“We ran outside and waited for it to come”: Resident experiences in response to a false earthquake early warning

Abstract

Ridgecrest, CA, experienced the Searles Valley earthquake sequence in 2019 and a “false” earthquake alert in 2020, providing a unique opportunity to examine the effects of earthquake experience on future responses to informational cues to action (i.e., earthquake alert), as well as reactions to a “false” alert. We conducted in-depth interviews with 41 residents using the protective action decision-making model as a theoretical framework. Interviewees reported a variety of environmental cues that signaled the onset of an earthquake, including sensing a foreshock, hearing the earth rumble, hearing objects fall to the floor and break, and observing unusual animal behavior. Fewer individuals received social cues to action. More individuals reported performing “drop, cover, and hold on,” and fewer reported standing in a doorway in response to the 2020 alert than had done so in the prior 2019 earthquake. Several respondents reported maintaining protective actions well after the “false” alert was issued, and many waited more than 5 min before determining there was no threat present. Prior experience of the 2019 earthquake series affected perceptions of the earthquake alert and what actions to take; however, there was limited knowledge of how the ShakeAlert system worked to monitor, detect, and model earthquakes via earthquake early warning to persons at risk. Findings indicate there is a need for additional public education about ShakeAlert-powered earthquake early warning, including how far in advance one can expect to receive an alert, as well as the protective actions one should take and when to take them.

Keywords

Earthquake early warning public response false alert protective actions risk communication

Introduction

Early warnings delivered to persons at risk of an earthquake have moved from possibility to reality in the last decade. ShakeAlert-powered earthquake early warnings (EEWs), delivered over wireless emergency alerts (WEAs), Android Earthquake Alerts (AEAs), or apps like MyShake, can now deliver life-saving warnings to persons within seconds of detection and potentially up to a minute’s notice (University of California, Berkeley, n.d.). Notification of an earthquake prior to its arrival requires the message receiver to have some level of understanding of the system and why a message is showing up on their mobile device, as well as some knowledge of the actions they should take, such as drop, cover, and hold on (DCHO) when instructed to do so (McBride et al., 2022b). Taking an action like DCHO absent an environmental cue that validates the need for action also indicates some level of trust that the alert is accurate and reliable (Bostrom et al., 2022). Without this belief in the veracity of the alert, message receivers are unlikely to act (Becker et al., 2022).

Since October 2019, when ShakeAlert-powered EEWs began to be delivered through WEAs for events magnitude 5.0 and above, there have been 21 successful WEAs issued as of 11 September 2023 (personal communication—S. McBride, 30 October 2023). This demonstrates the effectiveness of the ShakeAlert system to monitor, detect, and notify individuals of potential shaking prior to an actual shaking event (ShakeAlert, 2023). Such events can provide evidence of the accuracy of this new technology, bolstering the calls to action for those who may be at risk. Indeed, in anticipation of future shaking, the US Geological Survey expanded their questionnaire on the Did You Feel It? webpage (US Geological Survey, 2023), to include items about the receipt of an EEW over various devices (Goltz et al., 2023).

One event stands out, however, because the ShakeAlert-powered EEW was issued not in advance of an earthquake, but by accident, to a population that was not at risk. Furthermore, this same population had previously experienced one of the most severe shaking events in California in the recent 20 years (McMillan et al., 2019). Because this same population experienced severe shaking fewer than 12 months prior and then later received an alert, which was not followed by felt earth shaking, this unique community “event” has created an opportunity to learn a great deal about false alerts, as well as environmental, social, and informational cues that commonly affect protective action decision-making.

In this article, we apply the protective action decision model (PADM) (Lindell and Perry, 2004, 2012) to investigate the behavioral responses of residents in the town of Ridgecrest, California, and surrounding areas during the 4 and 5 July 2019 Searles Valley earthquake sequence (Wald and Collett, 2021) and the 21 May 2020 false alert (Salahieh, 2020). We show how on July 4 and 5 environmental cues affected the pre-decisional and decisional processes leading to protective action (PA) behaviors. We then show how residents responded to the social and informational cues received on 21 May, when they were warned for shaking that never occurred. We demonstrate the link between prior earthquake experience and PAs taken in response to a “false” EEW.

Literature review

Decision-making

The PADM provides a framework that describes the process individuals engage in as they make PA decisions in response to natural and technological hazards (Lindell and Perry, 2004, 2012). Decision-making begins when people receive a “cue” from the environment (such as feeling or observing an approaching threat), from people (a “social cue” such as observing neighbors taking an action to protect themselves), or from an information channel (such as a warning message delivered via opt-in messaging or other media) (Lindell and Perry, 2012). When a cue is received, the individual will initiate a pre-decision process that leads to a PA decision being made. The pre-decision process includes three parts: receiving a cue, choosing to pay attention to it, and comprehending the meaning of it. The activities of this stage, after cues are processed, lead to decision-making activities (Lindell and Perry, 2012).

Following the pre-decision stage, individuals must assess the credibility of the threat, including its potential impacts, and its personal relevance to themselves (or how susceptible they are to the potential impacts). The more certain, severe, and immediate the threat is perceived to be, the more likely the individual is to take action to protect themselves (Perry et al., 1981). The decision-making process may include searching for information about the threat, PAs, and determining what actions others are taking to protect themselves (Lindell and Perry, 2012). The more ambiguous the situation—that is, the more uncertain the receiver is about the environmental, social, or informational cues—the more likely it is that one will seek additional information and identify the actions of others (Wood et al., 2018).

Most hazards, such as wildfire, tornado, hurricane, blizzard, and fog, are arguably accompanied by some sort of environmental cue such as the smell of smoke, seeing the colors in the sky change, hearing sounds approaching, or feeling temperatures dropping or winds increasing. Each of these cues provides a bit of data that people at risk may begin to compare to their prior experience or knowledge. Reliance entirely upon a warning message, absent an environmental cue, is less common for natural hazard events. Indeed, the decision to take an action while fully reliant upon an informational cue, such as a warning message, is most commonly seen in response to unobservable threats such as toxic chemical releases (Carson, 1962) or infectious disease (Johnson, 2019). In response to those types of threats, perceived risk is exacerbated by uncertainty about the threat (Hoti et al., 2020), the potential severity of harm resulting from exposure (Prati et al., 2012), and affective dimensions (Slovic, 2010) such as outrage (Sandman, 1989), fear and anger (Lu et al., 2013), worry (Parker et al., 2020), and dread (Johnson, 2019).

Response to cues during an earthquake threat

Environmental cues

Compared with research on other environmental threats, research on immediate response to felt earth shaking is limited (Lindell et al., 2016). However, in a review of the literature, Lindell et al. (2016) found that “people’s immediate responses to earthquakes typically are initiated by ambiguous environmental cues such as a rumbling noise or the ground shaking” (p. 3). Further examples of environmental cues may include observing objects swaying or falling, feeling the ground become unstable, and observing the ground rip open. Individuals have also recorded unusual animal behaviors as a pre-quake signal (Siddiqui, 2022), which have been suggested to serve as biosensors to predict earthquakes (Manda, 2023).

Social cues

In some cases, PAs appear to follow social cues, such as observing others taking actions such as DCHO (dropping to the ground, covering their heads/bodies, and holding onto something), or alternatively, running away (McBride et al., 2022a). Importantly, those who have limited earthquake experience or limited pre-event education/awareness may rely on the social cues from others to interpret the shaking for them prior to acting (Alexander, 1990; Lindell et al., 2016).

Informational cues

In contrast with environmental or social cues, limited research has been conducted on informational cues, such as warning messages issued in advance of an earthquake. For example, Nakayachi et al. (2019) examined the effectiveness of EEW as an informational cue in Japan based on experiences of residents who received official warnings before the earthquake shaking occurred. In two separate studies, they found that the primary reaction of those who received the alerts was to remain still or motionless as they mentally prepared to feel earth shaking. They found that residents perceived the EEW to be effective because it gave them the ability to mentally prepare themselves for shaking; not because it spurred them to take a PA. In some cases, EEW may inhibit PA due to the attention placed on the alert at the time when they should be motivated to act (Nakayachi et al., 2024). Similarly, in Peru, where EEWs are disseminated using smartphones, research found that although EEWs were generally appreciated and understood by the public, only a fraction of those persons who received an alert as an informational cue before a recent earthquake took PA (Bossu et al., 2022). In another study, researchers assessed the effects of Google’s Android Earthquake Alert System on PA taking prior to felt shaking for an earthquake in Aotearoa New Zealand (Vinnell et al., 2023). They found that few participants used the time to protect themselves from shaking following this unofficial alert.

In the United States, the 2020 Lone Pine earthquake in eastern California was one of the first instances in which people may have received an EEW and also felt the earthquake (United States Geological Survey (USGS) ShakeAlert, 2020b). Research suggests that some people in the area had the dual experience of receiving an alert (an informational cue) prior to sensing environmental cues (Hauksson et al., 2021). By examining social media messages post-shaking, researchers found that individuals had questions about whether more earthquakes were imminent; however, the experience of receiving the informational cue and the actions individuals took to protect themselves following receipt of the EEW prior to shaking is not known. Additional EEWs sent the same year (Martin and Wenzke, 2020; USGS ShakeAlert, 2020a, 2020c) included alerts for three other California earthquakes (24 January, M4.6 near Barstow, CA; 4 April, M4.9 near Anza, CA; and 10 May, M4.5 near Ocotillo Wells, CA).

PAs

Actions performed in response to environmental cues experienced in an earthquake such as ground shaking regularly include taking cover by sheltering under a desk or doorway (Arnold et al., 1982; Bourque et al., 1993; Goltz et al., 1992; Lindell et al., 2016; Prati et al., 2012), freezing in place (Arnold et al., 1982; Bourque et al., 1993; Goltz et al., 1992; Lindell et al., 2016; Prati et al., 2012), and immediately evacuating or escaping from danger (Arnold et al., 1982; Goltz et al., 1992; Lindell et al., 2016; Ohta and Ohashi, 1985; Prati et al., 2012). Similar findings have been recorded by Goltz et al. (2020).

Alerts and EEW

Earthquake alert scenarios have been developed to identify 15 alert types, including (1) successful alerts (a message was issued in time to act and people felt shaking); (2) no alert (an earthquake occurs and shaking is felt, but the alerting threshold was not met); (3) inaccurate alerts (i.e., event is mislocated and no shaking is felt in the alerted area or an EEW is issued by the magnitude or location is wrong or the EEW is issued but the magnitude is underestimated); (4) simultaneous and late alerts (i.e., an earthquake occurred that met alerting thresholds, but people were too close to the epicenter to receive a timely alert); and (5) false alert (one or more of the alerting delivery platforms or systems experienced a technical issue that caused a ShakeAlert-powered message to be issued in the absence of an actual earthquake) (see McBride et al., 2020).

Because of the variety of scenarios and conditions under which the EEW system might send a message, the USGS designed a set of post-alert messages, one for each alerting type that could be issued via WEA, push notification/cell phone app, or social and broadcast media. These messages could be delivered in the immediate aftermath of an errant EEW to explain what had taken place and affirm the PAs taken by the recipient (McBride et al., 2020). Following the false alert issued to the town of Ridgecrest, CA, on 21 May 2020, the USGS issued a post-alert message that stated “USGS ShakeAlert Message cancelled. Investigating. If you protected yourself, well done,” Through a series of interviews and focus groups, researchers found that the post-alert message was lacking in key information that message receivers most wanted (Sutton et al., 2023). This included the status of the earthquake itself and whether the risk was no longer present. Research participants had one primary concern after receiving the initial EEW, they wanted to know “am I safe?”

As demonstrated in the review above, prior research has largely focused on links between environmental cues, social cues, and behaviors in response to felt earth shaking; scholarship examining the relationship between informational cues, or warning messages, and earthquake-related behaviors has been limited. Even less attention has been given to behaviors in response to an earthquake alert that turns out to be false. Furthermore, the links between prior earthquake experience and warning messages have not been studied. In this study, we draw from 41 interviews with members of the public who experienced both the 2019 Searles Valley earthquake sequence and the 2020 false alert to learn by addressing the following research questions (RQs):

RQ1a: What was the initial cue to action in the 2019 Searles Valley earthquake sequence?

RQ1b: What were the immediate behavioral responses to the 2019 Searles Valley earthquake sequence?

RQ2a: What was the initial cue to action in the 2020 EEW in Searles Valley, CA?

RQ2b: What were the immediate behavioral responses to the 2020 EEW in Searles Valley, CA?

RQ3: What is the relationship between the earthquake experienced in 2019 and the EEW issued in 2020?

Context

On 4 July 2019, a 6.4 magnitude earthquake struck Searles Valley, CA, in the area of Ridgecrest, California, at 17:33:49 UTC (10:33 a.m. PST) (Wald and Collett, 2021). This quake was later identified as a foreshock of significant size and was followed by nine aftershocks above a magnitude of 4.0 throughout the remainder of the day (Wald and Collett, 2021). This was the strongest earthquake to occur in California in more than 20 years and resulted in the declaration of a state of emergency for the city of Ridgecrest (McMillan et al., 2019) leading to resource deployment by the Federal Emergency Management Agency (FEMA) to aid in the response (Sanchez et al., 2019). Minor injuries were reported due to broken glass and fallen objects (Bloom et al., 2019), while there were additional reports of damage to roads, structures, water lines, and critical infrastructure due to shaking and resultant fires (Bloom et al., 2019).

The following day, on 5 July 2019, a 7.1 magnitude mainshock occurred at 03:19:53 UTC (8:19 p.m. PST) in Searles Valley, CA (Wald and Collett, 2021). This resulted in additional damage to roads, buildings, and infrastructure in Ridgecrest and surrounding areas, including the Naval Air Weapons Station at China Lake. This series of earthquakes, which caused approximately US$100 million in damage (Sanchez et al., 2019), also resulted in significant emotional distress among residents (Shatkin, 2020; Woodyard, 2019).

Over those 2 days, shaking was felt as far away as Los Angeles (152 miles), where the city had previously promoted the adoption and use of an earthquake alerting app called ShakeAlertLA. ShakeAlertLA was operational at the time of the July 4 earthquake. The alerting threshold, however, was set to issue an alert for earthquakes that reach a magnitude of at least 5 and an a Modified Mercalli Intensity (MMI) of at least 4, resulting in an alert not being issued. When ShakeAlertLA did not alert individuals, residents who appeared to be unfamiliar with the threshold requirements reported via social media that the application and the EEW system failed (Bogel-Burroughs, 2019; Epstein and Cha, 2019). Goltz and colleagues (Goltz et al., 2020) assessed the human behavioral response to the Searles Valley earthquake sequence using data submitted to Did You Feel It (US Geological Survey, 2019). They found that the two most active responses reported were to go to a doorway and to run outside; DCHO was the least frequent action taken.

Just 8 months after the Searles Valley earthquake sequence, on 21 May 2020, at 2:10 p.m. PST, the USGS ShakeAlert system mistakenly issued a ShakeAlert-powered EEW for the area of Ridgecrest, CA, during a test of the system (see Figure 1).

The alert was delivered via WEA to all cellular devices within the alert area that were equipped to receive a WEA. WEAs operate when the device is within the geographic vicinity of a cellular tower delivering the message; they do not require an additional app to be downloaded to a phone by the user (Sutton and Kuligowski, 2019). WEAs are automatically enabled on most smartphones, though users can turn this feature off through their device settings. When an EEW is sent via WEA, it triggers a loud tone and displays a message on the phone screen (FEMA, 2023).

It is unknown how many phones received the EEW alert on 21 May 2020. However, no earthquake followed the ShakeAlert-powered EEW. It was later determined by the USGS that the message had been mistakenly sent during a test of the system (Salahieh, 2020). Unfortunately, those who received the message represented the same population that had only 8 months prior experienced one of the most damaging series of earthquakes in the United States in more than a decade.

Data collection and analysis

In 2022, we conducted a series of interviews with persons who lived in the Ridgecrest region to learn about their experiences and actions in response to the July 2019 earthquakes and the May 2020 EEW. Interviews were coded inductively to identify themes related to environmental, social, and informational cues to action, as well as the actions taken to protect oneself from harm. We describe our data collection and analysis next.

Data collection

Interviews were conducted with 41 participants who were recruited from four Facebook groups relating to the Ridgecrest, CA, area. Facebook has been used successfully as a recruitment tool in social science research where populations are difficult to reach through traditional panel studies (Whitaker et al., 2017) or contact networks (Khatri et al., 2015). Researchers have argued that the sheer size of the user base of Facebook implies that even traditionally underrepresented groups, like those that have been affected by disaster, are likely to have a relatively large presence (Kosinski et al., 2016). Indeed, several prior disaster studies have used Facebook as a tool for recruitment for rapid response research where data are perishable and for targeted recruiting of difficult-to-reach populations (DeYoung et al., 2019; Hugelius et al., 2017; Mongold, 2020).

To access potential participants in these closed groups, the principal investigator of the study gained permission from group administrators to post a digital flyer inviting people to sign up for interviews via a Google Form. Selected participants were required to be at least 18 years of age and have experienced both the July 2019 and May 2020 events of focus for this study. Using the provided contact information, potential participants were contacted by a project research assistant (RA) to schedule an interview via the video conferencing software, Zoom. The use of Zoom in an academic research setting allows researchers to conduct interviews remotely, while reducing the cost and burden of travel while allowing for a greater breadth of perspectives to be represented (Archibald et al., 2019). Participants were given a US$40 Amazon gift card for their time participating in the study.

Interviews were conducted between 12 April and 13 July 2021 and were 26 to 99 min in length with an average of 46.78 min. Each interview was audio recorded and automatically transcribed using Otter.ai, an online audio transcription service. Transcripts were cleaned and validated by a project RA, and thematic analysis was conducted using spreadsheets and word processing tools.

Participants

A total of 41 participants were interviewed. Of these, 41, 66% (n = 27) described themselves as female. Participant age ranged from 20 to 70 years; 15% (n = 6) were between 20 and 30 years of age; 34% (n = 14) were between 31 and 40 years of age; 19% (n = 8) were between 51 and 60 years of age; and 15% (n = 6) were between 61 and 70 years of age. The amount of time that participants reported living in Southern California ranged across participants from 1 month to 66 years, with an average of 28.1 years. More than 75% of participants (n = 31) reported having previous earthquake experience prior to the events of 4–5 July 2019.

Procedure

After providing informed consent (University at Albany institutional review board (IRB) protocol #IRB00000589, 24 June 2020), interviewees participated in a semi-structured interview focusing on their experience in response to the July 2019 earthquakes in Ridgecrest, CA, followed by their experiences and perceptions of the 21 May 2020 EEW message. To elicit recall, interviewees began by describing their experience of the 2019 earthquakes, including what they remembered about the earthquake (i.e., what they observed, felt, and heard), the impacts and severity of damage they experienced, as well as their thoughts and emotions related to the event. Initiating an interview by asking respondents about their memories and emotions helps to transport the interviewee and to tap into their lived experience (Phillips, 2014). It also helps to develop rapport with the interviewer, as it establishes their interest in the experiences of the participant and creates an environment of safety for discussing memories about a disaster that may be unsettling (Kim et al., 2023). Following the initial discussion about their experiences of the July earthquakes, questions were asked about how they decided to take action, and what actions they and others took. A similar process was followed for their experiences of the May 2020 EEW messages. They were invited to describe their memories about the event, including their emotional responses, and to provide details about how they decided to take action, and what actions they and others took following the initial message and the post-alert message.

Demographic information was also collected from each participant, including gender, race, age, education, residency in California, and prior earthquake experience.

Data analysis

Transcripts were cleaned, and data were organized by question using Microsoft Excel spreadsheets. Data were then coded using an inductive approach and guided by the RQs about cues to action, decision-making, and PAs taken. Two researchers coded the data; the first author and an RA individually reviewed all of the participant responses to each category and identified emerging themes using constant comparative method of qualitative analysis (Hewitt-Taylor, 2001). They then met together to discuss the themes they identified as a form of peer debriefing (Spall, 1998) to compare and contrast their results. Once agreement was reached, the RA completed the remaining coding activities.

Results

In the next section, we present our results chronologically and organized by RQs. Frequencies are presented in Tables 1 to 3.

Table 1.

Reported experiences of and responses to the July 4–5 earthquake

Cue/emotion/action type (N = 41)	4 July 2019 (n)	5 July 2019 (n)
Environmental cues
Sound	6	5
Foreshock	11	1
Shaking	37	41
Animals	1	1
Water movement	3	N/A
Dust cloud	1	N/A
Objects breaking	9	4
Light swinging	2	1
Social cues
Neighbors outside	3	N/A
Heard screaming	1	N/A
Emotions
Excited/fascinated	10	5
Afraid/scared/terrified	24	35
Anxious/stressed/panicked/worried/nervous	25	23
Confused/uncertain	12	N/A
Angry/annoyed	1	N/A
Actions
Went outside	6	8
Stood in doorway	8	4
Stood still	9	N/A
Held onto objects that could fall on them	2	1
Under a table/desk	4	7
Stayed on couch	2	3
Laid down	N/A	4

Table 2.

Reported experiences of and responses to the May 21 EEW

Cues/emotions/actions (N = 39)	21 May 2020 (n)
Environmental cues
Animals	1
Social cues
Neighbors outside	1
EE warning	37
Did not receive, but heard/saw someone else’s EEW	2
Emotions
Excited/fascinated	1
Afraid/scared/terrified	8
Anxious/stressed/panicked/worried/nervous	15
Confused/uncertain	2
Angry/annoyed	6
Tricked	2
Thankful	1
Actions
Went outside	12
Stood still	10
Held onto objects that could fall on them	1
Under a table/desk	2

Table 3.

Reported duration of and reasons for stopping protective actions (PA) taken following May 21 EEW

Protective action duration/reasons for stopping (N = 39)	21 May 2020 (n)
Protective action duration
Less than 5 min	3
Between 5 and 29 min	5
30 min or more	5
Reasons for stopping protective action
Time passed without shaking	13
Heard it was false from a friend in USGS	1
Saw it was false from a Ridgecrest resident on Facebook	3
Saw USGS post they were testing their broadcasting system	1
No earthquake on USGS earthquake map	2
The follow-up WEA arrived	1

July 4 and 5 earthquake event

Environmental cues

In response to the July earthquakes, interviewees described a variety of environmental cues that signaled an earthquake was about to occur or was presently occurring. For example, 27% (n = 11) of interviewees claimed that a few minutes prior to the 6.4 magnitude earthquake, a foreshock had captured their attention. Nearly a quarter (22%, n = 9) reported hearing things falling to the floor and breaking as their first sign. In total, 90% (n = 37) of participants said that they felt the shaking on July 4; the remaining 10% (n = 4) were driving, boating, or out of town; however, they recalled being immediately notified about the earthquake by friends and family. All interviewees (100%; n = 41) said they felt the shaking on July 5.

In contrast with the felt environmental cues, one woman, who owns livestock, stated that she knew something was coming based on her observations of the animals,

What I remember was trying to figure out, starting around 9:30, why the peafowl were alerting so much. They were just screaming! I couldn’t find anything going on in their habitat. Then just about ten minutes before the quake, the horse and donkey and the burro just started running, running, running. I was like “Okay. Something’s happening. We are going to have something big.”

She described a similar experience in relation to the following day, July 5, when the magnitude 7.1 earthquake occurred, stating “I’m sitting outside again watching the same thing unfold with the peafowl, and the horse, and the burro.”

An additional 14% of participants (n = 6) on July 4 and 12% (n = 5) on July 5 described the sound as a key environmental cue, saying that they heard the earthquake before they felt it. Participants compared the sound of the earthquake to that of a freight train, an 18-wheeler ramming into a building, rushing water, or the nearby military base’s bomb tests followed by loud rumbling.

Social cues

Few interviewees (n = 3) described observing the behaviors of others as a cue to action. One participant who had moved to Ridgecrest just 1 month prior to the earthquake sequence said that on July 4 she struggled to remember what she was supposed to do in an earthquake before she finally went outside, where she found her neighbors gathered. Interviewees who were not in town at the time of the first earthquake (n = 4) described receiving phone calls or text messages from neighbors, friends, and family members to alert them; they all returned immediately.

Decision-making

Interviewees described how they considered various factors when making a decision about how to protect themselves. In some cases, the interviewees directly referenced factors that they considered such as personal or family safety. For instance, one participant, a husband and father in his 30s, explained,

I figured whatever we have at home could be replaced eventually. The important stuff was my wife and kid. It’s okay.

In contrast, some participants described concerns about physical harm that might come from their walls, ceilings, or entire houses collapsing around them (n = 6; 15%); falling trees, telephone poles, or transformers (n = 4; 10%); and roads/infrastructure being severely damaged as a result of the shaking (n = 2; 5%). One interviewee, a man in his 40s, explained these concerns, saying,

Generally, in my experience, the earthquake is not as bad as the damage and things that can happen later, a gas line can catch fire, or a cracked support beam which, in an aftershock, may cause a roof to cave in.

Several interviewees who lived in mobile homes also described a fear of their trailer falling off the jacks, a hazard that affected many homes because of the July 4 and 5 earthquakes (KGET News, 2019; Maison and Martinez, 2020).

PAs

Although most participants had experienced earthquakes before, many described being confused about the right thing to do when they felt the shaking. On July 4, 20% of participants (n = 8) described going outside during the shaking; an equal number (n = 8; 20%) stood in a doorway. Behaviors on July 5, however, differed for some. At the time of or following the earthquake, 20% (n = 8) went outside, but 12 % (n = 5) described dropping to the ground and covering under a desk or table. Only 10% of participants (n = 4) said they stood in a doorway.

Informational cue

On 21 May 2020, almost a year after the July 2019 earthquake sequence, Ridgecrest residents received an alert telling them that an earthquake was coming. For many, their initial thoughts about the alert (n = 11; 27%) included wondering what the alert was about, who it was from, and why they received it. One person’s question: “they can predict earthquakes now?” was a common refrain heard from many interviewees.

Many who received the EEW related the message to their experiences on July 4 and 5. For example, one interviewee who worked at the Naval Base (36-year-old, male) said,

My heart jumped in my throat because it brought back all the memories of the previous year. [The 7.1 earthquake on July 5 was] the scariest thing that’s ever happened in my life.

Another participant, a woman in her 20s, reported that she and her elderly neighbor both experienced anxiety attacks following the July 4 and 5 earthquakes, causing her to take the EEW very seriously, saying,

When we first saw the text that definitely made us panic. Our hearts were racing trying to get out. Just thinking “Oh no. Not again.” That was really scary. It shook both of us up pretty bad. We both got really bad anxiety attacks. It took probably the rest of the day for me to be calmed down. After they said it was an accident is when I finally started trying to calm down. It took a while for them to send out the thing that said it was an accident, though, so we had no idea. It brought all those memories flooding back and it was a panic. It was not a great mistake to happen to the community that had to deal with that.

Social cues

Many interviewees described noticing social cues around them; in particular, they saw neighbors outside their homes. Importantly, the EEW was issued at the height of the COVID-19 pandemic when much of the state of California was on lockdown. This meant that neighbors had restricted contact with one another, as the pandemic confined many to their homes. Seeing other people going outside was an unusual occurrence when they were advised to remain isolated. In some cases, where individuals were at work when they received the alert, interviewees described discussing the alert with their coworkers to determine what they should do in response (n = 2; 5%). This interaction with others helped to interpret the meaning of the alert and affirm the (lack of) actions they would ultimately take.

Behaviors in response to the alert

Nearly all (n = 39) of our interviewees received the alert on their phones (95%) prompting 10 of them (24%) to go directly outside while another 10 (24%) described waiting for shaking to begin before acting.

One interviewee described her behavior and that of those around her as taking immediate action and related this to their prior experience, saying:

My church actually received significant damage [during the 2019 earthquakes]. My children were inside the church, and I was outside. I got the alert and immediately went screaming into the building thinking it was going to collapse on them. Knowing that they were in this unstable building at the time was what made it very terrifying. Then they ran outside into the parking lot with me, along with two other adults that were with me at the time.

Another woman described running to her children who were on the trampoline across the yard explaining how she thought, “I better hurry and get to the other side because if that powerline comes down the kids are going to be trapped on the other side.”

Of the 39 interviewees who received the alert, only two interviewees (5%) described following the instructions in the message: DCHO. One interviewee described getting under a table with family members and the other hid under her desk at work.

Post-alert behaviors—ending the action

In the minutes after receiving the EEW, almost a third of the interviewees (n = 13; 31%) remained alert, a few of whom described themselves as being “on stand-by” in case shaking was still coming, eventually disregarding the EEW as time passed. Some interviewees described their belief that earthquakes are hard to predict and, if they were predictable at all, many participants believed, the alert would immediately precede the shaking. Some interviewees (n = 3; 7.3%) waited only a short time before ending their PAs (e.g., 5 min or less). One interviewee explained the alert to their family saying, “What I think is that this is bogus because we would have been shaking by now.” This was a common interpretation from many of those interviewed from Ridgecrest.

Others, however, waited much longer (n = 5; 12%)—in one case, for about 30 min, and in another, a full hour. One interviewee, who had strong memories of the physical impacts experienced during the July 5 earthquake, described running to the home of family members who lived nearby to alert them. Once there, they took cover and “stayed there for about an hour before we came out, just to be safe.” Another interviewee who ran outside said, “We sat for a half an hour and then we’re like, ‘Well I guess it just blew past us.’”

In some cases, interviewees searched for information online to confirm that they were not in danger. For example, three interviewees remembered seeing that the alert was false while looking for information on Facebook. Others searched online for information from the US Geological Survey, going to the organization’s webpage to learn about EEWs or searching for evidence of a recent earthquake that may have been too small to feel or that had occurred elsewhere. When no information could be found, they disregarded the alert and returned to their normal activities. One interviewee, however, described the experience of herself and her neighbor who argued that they could not “calm down” until the post-alert message was issued by USGS approximately an hour after the false alert was received.

Discussion

July EQ

We find that the initial cue to action and the immediate behavioral responses of our interviewees to the 2019 Searles Valley earthquake sequence correspond with what has been reported following other earthquake events (Arnold et al., 1982; Bourque et al., 1993; Goltz et al., 1992, 2020; Lindell et al., 2016; Ohta and Ohashi, 1985; Prati et al., 2012). In particular, participants described feeling the earth shaking, seeing objects sway, fall, or break, and hearing the earthquake approaching. Their actions are also similar to those taken in response to historical events; most participants reported standing still, bracing themselves, or running outside. Only 11 interviewees performed DCHO, and this was primarily in response to the earthquake on July 5.

May EEW

In response to the EEW issued on 21 May 2020, participants reported that the EEW informational cue resulted in immediate action for half of them, where nearly 25% described going outside and another quarter responded by bracing themselves physically and/or mentally. Many of those who took action also reported that their prior earthquake experiences in 2019 were key drivers of their behavioral response; they experienced significant shaking and the mental or emotional effects of those events caused them to believe and act upon the EEW. In some cases, those actions continued for more than 30 min while they remained alert and in a place that they believed to be safe from falling objects, primarily outside of buildings.

More than 25 interviewees indicated that the EEW served primarily to alert them to pay attention to social and environmental cues, to take some form of action, and to maintain awareness of potential shaking for several minutes following the receipt of the message. While some quickly assessed that the EEW was “false,” because the shaking did not immediately follow the message, others spent considerable time seeking additional information to confirm the message was mislocated or sent in error prior to resuming their activities. One recalled waiting for a post-alert message to reassure her that there was no current earthquake risk.

The May 2020 alert was the first instance of an EEW sent to the Ridgecrest area and interviewees appeared to have limited prior knowledge or understanding of EEW, how it would “work,” and the relationship between EEW and earthquake prediction. As described by several participants, the ability to be alerted prior to feeling an earthquake was perceived as earthquake prediction rather than as a result of monitoring the environment, modeling the earth’s movements, and alerting people who may be at risk. Clearly, there are opportunities to increase public understanding of the ShakeAlert system and what people can expect when receiving an alert.

There was also evidence that this population expected that the EEW was accurate and that it worked correctly to alert them to potential shaking, suggesting trust in the system, even though the processes of monitoring, detection, and modeling were largely unfamiliar to the Ridgecrest population. Novel technologies, like ShakeAlert-powered alerts, represent a type of automated system, where tasks require limited human involvement to function. Wojton et al. (2020) cite the factors that affect trust in automated systems, including “a person’s knowledge of why the system was built and how the designer intended for users to employ it” (p. 2) (Lee and See, 2004) and the performance of the system. Trust can deteriorate if “users’ expectations for performance do not match their experience” (p. 2) (Madhavan and Wiegmann, 2007). Wojton et al. (2020) explained that “as users gain experience with automated systems, knowledge of performance—the systems’ ability to execute specific tasks—replaces a person’s knowledge of the system’s purpose as the primary driver of trust” (p. 2) (Hoff and Bashir, 2015). Importantly, greater trust in automated systems will result in reliance behavior. With this experience of performance failure, where a false alert was issued, trust may deteriorate. The more recent examples of accurate EEWs delivered via ShakeAlert, where shaking was detected and felt by alerted populations should help to instill trust among those who questioned its ability to function effectively and reinforce trust among those who have not yet experienced a live event. This is, of course, a key RQ that should be investigated in future studies.

What also remains to be seen is the role of an EEW and how it affects public behavior. Will it initiate awareness of one’s surroundings before shaking begins? Will it reinforce the seriousness of forthcoming environmental cues? Will it serve as a cue to observe the behaviors of others who are nearby? Or will it be the informational cue that initiates PA?

Limitations and future research

This research was conducted in the winter of 2021, nearly a year after the false alert was received and close to 2 years after the July 2019 EQs. Therefore, we recognize that there may be some gaps in memory related to the earthquakes and the false alert. However, the interview guide was designed to help participants reflect on their experience of the earthquakes, tapping into their memories of what it felt like, what they saw, what they heard, and what they did, during each event, helping to trigger accurate recall. This was followed by questions about memories related to the EEW (what they heard, saw, and felt) and what they did in response to that message. Given the length of time between the events and data collection, respondent memories could be inaccurate, nonetheless. The fact that the earthquake was the largest in the area in the prior two decades made it a particularly salient experience, however, likely improving recall.

We also recognize that this was a rather limited sample of participants (N = 41), from a rural area in Southern California, recruited from a social media group who were interested in talking to a researcher about their earthquake experiences. The sample is not representative of the broader population that may be affected by earthquakes or may receive an EEW. In particular, it is biased toward those who have prior experience with a significant earthquake event and also experienced a false alert. It is possible that those who have not previously experienced an earthquake prior to receiving an EEW will have a different response when environmental cues are absent. It is also possible that knowledge of a system’s purpose and how it has functioned since May 2020 may affect interpretation of the informational cues provided in the EEW and, therefore, the actions that follow. Future research should include studies with persons who have experienced an accurate EEW to see how the informational cue triggers actions and what actions are taken. Future research should also consider the effect of participants’ prior knowledge of the ShakeAlert system (both its purpose and performance), and how this motivates behavioral response. In particular, research should examine how knowledge of system purpose and performance affects behaviors following false alerts that deliver primarily informational cues, as well as following accurate alerts that are accompanied by environmental and social cues to action.

Conclusion

This research investigated behavioral responses to a series of earthquakes followed by a false EEW in the California desert town of Ridgecrest, CA. We found that during shaking, our participants who were alerted via environmental cues took PA primarily by running outside or bracing themselves when they began to feel the earthquake. During the false EEW, participants paid attention to the alert, which served as an informational cue. Few, however, took the recommended PA of DCHO, and many waited more than 5 min before determining there was no threat present. Prior experience of the 2019 earthquakes affected perceptions of the EEW and what actions to take; however, there was limited knowledge of how the ShakeAlert system worked to monitor, detect, and model earthquakes via EEW messages to persons at risk.

Given this research, we conclude there is a need for additional public education around the ShakeAlert-powered EEW. In particular, education is needed about how far in advance one can expect to receive an alert, as well as the PAs one should take and when to take them. These findings also suggest there is belief among members of the public that the system performs as it was designed, and future accurate alerts will reinforce the ability of the system to perform effectively. Thus, even if there is a decline in trust due to a false EEW, future accurate alerts that provide an early informational cue and an opportunity to get to safety will reinforce the system performance and facilitate reliance behavior.

Footnotes

Acknowledgements

Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This material is based upon work supported by the National Science Foundation (grant no. 2051902).

Data and resources

Due to privacy and ethical concerns, neither the data nor the source of the data can be made available.

References

Alexander

(1990) Behavior during earthquakes: A Southern Italian example. International Journal of Mass Emergencies & Disasters 8(1): 5–29.

Archibald

Ambagtsheer

Casey

Lawless

(2019) Using zoom videoconferencing for qualitative data collection: Perceptions and experiences of researchers and participants. International Journal of Qualitative Methods18: 160940691987459.

Arnold

Eisner

Durkin

Whitaker

(1982) Occupant behavior in a six-storey office building fallowing severe earthquake damage. Disasters 6(3): 207–214.

Becker

Vinnell

McBride

Nakayachi

Doyle

EEH

Potter

Bostrom

(2022) The effects of earthquake experience on intentions to respond to earthquake early warnings. Frontiers in Communication 7: 857004.

Bloom

Pascucci

Myers

(2019) Magnitude 6.4 earthquake rocks SoCal on 4th of July; state of emergency declared in Ridgecrest. Available at: https://ktla.com/news/earthquakes/6-6-earthquake-jolts-southern-california-on-4th-of-july/ (accessed 22 November 2022).

Bogel-Burroughs

(2019) Californians’ alert apps didn’t sound for 2 big earthquakes. Why not? Available at: https://www.nytimes.com/2019/07/08/us/california-earthquake-alert-app.html (accessed 22 November 2022).

Bossu

Finazzi

Steed

Fallou

Bondár

(2022) “Shaking in 5 seconds!”—Performance and user appreciation assessment of the earthquake network smartphone-based public earthquake early warning system. Seismological Research Letters 93(1): 137–148.

Bostrom

McBride

Becker

Goltz

Groot

R-MD

Peek

Terbush

Dixon

(2022) Great expectations for earthquake early warnings on the United States West Coast. International Journal of Disaster Risk Reduction 82: 103296.

Bourque

Russell

Goltz

(1993) Human behavior during and immediately after the earthquake. The Loma Prieta, California, Earthquake of October 17(1989): B3–B22.

10.

Carson

(1962) Silent Spring. New York: Fawcett Cress.

11.

DeYoung

Sutton

Farmer

Neal

Nichols

(2019) “Death was not in the agenda for the day”: Emotions, behavioral reactions, and perceptions in response to the 2018 Hawaii Wireless Emergency Alert. International Journal of Disaster Risk Reduction 36: 101078.

12.

Epstein

Cha

(2019) California had its largest earthquake in years—Then an even bigger one hit. Available at: https://www.washingtonpost.com/nation/2019/07/06/california-earthquake/ (accessed 22 November 2022).

13.

FEMA (2023) Integrated public alert & warning system. Available at: https://www.fema.gov/emergency-managers/practitioners/integrated-public-alert-warning-system (accessed 22 November 2022).

14.

Goltz

Park

Quitoriano

Wald

(2020) Human behavioral response in the 2019 Ridgecrest, California, earthquakes: Assessing immediate actions based on data from “did you feel it?” Bulletin of the Seismological Society of America 110: 1589–1602.

15.

Goltz

Russell

Bourque

(1992) Initial behavioral response to a rapid onset disaster: A case study of theOctober 1, 1987 Whittier Narrows earthquake. International Journal of Mass Emergencies & Disasters 10(1): 43–69.

16.

Goltz

Wald

McBride

DeGroot

Breeden

Bostrom

(2023) Development of a companion questionnaire for “did you feel it?”: Assessing response in earthquakes where an earthquake early warning may have been received. Earthquake Spectra 39(1): 434–453.

17.

Hauksson

Olson

Grant

Andrews

Chung

Hough

Kanamori

McBride

Michael

Page

Ross

Smith

Valkaniotis

(2021) The normal-faulting 2020 Mw 5.8 Lone Pine, Eastern California, earthquake sequence. Seismological Research Letters 92(2A): 679–698.

18.

Hewitt-Taylor

(2001) Use of constant comparative analysis in qualitative research. Nursing Standard 15(42): 39–42.

19.

Hoff

Bashir

(2015) Trust in automation: Integrating empirical evidence on factors that influence trust. Human Factors: The Journal of the Human Factors and Ergonomics Society 57(3): 407–434.

20.

Hoti

Perko

Thijssen

Renn

(2020) Radiation risks and uncertainties: A scoping review to support communication and informed decision-making. Journal of Radiological Protection 40(2): 612–632.

21.

Hugelius

Adolfsson

Gifford

Örtenwall

(2017) Facebook enables disaster research studies: The use of social media to recruit participants in a post-disaster setting. PLoS Curr 9: ecurrents.dis.f4a444e1f182776bdf567893761f86b8. DOI: 10.1371/currents.dis.f4a444e1f182776bdf567893761f86b8

22.

Johnson

(2019) Americans’ views of voluntary protective actions against Zika infection: Conceptual and measurement issues. Risk Analysis 39(12): 2694–2717.

23.

KGET News (2019) Mobile home knocked off its foundation from Ridgecrest earthquake. Available at: https://www.youtube.com/watch?v=RgteU50P3mg (accessed 22 November 2022).

24.

Khatri

Chapman

Glasbey

Kelly

Nepogodiev

Bhangu

Fitzgerald

STARSurg

Committee

(2015) Social media and Internet driven study recruitment: Evaluating a new model for promoting collaborator engagement and participation (P. S. Sullivan, ed.). PLoS ONE 10(3): e0118899.

25.

Kim

Williams

Eldridge

Reinke

(2023) Digitally shaped ethnographic relationships during a global pandemic and beyond. Qualitative Research 23(3): 809–824.

26.

Kosinski

Wang

Lakkaraju

Leskovec

(2016) Mining big data to extract patterns and predict real-life outcomes. Psychological Methods 21(4): 493.

27.

Lee

See

(2004) Trust in automation: Designing for appropriate reliance. Human Factors: The Journal of the Human Factors and Ergonomics Society 46(1): 50–80.

28.

Lindell

Perry

(2004) Communicating Environmental Risk in Multiethnic Communities. Thousand Oaks, CA: SAGE.

29.

Lindell

Perry

(2012) The protective action decision model: Theoretical modifications and additional evidence. Risk Analysis 32(4): 616–632.

30.

Lindell

Prater

Huang

Johnston

Becker

Shiroshita

(2016) Immediate behavioural responses to earthquakes in Christchurch, New Zealand, and Hitachi, Japan. Disasters 40(1): 85–111.

31.

Xie

Zhang

(2013) Focusing on appraisals: How and why anger and fear influence driving risk perception. Journal of Safety Research 45: 65–73.

32.

McBride

Bellizzi

Ake

Sumy

Gin

Stewart

Santos-Hernandez

Ohlendorf

Cochran

Ball

Becker

Vinnell

Johnston

Krippner

Leonard

Baldwin

(2022a) Analyzing behavioral responses captured on video to the Hunga Tonga-Hunga Ha‘apai eruption, atmospheric shockwaves, and tsunami, 2022, SY42A-03. Presented at the AGU Fall Meeting Abstracts.

33.

McBride

Bostrom

Sutton

de Groot

Baltay

Terbush

Bodin

Dixon

Holland

Arba

Laustsen

Liu

Vinci

(2020) Developing post-alert messaging for ShakeAlert, the earthquake early warning system for the West Coast of the United States of America. International Journal of Disaster Risk Reduction 50: 101713.

34.

McBride

Smith

Morgoch

Sumy

Jenkins

Peek

Bostrom

Baldwin

Reddy

De Groot

Becker

Johnston

Wood

(2022b) Evidence-based guidelines for protective actions and earthquake early warning systems. Geophysics. Society of Exploration Geophysicists 87: 1JF-WA102.

35.

McMillan

Haskell

Stallworth

Rand

(2019) Ridgecrest earthquake: 6.4 temblor, strongest in Southern California in 20 years, strikes Mojave Desert. Available at: https://abc7.com/earthquake-usgs-mojave-desert-california/5378574/ (accessed 22 November 2022).

36.

Madhavan

Wiegmann

(2007) Similarities and differences between human–human and human–automation trust: An integrative review. Theoretical Issues in Ergonomics Science 8(4): 277–301.

37.

Maison

Martinez

(2020) Mobile home damage in 2019 Ridgecrest earthquakes. Available at: https://learningfromearthquakes.org/2019-07-04-searles-valley/images/2019_07_04_Searles_Valley/PDFs/RidgecrestMobileHomes.pdf (accessed 22 November 2022).

38.

Manda

(2023) Laboratory animals as potential biosensors to predict earthquakes. Lab Animal 52(5): 97–98.

39.

Martin

Wenzke

(2020) Magnitude 4.6 earthquake rattles barstow area. Available at: https://ktla.com/news/local-news/4-5-preliminary-magnitude-earthquake-strikes-near-barstow-u-s-g-s/ (accessed 22 November 2022).

40.

Mongold

(2020) Coastal versus inland hurricane evacuation behavior analysis. PhD Thesis, University of Delaware, Newark, DE.

41.

Nakayachi

Becker

Potter

Dixon

(2019) Residents’ reactions to earthquake early warnings in Japan. Risk Analysis: An International Journal 39(8): 1723–1740.

42.

Nakayachi

Yokoi

Goltz

(2024) Human behavioral response to earthquake early warnings (EEW): Are alerts received on mobile phones inhibiting protective actions? International Journal of Disaster Risk Reduction 105: 104401.

43.

Ohta

Ohashi

(1985) Field survey on occupant behavior in an earthquake. International Journal of Mass Emergencies & Disasters 3(1): 147–160.

44.

Parker

Finucane

Ayer

Ramchand

Parks

Clancy

(2020) Persistent risk-related worry as a function of recalled exposure to the deepwater horizon oil spill and prior trauma. Risk Analysis 40(3): 624–637.

45.

Perry

Lindell

Greene

(1981) Evacuation planning in emergency management. Battelle Human Affairs Research Centers series. Lexington Books. Available at: https://books.google.com/books?id=YSkiAAAAMAAJ (accessed 22 November 2022).

46.

Phillips

(2014) Qualitative Disaster Research. New York: Oxford University Press.

47.

Prati

Catufi

Pietrantoni

(2012) Emotional and behavioural reactions to tremors of the Umbria-Marche earthquake. Disasters 36(3): 439–451.

48.

Salahieh

(2020) Technical error results in Ridgecrest residents getting earthquake alerts telling them to take cover. Available at: https://ktla.com/news/local-news/technical-error-results-in-ridgecrest-residents-getting-earthquake-alerts-telling-them-to-take-cover/ (accessed 22 November 2022).

49.

Sanchez

Nguyen

Antczak

(2019) California assessing damage after biggest quake in 20 years. Available at: https://lasvegassun.com/news/2019/jul/06/california-assessing-damage-after-biggest-quake-in/ (accessed 22 November 2022).

50.

Sandman

(1989) Hazard versus outrage in the public perception of risk. In: Covello

McCallum

Pavlova

(eds) Effective Risk Communication. Boston, MA: Springer, pp. 45–49.

51.

ShakeAlert

(2023) What is ShakeAlert? Available at: https://www.shakealert.org/ (accessed 29 November 2023).

52.

Shatkin

(2020) Yes, the 5.5 earthquake you felt was a Ridgecrest aftershock. Available at: https://laist.com/news/55-earthquake-ridgecrest-aftershock-shakes-southern-california (accessed 22 November 2022).

53.

Siddiqui

(2022) An assessment on earthquake precursory signals for applications of monitoring and mitigation strategies. ECS Transactions 107(1): 12473–12491.

54.

Slovic

(2010) The feeling of risk: New perspectives on risk perception. Earthscan risk in society series, Earthscan, London; Washington, DC.

55.

Spall

(1998) Peer debriefing in qualitative research: Emerging operational models. Qualitative Inquiry 4(2): 280–292.

56.

Sutton

Kuligowski

(2019) Alerts and warnings on short messaging channels: Guidance from an expert panel process. Natural Hazards Review 20(2): 04019002.

57.

Sutton

Wood

Crouch

Waugh

(2023) Public perceptions of US earthquake early warning post-alert messages: Findings from focus groups and interviews. International Journal of Disaster Risk Reduction 84: 103488.

58.

US Geological Survey (2019) M 6.4—Ridgecrest earthquake sequence. Available at: https://earthquake.usgs.gov/earthquakes/eventpage/ci38443183/dyfi/intensity (accessed 22 November 2022).

59.

US Geological Survey (2023) Did you feel it? Available at: https://earthquake.usgs.gov/data/dyfi/ (accessed 29 November 2023).

60.

University of California, Berkeley (n.d.) FAQ. Available at: https://myshake.berkeley.edu/faq.html#notifications (accessed 29 November 2023).

61.

USGS ShakeAlert (2020a) M 4.5–17km SE of Ocotillo Wells, CA. Available at: https://earthquake.usgs.gov/earthquakes/eventpage/ci38488354/executive (accessed 22 November 2022).

62.

USGS ShakeAlert (2020b) M 5.8–18km SSE of Lone Pine, CA. Available at: https://earthquake.usgs.gov/earthquakes/eventpage/ci39493944/executive (accessed 22 November 2022).

63.

USGS ShakeAlert (2020c) USGS ShakeAlert on X—Anza. Available at: https://twitter.com/USGS_ShakeAlert/status/1246265652471455744 (accessed 22 November 2022).

64.

Vinnell

Tan

Prasanna

Becker

(2023) Knowledge, perceptions, and behavioral responses to earthquake early warning in Aotearoa New Zealand. Frontiers in Communication 8: 1229247.

65.

Wald

Collett

(2021) The 2019 Ridgecrest, California earthquake sequence. Available at: https://earthquake.usgs.gov/storymap/index-ridgecrest.html (accessed 22 November 2022).

66.

Whitaker

Stevelink

Fear

(2017) The use of facebook in recruiting participants for health research purposes: A systematic review. Journal of Medical Internet Research 19(8): e290.

67.

Wojton

Porter

Lane

Bieber

Madhavan

(2020) Initial validation of the trust of automated systems test (TOAST). The Journal of Social Psychology 160(6): 735–750.

68.

Wood

Mileti

Bean

Liu

Sutton

Madden

(2018) Milling and public warnings. Environment and Behavior 50(5): 535–566.

69.

Woodyard

(2019) “Scared to death”: 7.1 earthquake follows 6.4, leaving Ridgecrest residents fearful, resilient. Available at: https://www.usatoday.com/story/news/nation/2019/07/06/california-earthquake-7-1-follows-6-4-rattling-ridgecrest-residents/1663774001/ (accessed 22 November 2022).