SENTIMENT ANALYSIS AND EFFECTIVE SOCIAL MEDIA COMMUNICATION: A LOW-INCOME COUNTRY CASE DURING A COVID-19 PANDEMIC

Objective: This research investigates how the government of a low-income country can use sentiment analysis on social media and improve its communication during a national crisis. Method: This research scraped citizens' comments on the Brazilian Health Department's Facebook page regarding the COVID-19 pandemic (N = 106,292). Data were cleaned, and sentiment analysis was performed using social media comments (N = 93,715) employing the software LIWC. Originality/Relevance: The comprehension about the emotions perceived by citizens during a crisis is the most relevant aspect of the study. Mainly, we discuss that videos make sentiments stronger and impacts the perceptions about press conferences and informative posts content. Results: Findings revealed that the government posts in social media were composed of three categories: informative, press conference, and prevention. Further, it demonstrated that when these posts were made with video (vs. picture), citizens' positive emotions, social aspects, perceptual aspects, work aspects, and death perceptions were stronger. It happens because when the government uses video (vs. picture), people have more vividness regarding the post, increasing the intensity of sentiments. Theoretical/methodological/practical contributions: This research contributes, in a practical manner, by providing evidence of how citizens feel during a pandemic and what is the best manner to attenuate their negative sentiments. It also contributes by applying a sentiment analysis approach, a method that has been growing in the last years, mainly in the discussions about how consumers and people feel concerning their consumption and daily life experiences. It can be helpful for the communication strategies made by the government during a national crisis.