right wing–leaning sources, including legacy media (eg, The Jerusalem Post) and fake news sources (eg, True Pundit).
The websites that were tweeted represented diverse communication sources, with traditional news media making the top shared domains. Both the prominence of legacy news media in websites shared and the emergence of nontraditional media outlets, such as tabloids, vlogs, and other social media, exemplify processes of intermediate agenda setting in the new media environment. These processes were previously documented in political content [
In this new media environment, official health sources like the CDC and the NIH have had some success in disseminating their information, indicated by their inclusion in the lists of the most-tweeted domains (ie, CDC) and the most–consistently shared domains (ie, CDC and NIH). This is important, as governmental sources have been shown to provide credible, high-quality information compared to other sources, including the media [
Given the prominence of traditional media with their established gatekeeping, checks, and balances, it is not surprising that most stories shared in websites did not include obvious vaccine-opposing content. This finding is consistent with previous content analyses of tweets, reporting that vaccine-opposing content comprised a minority of the overall discussions on Twitter [
Our findings also point at the salience of international content. In addition to large, global media institutions like Reuters or CNN, and US-based newspapers and tabloids, some British newspapers, most notably The Guardian, were heavily tweeted in our data. Moreover, both the Israeli newspaper, The Jerusalem Post, and the Nigerian digital newspaper, The Cable, were included in list of the most-shared websites thanks to the viral story about an alleged Israeli COVID-19 vaccine. Similar to other smaller media organizations that were heavily tweeted, these two foreign, small newspapers demonstrated the opportunity of small players to advance their agenda in this new media environment by becoming “viral” through provision of sensational narratives. In the case of the Nigerian newspaper, inauthentic targeted attempts to spread this story contributed to its popularity, demonstrating the importance of deliberate manipulation in the new social media environment. It is likely that this rapidly changing environment, characterized by a “bursty” pattern of website sharing and the need to continuously provide new and sensational narratives as well as information and inauthentic spread strategies, poses unique challenges for governmental and official health sources. In addition, social media cross-platform sharing was evident in the prominence of YouTube as the third most-shared domain, as well as the most–consistently shared website over time. Future studies should further explore the content included in the different platforms, as well as users’ interpretation of this content and their motivation to engage in sharing it.
Finally, these findings are important in revealing patterns of propagation of this external content and these links. Despite the known presence of bots and other inauthentic propagation strategies of vaccination-related content on Twitter [
These results have important implications that can inform interventions, policies, and future research. At the most basic level, our findings indicate that sharing links to websites is a common strategy in Twitter conversations on the topic. In fact, shared websites were more common than hashtags, which have become synonymous with Twitter. Hashtags are frequently researched due to their use in creating discussion communities on social media [
Our analysis also revealed that websites were tweeted in a “bursty” pattern, indicating heterogeneity of a large number of sources, stories, and topics shared. The results regarding the increased number and diversity of external links shared at the time of data collection are consistent with other studies that documented the ebbs and flows of the “infodemic.” A recent study suggested that this increase was motivated by both uncertainty and state-sponsored propaganda [
The strengths of this study stem from its analysis of a large data set that was collected at a historically important period. Moreover, we employed a triangulation of computational methods and human coding to study a previously unexplored communication strategy within vaccine discourse on Twitter. However, this study is not without limitations. First, our tweets were collected by searching common vaccine-related keywords and hashtags. While these keywords and hashtags were identified following an extensive literature review and analysis of tweets by multiple research teams, it is possible that some emerging keywords and hashtags were not included. Future studies could apply additional computational methods, such as the Analysis of Topic Model Networks [
Additional limitations are grounded in our focus on tweets in English and on a specific time frame. Future studies should expand research to include additional languages and time frames, particularly during and following the rollout of the COVID-19 vaccine. In addition, some vaccine-related tweets, particularly those advancing vaccine-opposing messages, were deleted by Twitter by the time of analysis. Hence, the actual number of antivaccination tweets shared might be higher than what we were able to report, and their content might be somewhat different from what was collected. Moreover, we have focused on vaccine-related tweets that were part of the COVID-19 Twitter conversations. Although our data set is unique in including all related tweets rather than a sample, our findings do not apply to vaccine-related discourse on Twitter that was not part of the pandemic discourse. Moreover, our study focused on the first 20 weeks of the pandemic. Future studies should compare our findings regarding website sharing with similar content following the implementation of the COVID-19 vaccination campaigns globally. In addition, we focused on the content and propagation of the 20 most-tweeted websites and domains, and these findings might not apply to other links shared in this data set.
These findings are important in advancing understanding of website sharing in vaccine-related tweets, its use, and its dynamics. The analysis revealed that Twitter users share websites as part of their vaccine messages in COVID-19 conversations and that some of this sharing revealed inauthentic, deliberate attempts to spread this content. Our data included tweets that were posted in the first 5 months of the pandemic and showed that vaccine-related tweets were prominent in the pandemic-related Twitter discourse from its inception. Future research should examine the following months, as it is likely that with the advances in vaccine development, these conversations have increased in frequency and perhaps included different information sources.
The findings of this study pave the way for future studies that would answer additional questions. First and foremost, future research should expand the scope of this study by examining websites shared after June 2020, especially as new COVID-19 vaccines were approved and disseminated, and as information became available about their safety and efficacy. Given that our findings encompass the period prior to the approval and dissemination of the specific COVID-19 vaccines, they shed light on early communication on the topic rather than the specific risks and benefits of these vaccines.
In view of the global importance of the pandemic and vaccinations, future studies should also expand the scope of analysis to include additional languages other than English. In addition, it is important to consider, measure, and analyze additional aspects and implications of our work. For instance, to date, studies did not explore the impact of visual content on vaccine-related messages over social media. Future studies should expand the scope of this study’s analysis by exploring the visual content of vaccine-related tweets, websites, and of YouTube videos on the topic and the visual impact on propagation of the content over social networks. Similarly, future studies should explore the content and propagation of additional URLs in addition to the 20 most-tweeted websites explored in this study.
Finally, we call for hypothesis-driven communication interventions that would not only measure how and why antivaccination messages propagate over social media [
application programming interface
Centers for Disease Control and Prevention
messenger RNA
National Institutes of Health
World Health Organization
This research was, in part, funded by the Omar N Bradley Foundation through a General Omar N Bradley Research Fellowship in Mathematics Grant (IC) and was conducted with support from the University of New Mexico Women in STEM Faculty Development Fund (TG). We would like to thank Melanie Catron, Zohar Griffith, Kenneth Hafer, Paulina Majewska, and Rachel Taylor for their help in coding the data.
None declared.