|
Abstract
|
On January 13, 2018, at 8:07 a.m. Hawaii Standard Time, an errant emergency alert was sent to persons in Hawaii. An employee at the Hawaii Emergency Management Agency (EMA) sent the errant alert via the Wireless Emergency Alert (WEA) system and the Emergency Alert System (EAS) during a ballistic missile preparedness drill, advising persons to seek shelter from an incoming ballistic missile. WEA delivers location-based warnings to wireless carrier systems, and EAS sends alerts via television and radio (1). After 38 minutes, at 8:45 a.m., Hawaii EMA retracted the alert via WEA and EAS (2). To understand the impact of the alert, social media responses to the errant message were analyzed. Data were extracted from Twitter* using a Boolean search for tweets (Twitter postings) posted on January 13 regarding the false alert. Tweets were analyzed during two 38-minute periods: 1) early (8:07-8:45 a.m.), the elapsed time the errant alert circulated until the correction was issued and 2) late (8:46-9:24 a.m.), the same amount of elapsed time after issuance of the correction. A total of 5,880 tweets during the early period and 8,650 tweets during the late period met the search criteria. Four themes emerged during the early period: information processing, information sharing, authentication, and emotional reaction. During the late period, information sharing and emotional reaction themes persisted; denunciation, insufficient knowledge to act, and mistrust of authority also emerged as themes. Understanding public interpretation, sharing, and reaction to social media messages related to emergencies can inform development and dissemination of accurate public health messages to save lives during a crisis.
The rapid dissemination of public health messaging is a component of information management, one of the six core domains of public health preparedness (3). The information management domain addresses public health communication and includes two capabilities: 1) emergency public information and warning and 2) information sharing. Emergency public information and warning is an essential capability for state and local public health preparedness and consists of the ability to develop, coordinate, and disseminate information, alerts, warnings, and notifications to the public and incident management responders. The information sharing capability consists of the ability to conduct multijurisdictional, multidisciplinary exchange of health-related information and situational awareness data among all levels of the government and the private sector (3).
Using Sysomos† (version 1.47; Meltwater), an analysis of social media data from Twitter was performed by conducting a Boolean search to identify relevant tweets. The search used the terms "missile and Hawaii," "ballistic," "shelter," "drill," "threat," "alert," or "alarm" to identify tweets posted on the morning of January 13, 2018. Twitter data were used for this analysis because they are available in the public domain and easily accessible. Retweets (reposting the same tweet) and quote tweets (reposting the tweet with a comment at the top of the tweet) were excluded to limit the analysis to initial tweets.
All tweets were stratified into one of two periods. The early period consisted of tweets sent during the initial 38 minutes (8:07-8:45 a.m.), and the late period consisted of those sent in the 38 minutes after the false alarm retraction message was issued via EAS and WEA at 8:45 a.m. Tweets were coded using grounded theory, which is a systematic approach to analyze qualitative data and develop theories from those data (4). Themes were identified until theoretical saturation was reached. Atlas.ti§ software (version 8; Atlas.ti Scientific Software Development) was used for all exploratory qualitative analysis.
A total of 127,125 tweets were identified; after excluding 69,151 (54%) retweets and 43,444 (34%) quote tweets, 14,530 (11%) initial tweets remained for analysis. Among these, 5,880 (40%) were sent during the early period, and 8,650 (60%) were sent during the late period.
Four themes emerged from the Twitter data during the early period: 1) information processing; 2) information sharing; 3) authentication; and 4) emotional reaction. Information processing was defined as any indication of initial mental processing of the alert. Many information processing tweets dealt with coming to terms with the imminent missile threat (Table). Information sharing consisted of any attempt to disseminate the alert, often directed at other persons' Twitter handles (user names). One Twitter user shared a tweet with the U.S. Indo-Pacific Command, the Defense Intelligence Agency, and the White House National Security Council. Authentication involved any attempt to validate the alert. Finally, emotional reaction was the expression of shock, fear, panic, or terror.
During the late period, the information sharing and emotional reaction themes persisted, and three new themes emerged: 1) insufficient knowledge to act; 2) denunciation; and 3) mistrust of authority (Table). These new themes are fundamentally different from those expressed during the early period and reflect reactions and responses to misinformation. Insufficient knowledge to act involved reacting to the lack of a response plan, particularly not knowing how to properly take shelter. Denunciation consisted of blaming the emergency warning and response, particularly criticizing the time it took to correct the alert. Finally, mistrust of authority involved doubting the emergency alert system or governmental response...
Language: en |