Next Article in Journal
Acknowledgment to Reviewers of Psych in 2021
Next Article in Special Issue
Gender Gap in Healthcare Worker—Patient Communication during the COVID-19 Pandemic: An Italian Observational Study
Previous Article in Journal
The Interventions of a Mobile Mental Health Unit on the Refugee Crisis on a Greek Island
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Psych
Volume 4
Issue 1
10.3390/psych4010005
psych-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
Please note that, as of 22 March 2024, Psych has been renamed to Psychology International and is now published here.
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Brief Report

Engagement Analysis of Canadian Public Health and News Media Facebook Posts and Sentiment Analysis of Corresponding Comments during COVID-19

by
Melissa MacKay
*,
Taylor Colangeli
,
Sydney Gosselin
,
Sophie Neumann
and
Andrew Papadopoulos
Department of Population Medicine, University of Guelph, Guelph, ON N1G 2W1, Canada
*
Author to whom correspondence should be addressed.
Psych 2022, 4(1), 60-70; https://doi.org/10.3390/psych4010005
Submission received: 16 December 2021 / Revised: 3 January 2022 / Accepted: 6 January 2022 / Published: 9 January 2022
(This article belongs to the Special Issue The Uncertain Communication during the COVID-19 Pandemic: People’s Reactions and Coping Strategies)
Browse Figures
Versions Notes

Abstract

:
During the COVID-19 pandemic, key stakeholders have used social media to rapidly disseminate essential information to the public to help them make informed health-related decisions. Our research examined how the public responded to official actors’ Facebook posts during COVID-19 and examined the comment sentiment and post engagement rates. CBC News and CTV News received a greater proportion of negative comments and a lower average post engagement rate compared with Healthy Canadians. Additionally, the proportion of negative and positive comments varied over time for all sources; however, over 30% of the comments for all three actors were consistently negative. Key stakeholders should monitor the public’s response to their social media posts and adapt their messages to increase the effectiveness of their crisis communication efforts to encourage the adoption of protective measures.

1. Introduction

The COVID-19 pandemic is an unprecedented crisis within a globalized and highly connected society. Communication regarding the risks and recommended behaviors associated with COVID-19 is complex, due to the uncertain nature of the disease, the rapidly evolving conditions, and the corresponding infodemic [1]. In Canada, after the number of COVID-19 infections levelled off in the summer months, they started to steadily rise again in September 2020, with an all-time high number of new daily infections (n = 7894) reported at the end of November 2020 [2]. During a time in which the spread of the virus rapidly increases, public adherence to risk-protective measures is critical to reduce the burden of disease [3,4,5]. Confusion and lack of clarity about changing decisions and recommendations may contribute to a lack of adherence [3], making it essential for effective risk communication strategies to influence risk perception and the adoption of recommendations.
Recent research has reiterated the importance of transparent, targeted and tailored information and communication channels, and the repetition and clarity of messages to increase understanding and adherence to COVID-19 recommendations [3,6,7]. Crisis communication guidelines state that key stakeholders must provide an accurate and timely crisis communication response at the onset, and for the duration of an emergency, sharing information to encourage the public to make informed decisions to protect their health [8,9]. Crisis communication can shape public beliefs and risk perception, and influence people to partake in risk-protective measures by promptly communicating accurate and consistent information [9,10,11,12].
Multiple communication channels are essential for effective crisis communication, and audience needs must be taken into consideration [13]. Social media is one such channel that provides rapid and extended reach to various audiences [13,14]. Crisis communication through social media can be used to increase awareness and improve preparedness during public health emergencies [15,16], and to encourage the adoption of risk-protective behaviors [17,18,19]. Evaluation of crisis communication on social media channels through analytics such as engagement metrics and sentiment analysis can help key stakeholders monitor the effectiveness of their communications, and the reach of their efforts [20,21]. Facebook, the most used social media network in Canada, noted a significant increase in use during the COVID-19 pandemic [22]. Facebook is widely used by crisis actors in Canada, including public health officials, the government, politicians, and news media, and allows the public to participate in the crisis communication response [14].
The reactions of the public are essential to monitor and assess, as they give insight into the effectiveness of crisis communication [23,24,25]. It is possible to assess the public’s response to crisis messages with a social media rhetorical arena, where multiple public actors are crisis communicators [23]. Monitoring crisis rhetorical arenas, such as Facebook pages, can indicate message acceptance and the uptake of recommendations by crisis publics [23]. Engagement metrics and sentiment analysis are tools actors can use to monitor the public’s reactions to crisis messages [20,24].
Sentiment analysis is used to determine the emotional tone behind a series of words and emoticons [26], and identify how people are more likely to share and talk about the topics they experience emotions over [24]. The literature varies on how sentiment influences engagement. Some studies have found positive sentiment increases likes and comments [27], and others have found both positive and negative sentiments promote engagement [24]. During the COVID-19 pandemic thus far, sentiment analysis has been used to measure the public’s response to public health messaging [28,29]. There have been mixed results, as some studies found a positive public response to these messages [28,30], whereas others noted a greater negative public response [29,31,32]. Importantly, negative comments left by other Facebook users have been shown to have persuasive effects on the views of other users [33].
Additionally, users can interact with Facebook posts by reacting to posts (including likes), commenting on posts, and sharing the post within their own social network. Engagement metrics signal active interaction with the content; shares can organically increase content views, comments indicate conversation, and reactions can indicate someone’s opinion to the content [34]. Reactions and shares provide indirect social information to the public about the social media messages. Although indirect social information has not been found to influence people’s judgements of the message and source, as comments have been [35,36], they have been found to impact personal attitudes [35]. People use reactions, comments, shares, etc., to evaluate information they see on social media. Exemplars or single interactions on social media can be perceived as the opinion shared by others [37]. Importantly to crisis communication, this exemplification can influence the public’s perception of risks, and the effectiveness of the official response [38].
Our research describes the public’s engagement and emotional response to crisis communication on Facebook by Canadian public health and news media. The objectives of this research were to:
  • Classify the engagement rate of Facebook posts over time;
  • Assess the proportion of negative sentiment on comments over time;
  • Assess how trends in the proportion of trinary sentiment (positive, neutral, or negative emotional response) of comments may affect the total number of comments per post.
This research builds upon prior research examining the quality and content of Facebook posts for guiding principles for effective crisis communication, and how this influenced the public’s sentiment to the messages. The results of this research will allow crisis actors, including those of public health, to better understand the public’s reaction to COVID-19 messaging in the first wave in Canada, which will help inform current and future messaging, and the evaluation of public response.

2. Materials and Methods

2.1. Data Collection

This research did not require an ethics approval, as it was a retrospective infodemiology study of publicly posted information on social media.
The advanced search function on Facebook was used to manually collect posts from Healthy Canadians, CTV News, and CBC News pages, and the corresponding comments. Healthy Canadians is the sole federal public health Facebook page and CBC and CTV News are the most subscribed to national news sources among Canadians [39]. CBC is a taxpayer funded and government run corporation that provides a variety of news channels across Canada in English, French, and eight Indigenous languages [40]. CTV News runs both national and local programming in English and is owned by Bell Media [41]. Two searches were run on each page with common COVID-19 terms/concepts at the time: COVID and coronavirus and masks and social distance. All posts made between December 2019 and June 2020 were collected from the three pages that included the key terms/concepts. Posts written in English and those relevant to the national or provincial scope were included. Comments were included if they included text and were in English. Posts shared by one of the three Facebook pages, but authored by another source, were excluded.
Researchers manually copied post text, the date of post, number of reactions, number of shares, and number of comments into an Excel [42] spreadsheet on a shared drive. An identification number was created for each post. Screenshots of each post were also captured and stored in a shared drive. Comments were also copied into an Excel spreadsheet and the corresponding post ID was used to connect the comments to the original Facebook post.

2.2. Data Analysis

Similar to Calderon et al. [43], our research team used SentiStrength (Java, version, ersion 2.3.7110.19972), a sentiment analysis software that runs an algorithm to estimate the sentiment of short and informal pieces of text [44]. SentiStrength assigns words a sentiment strength value which is positive (+1 not positive to +5 extremely positive) or negative (−1 not negative to −5 extremely negative) [45]. The most positive and most negative values in the text are then compared, and an overall trinary classification of positive, neutral, or negative is provided [45]. SentiStrength provides a machine learning approach to sentiment analysis that is more accurate than other machine-learning approaches, due to the ability to decode non-standard spelling, and booster the strength of words [44].
Pre-assigned sentiment scores for some words were modified, as they could differ for highly specific topics, such as the COVID-19 pandemic [46]. After testing the changes for increased accuracy, the modified words changed to neutral sentiment included: cancer, death, dying, emergency, ill, infect, isolate, risk, and sick. We also altered the acronym and idiom lists to reflect those that frequently appeared in the comments, such as “bs” and “shut up”. Finally, the program’s spelling correction list, booster word list, negating word list, emoticon list, and standard settings were used.

2.3. Statistical Analysis

The post engagement rate was calculated by adding the number of comments, reactions, and shares for each post, dividing by the total followers, and multiplying by 100 to get a percent [47]. Next, to examine the change in comment sentiment overtime, we obtained the monthly proportions of negative, neutral, and positive comments by dividing the monthly totals of the negative, neutral, and positive comments by the total number of comments analyzed for sentiment per month.
Lastly, linear regression analyses were conducted to model any potential relationship between the total number of comments, and the sentiment score of the posts for Healthy Canadians and the news media (CBC and CTV News). Posts with no comments (n = 6) were excluded, as were posts with fewer than 10 comments, as they were deemed unlikely to add information to the regression analysis. Once the sentiment scores were calculated, the univariate linear regression analyses were conducted in R Version 4.0.3 (R Core Team, Vienna, Austria).

3. Results

A total of 438 posts and 26,774 anonymized comments related to COVID-19 were manually collected. Healthy Canadians posts accounted for 26% of the total posts and CBC and CTV accounted for 36% and 39%, respectively (Table 1). As reported in our related research, negative sentiment of comments on included posts was found to be highest for all sources [48]. The pattern of sentiment across sources was found to be statistically significant with comments on posts made by Healthy Canadians evenly distributed across positive, neutral, and negative sentiment, whereas posts made by the news media evoked approximately 50% more negative comments than positive [48].
On average, the news media posted more and had the largest number of comments per post (Table 1). Healthy Canadian posts received 8% of the total comments and CBC and CTV accounted for 42% and 49% respectively. Despite this, Healthy Canadians had an average post engagement of 0.3%, which is above the average engagement rate of 0.27% across sectors on Facebook [49]. News media average post engagement fell below the benchmark.

3.1. Change in Post Engagement Rate over Time

Post engagement rate across sources varied (Figure 1), with the highest post engagement rate (6.1%) found for a post made by Healthy Canadians in March 2020. The highest post engagement rate found for a post made by CTV News was 3.3% in April 2020 and 2.5% for CBC News in March 2020.

3.2. Change in Comment Sentiment over Time

Our research team explored the change in negative comments between 1 March and 14 June 2020 for Healthy Canadians and the news media (Figure 2). This period was selected as March was the month during which the novel coronavirus outbreak was declared a pandemic and June 14th 2020 was the time at which the data collection began.
Both sources experienced a similar proportion of negative comments between March and June, with slight fluctuations. During this time, the proportion of negative comments for Healthy Canadians ranged from 29.73% to 37.43%, and for the news media they ranged from 34.39% to 37.63%. Of note is that during May, the news media received a larger proportion of negative comments (34.39%) in response to their posts compared with Healthy Canadians (29.73%). Overall, the news media typically received a greater proportion of negative comments, specifically between April and June.

3.3. Relationship between Sentiment Scores and Total Number of Comments Per Post

For both Healthy Canadians and the news media, regardless of the number of comments on a post, there was no linear relationship between the calculated sentiment score and total number of comments for each post (Healthy Canadians, B = −4.2 × 10−6, p = 0.958; news media, B = 3.06 × 10−5, p = 0.267), as shown in Table 2.

4. Discussion

Overall, the news media received a greater proportion of negative comments and lower post engagement compared with Healthy Canadians, although negative sentiment was highest for all sources. Although Healthy Canadians posts accounted for approximately ¼ of the posts, they received only 8% of the total comments. There was no linear relationship between the calculated sentiment score and total number of comments for each post (Healthy Canadians, B = −4.2 × 106, p = 0.958; news media, B = 3.06 × 105, p = 0.267), as shown in Table 2.

4.1. Monitor and Increase Social Media Engagement

Our research found that Healthy Canadians had a slightly higher than average post engagement rate, whereas news media was below this average. The engagement rate takes into account the number of interactions and followers to understand how many people are connecting with the information. A past study found increased likes on a Facebook post resulted in more positive attitudes towards the flu vaccination [35]. Reactions, including likes, can increase how much attention the public pays to the post [35], making it important to understand and incorporate aspects of messaging that increase likes within key actor’s followers. A study on the platform Sina Weibo in China during COVID-19 found that social media functions, such as mentions and hashtags, as well as responding to comments, can positively influence sentiment [24]. Additionally, shares organically spread posts throughout follower networks [34] and, similar to reactions, actors should assess what makes their posts go viral and aim to include those features. The use of guiding principles for effective crisis communication should be used to increase message acceptance and uptake [48]. Additionally, hashtags, media features such as images and videos, mentions, and replies to comments have been shown to increase engagement [24,25].

4.2. Monitor the Sentiment of Comments on Social Media to Correct for Exemplars

The large proportion of negative comments observed for all sources was concerning when related to the exemplification theory. The public may interpret the high number of negative comments to represent the opinions held by the majority, negatively impacting their judgements of both the information and the source [37]. This difference in the proportion of negative versus positive comments may be explained, in part, by the public’s trust in the source itself. Previous research found that although the media is often the first source of information for many during a public health emergency [50], studies with representative samples of the general public in countries in North America, Oceania, and Europe find them untrustworthy [51,52,53,54]. Therefore, due to distrust of the source, the public may be more likely to respond negatively to messages shared by the media. Nonetheless, the public may also judge the trustworthiness of a source’s crisis communication by analyzing its content for timeliness and honesty [55]. Although research has also shown that the public places greater trust in public health during infectious disease outbreaks and pandemics [50,52,55,56,57,58], the public may still perceive their messaging to be of poor quality due to lack of timeliness, clarity, consistency, and transparency [3,6], leading to a more negative response. Furthermore, a recent study examining COVID-12 vaccine hesitancy found that distrust in government and public health is related to hesitancy [59]. The study found that messaging that is targeted and tailored to the specific concerns of this group can increase message acceptance and influence hesitancy [59].
No relationship was observed between the aggregate sentiment and total number of comments for each post. Public health and the news media should monitor the comments on their social media posts, assessing their sentiment to gauge how people are reacting to the content due to the possible impact they have on others’ perceptions of the information and source. Social media monitoring is used by public health to detect disease outbreaks, analyze disease trends, and conduct social listening where opinions and reactions to official communications are monitored [60]. Social media management platforms, such as Hootsuite [61] or Sprout Social [62], allow communicators to automatically monitor engagement, mentions, and comments, and respond within the management tool [15]. There is no one way to monitor and analyze data for public health social media monitoring. Public health agencies should clearly define the purpose for monitoring and how the data will be used [60] and use automated monitoring and management to analyze public perceptions and needs during various stages of a pandemic [24].
Additionally, the use of guiding principles for effective crisis communication to maintain trust can increase positive sentiment and should be consistently applied [48].

4.3. Pay Close Attention to Sentiment When Key Crisis Events Occur

A change in comment sentiment was observed across time for Healthy Canadians and the news media. Between March and June, both sources experienced fluctuations in the proportion of negative comments. This may have been impacted by the key outbreak events and public health measures that took place during this time. The uncertainty that may have arisen from the unfamiliarity of events and measures, along with inconsistent messaging concerning earlier recommendations, may have evoked fear and anxiety amongst the public [63], contributing to the large proportion of negative comments. A study on Facebook posts in Macao found that focusing on different types of messaging during the various stages of a pandemic is important for maintaining engagement [25]. In the acute phase, messages that focus on efficacy and rumor control result in higher positive engagement, whereas during the chronic phases, support and resources result in higher positive engagement [25].

5. Limitations

Facebook’s algorithm also impacts the reach of posts based on ranking signals based on “users’ past behaviors” [64]. Moreover, individuals who comment on the social media posts may tend to exaggerate their responses, as crises impact how people understand and process information [63,65].
Additionally, the largest proportions of Facebook users in Canada in 2020 are between 25 and 34 (25.9%) and 35 and 44 years of age (19.2%) [66]. Consequently, the results should be generalized cautiously, and other social media platforms should be explored.
SentiStrength is unable to detect sarcasm and irony, commonly seen among the comments, causing an increase in positive ratings [44].

6. Conclusions

News media received a greater proportion of comments with negative sentiment and had a lower overall average post engagement compared with public health. Public health had an above average engagement post rate, meaning their posts were liked, commented on, and shared more than news media. Although the comments’ overall sentiment did not impact the total number of comments per post, the influence that negative comments can have on the opinions and behaviors of others is of importance. Public health and the news media should monitor the public’s comments and engagement via social media, directly address negative comments, and employ best practices for effective crisis communication to ensure that the public accepts crisis messages and ultimately adopts recommended risk-protective measures. Future research of a similar nature should aim to include a larger sample size, and include other social media platforms. In addition, future research should include participants, to better understand perceptions regarding official actor crisis communications, and what increases positive engagement.

Author Contributions

Conceptualization, M.M., T.C. and A.P.; methodology, M.M., T.C., S.G. and A.P.; validation, M.M., T.C., S.G. and A.P.; formal analysis, M.M., T.C. and S.G.; investigation, M.M. and T.C.; data curation, M.M. and T.C.; writing—original draft preparation, M.M. and T.C.; writing—review and editing, M.M., T.C., S.G., S.N. and A.P.; visualization, M.M., T.C. and S.G.; supervision, A.P.; project administration, A.P.; funding acquisition, M.M. and A.P. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the University of Guelph COVID-19 Research Development and Catalyst Fund [number 054624].

Institutional Review Board Statement

Ethics approval was not required for this retrospective analysis of publicly available data.

Informed Consent Statement

Not applicable.

Data Availability Statement

Available upon request from the corresponding author.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Banks, K. In the Midst of the Pandemic, Academics Are Fighting a Rising ‘Infodemic’. University Affairs. 2020. Available online: https://www.universityaffairs.ca/features/feature-article/in-the-midst-of-the-pandemic-academics-are-fighting-a-rising-infodemic/ (accessed on 5 August 2021).
  2. Esri Canada. Canadian Outbreak At-A-Glance. 2020. Available online: https://resources-covid19canada.hub.arcgis.com (accessed on 7 December 2020).
  3. National Collaborating Centre for Methods and Tools. Rapid Review Update 1: What Are Best Practices for Risk Communication and Strategies to Mitigate Risk Behaviours? National Collaborating Centre for Methods and Tools. Available online: https://www.nccmt.ca/uploads/media/media/0001/02/5f7d164da82e9565106ae14b871bbe89b45606ad.pdf (accessed on 24 March 2021).
  4. Government of Canada. Coronavirus Disease (COVID-19): Prevention and Risks. 2020. Available online: https://www.canada.ca/en/public-health/services/diseases/2019-novel-coronavirus-infection/prevention-risks.html (accessed on 13 November 2020).
  5. World Health Organization. Coronavirus Disease (COVID-19) Advice for the Public. 2020. Available online: https://www.who.int/emergencies/diseases/novel-coronavirus-2019/advice-for-public (accessed on 5 January 2021).
  6. Ghio, D.; Lawes-Wickwar, S.; Tang, M.Y.; Epton, T.; Howlett, N.; Jenkinson, E.; Stanescu, S.; Westbrook, J.; Kassianos, A.P.; Watson, D.; et al. What Influences People’s Responses to Public Health Messages for Managing Risks and Preventing Infectious Diseases? A Rapid Systematic Review of the Evidence and Recommendations. PsyArXiv 2020. Available online: https://psyarxiv.com/nz7tr/ (accessed on 27 October 2020).
  7. MacKay, M.; Colangeli, T.; Thaivalappil, A.; Del Bianco, A.; McWhirter, J.; Papadopoulos, A. A Review and Analysis of the Literature on Public Health Emergency Communication Practices. J. Community Health 2021. Available online: https://link.springer.com/10.1007/s10900-021-01032-w (accessed on 25 October 2021).
  8. CDC. CERC Manual|Crisis & Emergency Risk Communication (CERC). 2018. Available online: https://emergency.cdc.gov/cerc/manual/index.asp (accessed on 12 August 2021).
  9. Henry, B. Canadian Pandemic Influenza Preparedness: Communications strategy. Can. Commun. Dis. Rep. 2018, 44, 106–109. Available online: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6449096/ (accessed on 4 October 2020). [CrossRef]
  10. Quinn, P. Crisis Communication in Public Health Emergencies: The Limits of ‘Legal Control’ and the Risks for Harmful Outcomes in a Digital Age. Life Sci. Soc. Policy 2018, 14, 4. Available online: https://lsspjournal.biomedcentral.com/articles/10.1186/s40504-018-0067-0 (accessed on 9 June 2021). [CrossRef] [PubMed] [Green Version]
  11. Seeger, M.W. Best Practices in Crisis Communication: An Expert Panel Process. J. Appl. Commun. Res. 2006, 34, 232–244. Available online: http://www.tandfonline.com/doi/abs/10.1080/00909880600769944 (accessed on 14 June 2021). [CrossRef]
  12. Vaughan, E.; Tinker, T. Effective Health Risk Communication about Pandemic Influenza for Vulnerable Populations. Am. J. Public Health 2009, 99, S324–S332. Available online: http://ajph.aphapublications.org/doi/10.2105/AJPH.2009.162537 (accessed on 14 June 2021). [CrossRef] [PubMed]
  13. Seeger, M.; Pechta, L.; Price, S.; Lubell, K.M.; Rose, D.; Sapru, S.; Chansky, M.C.; Smith, B.J. A Conceptual Model for Evaluating Emergency Risk Communication in Public Health. Health Secur. 2018, 16, 193–203. Available online: https://www-liebertpub-com.subzero.lib.uoguelph.ca/doi/abs/10.1089/hs.2018.0020 (accessed on 5 August 2021).
  14. Veil, S.R.; Buehner, T.; Palenchar, M.J. A Work-In-Process Literature Review: Incorporating Social Media in Risk and Crisis Communication. J. Contingencies Crisis Manag. 2011, 19, 110–122. Available online: http://onlinelibrary.wiley.com/doi/abs/10.1111/j.1468-5973.2011.00639.x (accessed on 2 February 2021). [CrossRef]
  15. Wendling, C.; Radisch, J.; Jacobzone, S. The Use of Social Media in Risk and Crisis Communication. 2013. Available online: https://www.oecd-ilibrary.org/governance/the-use-of-social-media-in-risk-and-crisis-communication_5k3v01fskp9s-en (accessed on 11 August 2021).
  16. Chan, M.S.; Winneg, K.; Hawkins, L.; Farhadloo, M.; Jamieson, K.H.; Albarracín, D. Legacy and social media respectively influence risk perceptions and protective behaviors during emerging health threats: A multi-wave analysis of communications on Zika virus cases. Soc. Sci. Med. 2018, 212, 50–59. Available online: https://linkinghub.elsevier.com/retrieve/pii/S0277953618303630 (accessed on 25 October 2021). [CrossRef] [PubMed]
  17. Choi, D.-H.; Yoo, W.; Noh, G.-Y.; Park, K. The impact of social media on risk perceptions during the MERS outbreak in South Korea. Comput. Hum. Behav. 2017, 72, 422–431. Available online: https://linkinghub.elsevier.com/retrieve/pii/S074756321730153X (accessed on 25 October 2021). [CrossRef]
  18. Hassan, M.S.; Halbusi, H.A.; Najem, A.; Razali, A.; Williams, K.A.; Mustamil, N.M. Impact of Risk Perception on Trust in Government and Self-Efficiency During COVID-19 pandemic: Does Social Media Content Help Users Adopt Preventative Measures? Review. 2020. Available online: https://www.researchsquare.com/article/rs-43836/v1 (accessed on 25 October 2021).
  19. Oh, S.-H.; Lee, S.Y.; Han, C. The Effects of Social Media Use on Preventive Behaviors during Infectious Disease Outbreaks: The Mediating Role of Self-relevant Emotions and Public Risk Perception. Health Commun. 2020, 36, 972–981. [Google Scholar] [CrossRef] [PubMed]
  20. Alamoodi, A.; Zaidan, B.; Zaidan, A.; Albahri, O.; Mohammed, K.; Malik, R.; Almahdi, E.; Chyad, M.; Tareq, Z.; Hameed, H.; et al. Sentiment analysis and its applications in fighting COVID-19 and infectious diseases: A systematic review. Expert Syst. Appl. 2021, 167, 114155. Available online: https://www.sciencedirect.com/science/article/pii/S0957417420308988 (accessed on 5 August 2021). [CrossRef]
  21. Jiang, H.; Luo, Y.; Kulemeka, O. Social media engagement as an evaluation barometer: Insights from communication executives. Public Relat. Rev. 2016, 42, 679–691. Available online: https://linkinghub.elsevier.com/retrieve/pii/S0363811115300461 (accessed on 25 October 2021). [CrossRef]
  22. PR Newswire. Facebook Reports First Quarter 2020 Results. CISION PR Newswire. 2020. Available online: https://www.prnewswire.com/news-releases/facebook-reports-first-quarter-2020-results-301049682.html (accessed on 5 August 2021).
  23. Coombs, T.; Holladay, S. How publics react to crisis communication efforts: Comparing crisis response reactions across sub-arenas. J. Commun. Manag. 2014, 18, 40–57. Available online: https://www.researchgate.net/publication/263270956_How_publics_react_to_crisis_communication_efforts_Comparing_crisis_response_reactions_across_sub-arenas (accessed on 5 August 2021). [CrossRef]
  24. Chen, Q.; Min, C.; Zhang, W.; Wang, G.; Ma, X.; Evans, R. Unpacking the black box: How to promote citizen engagement through government social media during the COVID-19 crisis. Comput. Hum. Behav. 2020, 110, 106380. Available online: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7151317/ (accessed on 30 December 2021). [CrossRef] [PubMed]
  25. Pang, P.C.-I.; Cai, Q.; Jiang, W.; Chan, K.S. Engagement of Government Social Media on Facebook during the COVID-19 Pandemic in Macao. Int. J. Environ. Res. Public Health 2021, 18, 3508. Available online: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8036686/ (accessed on 30 December 2021). [CrossRef] [PubMed]
  26. MonkeyLearn. Sentiment Analysis: The Go-To Guide. MonkeyLearn, 2021. Available online: https://monkeylearn.com/sentiment-analysis/ (accessed on 5 August 2021).
  27. Ji, Y.G.; Chen, Z.F.; Tao, W.; Cathy Li, Z. Functional and emotional traits of corporate social media message strategies: Behavioral insights from S&P 500 Facebook data. Public Relat. Rev. 2019, 45, 88–103. Available online: https://linkinghub.elsevier.com/retrieve/pii/S0363811118303680 (accessed on 30 December 2021).
  28. Kumar, A.; Khan, S.U.; Kalra, A. COVID-19 pandemic: A sentiment analysis. Eur. Heart J. 2020, 41, 3782–3783. Available online: https://academic.oup.com/eurheartj/article/41/39/3782/5873149 (accessed on 30 December 2021). [CrossRef]
  29. de las Heras-Pedrosa, C.; Sánchez-Núñez, P.; Peláez, J.I. Sentiment analysis and emotion understanding during the COVID-19 pandemic in Spain and its impact on digital ecosystems. Int. J. Environ. Res. Public Health 2020, 17, 5542. [Google Scholar] [CrossRef] [PubMed]
  30. Manguri, K.H.; Ramadhan, R.N.; Mohammed Amin, P.R. Twitter Sentiment Analysis on Worldwide COVID-19 Outbreaks. Kurd. J. Appl. Res. 2020, 5, 54–65. [Google Scholar] [CrossRef]
  31. Ji, X.; Chun, S.A.; Wei, Z.; Geller, J. Twitter sentiment classification for measuring public health concerns. Soc. Netw. Anal. Min. 2015, 5, 1–25. [Google Scholar] [CrossRef]
  32. Shen, Y. COVID-19 Outbreak: Tweet Analysis on Face Masks. Towards Data Science. 2020. Available online: https://towardsdatascience.com/covid-19-outbreak-tweet-analysis-on-face-masks-27ef5db199dd (accessed on 15 October 2021).
  33. Winter, S.; Brückner, C.; Krämer, N.C. They Came, They Liked, They Commented: Social Influence on Facebook News Channels. Cyberpsychology Behav. Soc. Netw. 2015, 18, 431–436. Available online: https://liebertpub.com/doi/10.1089/cyber.2015.0005 (accessed on 5 August 2021). [CrossRef] [PubMed]
  34. Grady, D.A.; Hollifield, A.; Sturgill, A. (Eds.) The Golden Age of Data: Media Analytics in Study & Practice, 1st ed.; Routledge: New York, NY, USA, 2019; Available online: https://www.taylorfrancis.com/books/9781000713909 (accessed on 6 August 2021).
  35. Peters, R.G.; Covello, V.T.; McCallum, D.B. The Determinants of Trust and Credibility in Environmental Risk Communication: An Empirical Study. Risk Anal. 1997, 17, 43–54. Available online: http://onlinelibrary.wiley.com/doi/abs/10.1111/j.1539-6924.1997.tb00842.x (accessed on 6 October 2020). [CrossRef]
  36. Lee, E.-J.; Jang, Y.J. What Do Others’ Reactions to News on Internet Portal Sites Tell Us? Effects of Presentation Format and Readers’ Need for Cognition on Reality Perception. Commun. Res. 2010, 37, 825–846. [Google Scholar] [CrossRef]
  37. Peter, C.; Rossmann, C.; Keyling, T. Exemplification 2.0: Roles of direct and indirect social information in conveying health messages through social network sites. J. Media Psychol. Theor. Methods Appl. 2014, 26, 19. Available online: https://psycnet-apa-org.subzero.lib.uoguelph.ca/fulltext/2014-10231-004.pdf (accessed on 6 August 2021). [CrossRef]
  38. Wagner, D. Managing Negative Comments Posted on Social Media. PhD Thesis, Walden University, Minneapolis, Minnesota, 2003. Available online: https://scholarworks.waldenu.edu/cgi/viewcontent.cgi?article=2555&context=dissertations (accessed on 6 August 2021).
  39. Anthony, K. For Canadians, Trust in News Media Has Fallen: Study. 2019. Available online: https://mediaincanada.com/2019/09/13/for-canadians-trust-in-news-media-has-fallen-study/ (accessed on 5 August 2021).
  40. Canada Guide. News and Media. The Canada Guide. 2021. Available online: https://thecanadaguide.com/basics/news-and-media/ (accessed on 30 December 2021).
  41. Canadian Communications Foundation. CTV Television Network|History of Canadian Broadcasting. History of Canadian Broadcasting. 2021. Available online: https://www.broadcasting-history.ca/listing_and_histories/ctv-television-network (accessed on 30 December 2021).
  42. Microsoft Corporation. Microsoft Excel. 2018. Available online: https://office.microsoft.com/excel (accessed on 5 August 2021).
  43. Calderon, N.A.; Fisher, B.; Hemsley, J.; Ceskavich, B.; Jansen, G.; Marciano, R.; Lemieux, V.L. Mixed-initiative social media analytics at the World Bank: Observations of citizen sentiment in Twitter data to explore “trust” of political actors and state institutions and its relationship to social protest. In Proceedings of the 2015 IEEE International Conference on Big Data (Big Data), Santa Clara, CA, USA, 29 October–1 November 2015; pp. 1678–1687. [Google Scholar]
  44. Thelwall, M.; Buckley, K.; Paltoglou, G.; Cai, D.; Kappas, A. Sentiment strength detection in short informal text. J. Am. Soc. Inf. Sci. Technol. 2010, 61, 2544–2558. Available online: https://onlinelibrary.wiley.com/doi/abs/10.1002/asi.21416 (accessed on 14 June 2021). [CrossRef] [Green Version]
  45. SentiStrength. SentiStrength Results. Available online: http://sentistrength.wlv.ac.uk/results.php?text=FLUS+MUTATE+year+after+year+that%27s+why+they+can+never+get+the+flu+shot+right+and+yet+we%27re+supposed+&submit=Detect+Sentiment (accessed on 1 August 2021).
  46. Thelwall, M. Heart and Soul: Sentiment Strength Detection in the Social Web with. 2013. Available online: https://www.semanticscholar.org/paper/Heart-and-Soul-%3A-Sentiment-Strength-Detection-in-Thelwall/2d5c5bce531b454a9a79eaabcb835be5fd977ea1 (accessed on 5 August 2021).
  47. Socialbakers. Formulas Revealed: The Facebook and Twitter Engagement Rate. Available online: https://www.socialbakers.com/blog/467-formulas-revealed-the-facebook-and-twitter-engagement-rate (accessed on 5 August 2021).
  48. MacKay, M.; Colangeli, T.; Gillis, D.; McWhirter, J.; Papadopoulos, A. Examining Social Media Crisis Communication during Early COVID-19 from Public Health and News Media for Quality, Content, and Corresponding Public Sentiment. Int. J. Environ. Res. Public Health 2021, 18, 7986. Available online: https://www.mdpi.com/1660-4601/18/15/7986 (accessed on 6 August 2021). [CrossRef] [PubMed]
  49. Jipa. 2021 Social Media Industry Benchmarks|Socialinsider. Socialinsider Blog: Social Media Marketing Insights and Industry Tips. 2021. Available online: https://www.socialinsider.io/blog/social-media-industry-benchmarks/ (accessed on 5 August 2021).
  50. Henrich, N.; Holmes, B. Communicating During a Pandemic: Information the Public Wants About the Disease and New Vaccines and Drugs. Health Promot. Pract. 2011, 12, 610–619. Available online: http://journals.sagepub.com/doi/10.1177/1524839910363536 (accessed on 18 August 2021). [CrossRef] [PubMed]
  51. Aylesworth-Spink, S. Falling in Line: News Media and Public Health Response during the 2009 H1N1 Outbreak in Canada. Ph.D. Thesis, Queen’s University, Kingston, Canada, 2015. Available online: https://qspace.library.queensu.ca/handle/1974/12948?show=full (accessed on 5 August 2021).
  52. Gray, L.; MacDonald, C.; Mackie, B.; Paton, D.; Johnston, D.; Baker, M.G. Community responses to communication campaigns for influenza A (H1N1): A focus group study. BMC Public Health 2012, 12, 205. Available online: http://bmcpublichealth.biomedcentral.com/articles/10.1186/1471-2458-12-205 (accessed on 21 October 2020). [CrossRef] [PubMed] [Green Version]
  53. Cloes, R.; Ahmad, A.; Reintjes, R. Risk Communication During the 2009 Influenza A (H1N1) Pandemic: Stakeholder Experiences from Eight European Countries. Disaster Med. Public Health Prep. 2015, 9, 127–133. Available online: https://www.cambridge.org/core/product/identifier/S1935789314001244/type/journal_article (accessed on 25 October 2021). [CrossRef] [PubMed]
  54. Kok, G.; Jonkers, R.; Gelissen, R.; Meertens, R.; Schaalma, H.; de Zwart, O. Behavioural intentions in response to an influenza pandemic. BMC Public Health 2010, 10, 174. Available online: https://bmcpublichealth.biomedcentral.com/articles/10.1186/1471-2458-10-174 (accessed on 25 October 2021). [CrossRef]
  55. Gesser-Edelsburg, A.; Mordini, E.; James, J.J.; Greco, D.; Green, M.S. Risk Communication Recommendations and Implementation During Emerging Infectious Diseases: A Case Study of the 2009 H1N1 Influenza Pandemic. Disaster Med. Public Health Prep. 2014, 8, 158–169. Available online: https://www.cambridge.org/core/product/identifier/S1935789314000275/type/journal_article (accessed on 21 October 2020). [CrossRef] [Green Version]
  56. King, C.L.; Chow, M.Y.K.; Wiley, K.E.; Leask, J. Much ado about flu: A mixed methods study of parental perceptions, trust and information seeking in a pandemic. Influenza Other Respir. Viruses 2018, 12, 514–521. Available online: http://doi.wiley.com/10.1111/irv.12547 (accessed on 21 October 2020). [CrossRef] [PubMed] [Green Version]
  57. Lyu, S.-Y.; Chen, R.-Y.; Wang, S.S.; Weng, Y.-L.; Peng, E.Y.-C.; Lee, M.-B. Perception of spokespersons’ performance and characteristics in crisis communication: Experience of the 2003 severe acute respiratory syndrome outbreak in Taiwan. J. Formos. Med. Assoc. 2013, 112, 600–607. Available online: https://linkinghub.elsevier.com/retrieve/pii/S0929664612005876 (accessed on 21 October 2020). [CrossRef]
  58. McFadden, S.M.; Malik, A.A.; Aguolu, O.G.; Willebrand, K.S.; Omer, S.B. Perceptions of the adult US population regarding the novel coronavirus outbreak. PLoS ONE 2020, 15, e0231808. Available online: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0231808 (accessed on 16 December 2020). [CrossRef]
  59. Benham, J.L.; Atabati, O.; Oxoby, R.J.; Mourali, M.; Shaffer, B.; Sheikh, H.; Boucher, J.C.; Constantinescu, C.; Parsons Leigh, J.; Ivers, N.M.; et al. COVID-19 Vaccine–Related Attitudes and Beliefs in Canada: National Cross-sectional Survey and Cluster Analysis. JMIR Public Health Surveill. 2021, 7, e30424. Available online: https://publichealth.jmir.org/2021/12/e30424 (accessed on 30 December 2021). [CrossRef]
  60. Fung, I.C.-H.; Tse, Z.T.H.; Fu, K.-W. The use of social media in public health surveillance. West. Pac. Surveill. Response J. 2015, 6, 3–6. Available online: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4542478/ (accessed on 30 December 2021). [CrossRef]
  61. Hootsuite. Social Media Marketing & Management Dashboard. Available online: https://www.hootsuite.com/ (accessed on 30 December 2021).
  62. Chen, J. Why Brands Need a Social Media Monitoring Strategy. Sprout Social. 2020. Available online: https://sproutsocial.com/insights/social-media-monitoring/ (accessed on 30 December 2021).
  63. Renyolds, B. Crisis and Emergency Risk Communication: Pandemic Influenza; US Department of Health and Human Services: Washington, DC, USA, 2007. Available online: https://emergency.cdc.gov/cerc/resources/pdf/cerc-pandemicflu-oct07.pdf (accessed on 6 August 2021).
  64. Cooper, P. How the Facebook Algorithm Works in 2021 and How to Work with It. Hootsuite. 2021. Available online: https://blog.hootsuite.com/facebook-algorithm/ (accessed on 6 August 2021).
  65. Centers for Disease Control and Prevention. CERC: Psychology of a Crisis; US Department of Health and Human Services: Washington, DC, USA, 2019; p. 16. Available online: https://emergency.cdc.gov/cerc/ppt/CERC_Psychology_of_a_Crisis.pdf (accessed on 6 August 2021).
  66. Statista. Canada Facebook Users by Age 2021. Available online: https://www.statista.com/statistics/863754/facebook-user-share-in-canada-by-age/ (accessed on 6 August 2021).
Psych 04 00005 g001 550
Figure 1. Engagement Rate of Facebook Posts by Source Between January 2020 and June 2020.
Figure 1. Engagement Rate of Facebook Posts by Source Between January 2020 and June 2020.
Psych 04 00005 g001
Psych 04 00005 g002 550
Figure 2. Change in negative comments for public health and news media over time.
Figure 2. Change in negative comments for public health and news media over time.
Psych 04 00005 g002
Table
Table 1. Total number of posts and engagement across sources.
Table 1. Total number of posts and engagement across sources.
SourceFollowers * (n)Number of Posts (n)Comments (n)Reactions (n)Shares (n)Average Post Engagement Rate (%)
Healthy Canadians352,822112221165,11142,2290.3016
CBC News2,688,92015711,554113,043123,0820.0707
CTV News977,63616913,00999,1521479970.1787
* Total Followers in August 2020.
Table
Table 2. Results from the univariate regression analyses for the total number of user comments per Facebook post about COVID-19 predicting the sentiment score for Healthy Canadians and the news media.
Table 2. Results from the univariate regression analyses for the total number of user comments per Facebook post about COVID-19 predicting the sentiment score for Healthy Canadians and the news media.
Healthy Canadians
BB (95% CI)SE(B)p Value
(Intercept)1.97(1.91, 2.03)0.03<0.001
Number of comments−4.2 × 10−6(−1.7 × 10−4, 1.6 × 10−4)8.04 × 10−50.958
News media
BB 95% (95% CI)SE(B)p Value
(Intercept)1.85(1.82, 1.87)0.014<0.001
Number of comments3.06 × 10−5(−2.35 × 10−5, 8.47 × 10−5)2.75 × 1050.267
Note. B represents regression weights. CI represents confidence interval. SE(B) represents the standard error.
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

MacKay, M.; Colangeli, T.; Gosselin, S.; Neumann, S.; Papadopoulos, A. Engagement Analysis of Canadian Public Health and News Media Facebook Posts and Sentiment Analysis of Corresponding Comments during COVID-19. Psych 2022, 4, 60-70. https://doi.org/10.3390/psych4010005

AMA Style

MacKay M, Colangeli T, Gosselin S, Neumann S, Papadopoulos A. Engagement Analysis of Canadian Public Health and News Media Facebook Posts and Sentiment Analysis of Corresponding Comments during COVID-19. Psych. 2022; 4(1):60-70. https://doi.org/10.3390/psych4010005

Chicago/Turabian Style

MacKay, Melissa, Taylor Colangeli, Sydney Gosselin, Sophie Neumann, and Andrew Papadopoulos. 2022. "Engagement Analysis of Canadian Public Health and News Media Facebook Posts and Sentiment Analysis of Corresponding Comments during COVID-19" Psych 4, no. 1: 60-70. https://doi.org/10.3390/psych4010005

APA Style

MacKay, M., Colangeli, T., Gosselin, S., Neumann, S., & Papadopoulos, A. (2022). Engagement Analysis of Canadian Public Health and News Media Facebook Posts and Sentiment Analysis of Corresponding Comments during COVID-19. Psych, 4(1), 60-70. https://doi.org/10.3390/psych4010005

Article Metrics

Back to TopTop