COVID-19疫情期間新聞媒體報導對民眾疫苗接種意願影響初探

Abstract

2020年全球爆發嚴重特殊傳染性肺炎(COVID-19)疫情，對公共衛生系統造成重大威脅，為避免新興傳染病影響國人健康，台灣政府於2020年1月20日成立中央流行疫情指揮中心(以下簡稱指揮中心)，成立期間歷時約1,197天，推動多項防疫措施，並依疫情狀況及防治方向分為三大階段，包括：阻斷傳播期、提升免疫期、防疫與經濟並重期；提升免疫期(2021年5月至2022年3月)，台灣積極提升疫苗接種涵蓋率，然而在疫苗接種作業推動過程中，接種情形起伏波動明顯，據相關研究文獻與報告指出，民眾疫苗決策態度受複雜原因影響，但與資訊取得來源(例如：媒體報導)有關；因此，本研究期運用指揮中心監測與蒐集之網路媒體報導資料，透過網路爬蟲與情緒分析將報導資料情緒進行量化，結合回顧當前時空背景下媒體關注重點議題，探討媒體報導對民眾疫苗接種意願的影響。 本研究採用Python適合中文文本分析的SnowNLP情緒分析工具進行資料分析與處理，量化後的情緒分數數值介於 0 與 1 之間，值越趨近於 0代表情緒越負面，趨近於1代表情緒越正面，0.5則為中立；量化的情緒分數與疫苗接種人次數據，則進一步透過Kruskal-Wallis檢定、事後分析t檢定、時間序列分析、Pearson相關性分析等統計方法進行檢定分析，以釐清媒體情緒變化與疫苗接種人次變化的相關性。 研究結果顯示，疫情期間6家網路主流媒體(包括：Ettoday新聞雲、聯合新聞網、中時電子報、蘋果日報網站、中央社即時新聞、自由時報電子報)共發布12萬9,004則COVID-19相關報導，以疫情初期報導數量較多，隨後逐年下降；整體新聞標題平均情緒分數為0.417，顯示媒體傾向使用較負面情緒的文字撰擬新聞標題；新聞內文平均情緒分數為0.447，相較於新聞標題情緒較中立及穩定；推論型統計發現，各媒體間的新聞標題情緒具顯著差異(K-W=420.36，p < 0.0001)，事後t檢定僅蘋果與整體平均，及聯合新聞網與中時電子報，2組情緒分數未達顯著差異，其餘組別均達顯著差異；分週別分析結果顯示，以自由時報電子報及聯合新聞網、自由時報電子報及中時電子報，發生顯著差異次數最多。 在媒體報導對疫苗接種意願的影響方面，研究顯示2021年6月份(疫苗接種第14週)中時電子報的情緒分數達顯著下降(t=-3.550，p<0.05) ，隨後第15週第1劑接種人次發生驟降，雖未達顯著，但經回顧新聞內容推論，與出現大量AZ疫苗不良反應報導，及網路謠傳疫苗不實訊息有關；Pearon相關性分析結果顯示，第15週中時電子報的情緒分數與疫苗第2劑接種人次達顯著負相關(r=-0.78，p<0.05)，且該媒體報導內容以批判政府疫苗相關防疫政策為主；第21週蘋果日報網站的情緒分數與疫苗第2劑接種人次亦達顯著負相關(r=-0.77，p<0.05)，內容以國際疫情、疫苗不良反應與疫苗供應不足報導為主；顯示該2個週別當接種人次越高時，2家媒體報導的情緒分數呈現越負面，推論媒體報導情緒與民眾接種意願確實具相互影響關係。 本研究結論顯示，COVID-19疫情初期，台灣網路主流新聞媒體大量報導疫苗不良反應等負面訊息，以及網路流傳疫苗安全性相關的不實訊息，影響民眾對特定廠牌的疫苗接種意願，為導致疫苗猶豫發生的原因之一；當疫苗接種量上升時，部分媒體隨之降低其報導負面情緒，且報導以批判政府疫苗防疫政策、疫苗不良反應、疫苗供應量不足等內容為主，推論媒體可能具特定立場，或受其報導風格特色影響所致；本研究強調了媒體報導情緒對公共衛生行為的重要影響，尤其是民眾疫苗接種決策過程，研究建議在應對未來疫情或公共衛生危機中，媒體應保持報導的客觀和中立性，避免過度渲染負面情緒，政府推動的公共衛生策略也應考慮媒體報導的影響，加強正確資訊的傳遞，避免因負面報導或不實資訊影響公共健康安全。

In 2020, the COVID-19 pandemic posed a significant threat to public health systems worldwide. To safeguard the health of its citizens against emerging infectious diseases, the Taiwanese government established the Central Epidemic Command Center (CECC) on January 20, 2020. The CECC operated for approximately 1,197 days, implementing numerous epidemic prevention measures and dividing the response into three major phases based on the status and direction of the epidemic: transmission blocking, immunity enhancement, and balancing epidemic prevention and the economy. During the immunity enhancement phase (May 2021 to March 2022), Taiwan actively increased its vaccine coverage rate. However, the vaccination process experienced notable fluctuations. According to related studies and reports, public vaccine decision-making attitudes are influenced by complex reasons, including information sources such as media reports. Therefore, this study uses data from the CECC's monitoring and collection of online media reports. Through web crawling and sentiment analysis, the study quantifies the emotional content of the reports, combines it with a review of key issues under the current context, and explores the impact of media reports on public willingness to receive vaccinations. This study employs SnowNLP, a sentiment analysis tool suitable for Chinese text analysis in Python, for data processing and analysis. The quantified sentiment scores range between 0 and 1, where values closer to 0 indicate more negative emotions, values closer to 1 indicate more positive emotions, and 0.5 represents neutrality. The quantified sentiment scores and vaccination numbers are further analyzed using statistical methods such as Kruskal-Wallis test, post hoc t-test, time series analysis, and Pearson correlation analysis to clarify the relationship between changes in media sentiment and vaccination numbers. The study results show that during the pandemic, six major online media outlets (including Ettoday, United Daily News, China Times, Apple Daily, Central News Agency, and Liberty Times) published a total of 129,004 COVID-19 related reports. The number of reports was higher in the early stages of the pandemic and gradually decreased over time. The overall average sentiment score of news headlines was 0.417, indicating that media tended to use more negative emotions in writing headlines. The average sentiment score of news content was 0.447, which was more neutral and stable compared to the headlines. Inferential statistics found significant differences in sentiment among the headlines of different media outlets (K-W=420.36, p < 0.0001). Post hoc t-tests showed no significant difference only between Apple Daily and the overall average, and United Daily News and China Times, while other groupings showed significant differences. Weekly analysis indicated that significant differences occurred most frequently between Liberty Times and United Daily News, and Liberty Times and China Times. Regarding the impact of media reports on vaccination willingness, the study shows that in June 2021 (the 14th week of vaccination), the sentiment score of China Times significantly decreased (t=-3.550, p<0.05), followed by a sharp drop in the number of first-dose vaccinations in the 15th week, which, although not significant, was inferred from the news content to be related to numerous reports of adverse reactions to the AZ vaccine and online rumors of vaccine misinformation. Additionally, Pearson correlation analysis showed a significant negative correlation between China Times' sentiment score in the 15th week and the number of second-dose vaccinations (r=-0.78, p<0.05), with the media content mainly criticizing government vaccine policies. Similarly, in the 21st week, Apple Daily's sentiment score was significantly negatively correlated with the number of second-dose vaccinations (r=-0.77, p<0.05), with reports focusing on the international pandemic, adverse vaccine reactions, and vaccine shortages. This indicates that during these two weeks, as vaccination numbers increased, the sentiment scores of the two media outlets became more negative, suggesting a reciprocal influence between media sentiment and public vaccination willingness. In conclusion, the study highlights that in the early stages of the COVID-19 pandemic, Taiwanese mainstream online news media extensively reported on adverse vaccine reactions and circulated false information about vaccine safety, affecting public willingness to receive vaccines from certain brands and contributing to vaccine hesitancy. As vaccination rates increased, some media outlets reduced negative sentiment in their reports, mainly focusing on criticizing government vaccine policies, adverse reactions, and vaccine supply shortages. This suggests that media may have specific stances or be influenced by their reporting styles. The study emphasizes the significant impact of media sentiment on public health behaviors, especially in the decision-making process for vaccination. It recommends that in future pandemics or public health crises, media should maintain objective and neutral reporting, avoid excessive negative sentiment, and government public health strategies should consider the influence of media reports, strengthening the dissemination of accurate information to prevent negative reports or misinformation from affecting public health safety.