1. Introduction
In June 2021, the United States experienced unprecedented heat waves from coast to coast, with multiple states breaking all-time heat records (Di Liberto 2021). These extreme heat events were especially devastating in the Pacific Northwest region of the United States, with 100 deaths occurring in Washington State alone (Washington State Department of Health 2021). Indeed, extreme heat is the deadliest weather-related hazard, killing more Americans on average than flooding, tornadoes, and hurricanes combined (National Weather Service 2021). However, as compared with these other hazards, the public does not typically perceive heat as a dangerous threat (Hass and Ellis 2019; Kalkstein and Sheridan 2007; Lambrecht et al. 2021). In addition, vulnerable populations, such as older adults over the age of 65, generally do not believe they are at greater risk of adverse heat outcomes, thus decreasing the likelihood they will protect themselves during extreme heat events (Kemen et al. 2021).
Given these findings, there has been a rise in communication interventions intended to increase the public’s perceived severity of, and perceived vulnerability to, heat risks and increase their adoption of risk-mitigating behaviors (Bassil and Cole 2010). These communication interventions can be disseminated on social media channels to at-risk populations (Lambrecht et al. 2021; Li et al. 2018). Social media represents a direct channel for organizations, such as National Weather Service (NWS) weather forecast offices (WFOs), to communicate actionable information during extreme heat events to the public (Olson et al. 2019).
However, prior reviews of heat-related social media messages have found that NWS WFOs disproportionately emphasize the hazard’s intensity (i.e., temperature or heat index values), with less information focused on protective action guidance, impacts, and population vulnerabilities (Li et al. 2018). These assessments indicate that there is an opportunity for improvements to NWS WFO heat messaging on social media. Yet, more nuanced messaging recommendations cannot be made without first understanding how NWS risk communicators currently communicate about heat to at-risk publics.
Accordingly, the purpose of this study is to evaluate heat communication interventions of NWS WFOs on Twitter (now known as X). NWS WFOs are the official organizations that issue heat-related watches, warnings, and advisories to local geographical areas in the United States (Hawkins et al. 2017). WFOs also communicate the messages from the Weather-Ready Nation (WRN) program of the NWS by drawing from risk communication campaign materials developed at the federal level for dissemination on social media channels. As a free and publicly accessible risk communication channel, Twitter represents a way for NWS WFOs to communicate this information directly with the public(s) (Olson et al. 2019). Because of this, Twitter represents a site for researchers to observe organizational risk communication practices, including the words selected to communicate about hazardous conditions, vulnerable populations, and protective actions. Although prior Twitter researchers have taken a case study approach by investigating organizational follower numbers and public engagement with messages (see, e.g., Silver and Andrey 2019), our approach centers on quantifying the content and features of public facing messages that communicate about heat.
Specifically, we focus on the content of NWS WFO communication during threat periods—or periods in which there is an excessive heat watch, heat advisory, or excessive heat warning in effect for a particular location. We conduct a quantitative content analysis of tweets sent by seven WFOs during the summer of 2021 to identify language used to express the hazard and its impacts, vulnerable populations, and the actions people can take to protect themselves. From this analysis, we identify the frequency of embedded communication assumptions that may inhibit the effectiveness of NWS heat communication interventions. By identifying these assumptions and messaging gaps, improved NWS heat risk communication interventions can be developed and tested with the public in the future.
2. Literature review
As a result of climate change, extreme heat events will continue to increase in both frequency and severity worldwide (Seneviratne et al. 2021). To mitigate heat-related mortality and morbidity, numerous cities and regions across the world have developed heat-health warning systems (HHWSs; e.g., Matzarakis et al. 2020; Williams et al. 2022; Wu et al. 2020). HHWSs activate public health interventions that aim to reduce heat exposure and impacts when heat forecasts are predicted to reach a certain threshold (Kotharkar and Ghosh 2022). Although the thresholds and interventions vary depending on location, all HHWSs have a public communication component that focuses on notifying those at risk of dangerous heat conditions (Casanueva et al. 2019; Grundstein and Williams 2018; Kalkstein and Sheridan 2007).
A scoping review conducted by Mayrhuber et al. (2018) assessed the effectiveness of these public health heat interventions, including heat communication interventions, by examining 23 studies. They found that prior heat interventions that have been implemented to change risk perceptions and behavior for at-risk populations and their caregivers start from the premise of embedded assumptions that could decrease their persuasiveness. These assumptions are as follows:
-
Informing people about heat and heat dangers will lead to behavioral change.
-
At-risk individuals will recognize their own vulnerability, which will lead to concern and awareness.
-
The benefits of heat advice are commonly understood and taken seriously.
-
Caretakers of vulnerable groups have the capacity and training to engage in heat reduction measures.
As Mayrhuber et al. (2018) note, these four assumptions are “problematic, as they determine the actual effect of interventions and how vulnerable persons and groups may be reached” (Mayrhuber et al. 2018, p. 52). By conforming to these assumptions, risk communication interventions about extreme heat have largely focused on informing message recipients about the hazard itself, rather than conveying the seriousness of the threat and the actions they should take to prevent harm. These findings also suggest that communication about heat risk, heat vulnerability, and protective actions can, and must, be improved to prevent losses as excessive heat events continue to increase over time. We provide an in-depth overview of these four assumptions below, which informs our six research questions (RQs).
a. Assumption 1: Informing people about heat and heat dangers will lead to behavioral change
One common tool for communicating about heat risk is via excessive heat warnings, excessive heat watches, and heat advisories sent by the NWS (Hawkins et al. 2017). These terms—or “products” as the NWS describes them—constitute warning “signal words,” which are used to “attract attention to the warning and indicate the level of hazard present” (Wogalter et al. 2002, p. 221).
The goal of heat products is to “[increase] heat awareness for the population as a whole and implicitly encourage individual actions to protect their health during extreme heat events” (Benmarhnia et al. 2019, p. 5). Prior research has found that the public is often aware of when heat events occur and/or when heat products are in effect, yet do not typically change their behavior in response to heat-related warnings despite receiving this information (Bassil and Cole 2010; Grundstein and Williams 2018; Kalkstein and Sheridan 2007; Lane et al. 2014; Madrigano et al. 2018; Sheridan 2007). This finding is unsurprising, as informing someone about a risk is a separate act (with separate goals) from persuading an individual to do something, such as adopting heat risk mitigating behaviors (Rossi and Macagno 2021).
Specifically, informing helps increase comprehension and understanding of an issue, whereas persuasion helps increase the likelihood of individuals accepting protective action recommendations. And although informing and persuading both rely on logical proofs (i.e., the relationship between premise and conclusion), persuasion—or changes in attitudes, intentions, and behaviors—requires the message sender to provide additional information intended to increase message recipients’ perceived severity of the situation and their susceptibility (Witte 1992). For example, assume that a message states, “you should hydrate yourself (conclusion) because there is a heat advisory in effect (premise).” For this information to increase one’s likelihood of accepting the message conclusion (i.e., hydrate), the message recipient must be able to infer the relationship between heat advisory, its consequences, and the need to be hydrated. Here, message recipients must (i) know what a heat advisory is and why it is dangerous, (ii) see themselves at risk, (iii) understand how the recommended behavior will protect themselves, and (iv) have the resources and confidence needed to act (Reynolds and Seeger 2005). Therefore, it is not “the information per se which is convincing or not, but what the receivers make of it” (Wänke and Reutner 2010, p. 2).
Mayrhuber et al.’s (2018) first assumption also focuses on how heat dangers are communicated. This type of information can include a description of the hazard—or how heat and its intensity are described via meteorological conditions (e.g., temperature, heat index; Li et al. 2018), as well as information related to how heat will personally impact an individual (Mileti 2018). These types of information can increase one’s perceived severity, which strongly predicts the extent to which individuals adopt heat mitigation behaviors (Ban et al. 2019).
Thus, to measure the frequency of Mayrhuber et al.’s (2018) first assumption, which focuses on how heat (i.e., the hazard) is expressed to the public, we look at the frequency of heat signal words, heat descriptions, and heat impacts via the following three RQs:
-
RQ1—How frequently do NWS WFO tweets include signal words?
-
RQ2—How frequently do NWS WFO tweets include a description of heat?
-
RQ3—How frequently do NWS WFO tweets include the dangers (i.e., impacts) of heat?
Next, we turn to Mayrhuber et al.’s (2018) second assumption, which speaks to message recipients’ perceived vulnerability.
b. Assumption 2: At-risk individuals will recognize their own vulnerability, which will lead to concern and awareness
For messages to motivate action, message recipients must personalize their risk, defined as “the perceived implications of the risk being communicated on the receivers themselves” (Mileti and Peek 2000, p. 184)—or the belief that one will be severely affected in some capacity. Behavior change depends on an individual’s ability to accurately assess the severity of the situation and feel vulnerable or susceptible to its effects (Witte 1992).
However, prior research has found those at higher risk of experiencing heat-related impacts typically do not perceive themselves as vulnerable (Bassil and Cole 2010); instead, vulnerable populations often feel that heat-related messages do not personally apply to them. For example, older adults (often described as “elderly” but typically defined as those over the age of 65) are some of the most vulnerable to heat consequences because of a decrease in thermoregulation ability that comes with age (Cheng et al. 2018). Despite this risk, older adults often do not believe that they are more vulnerable to negative heat and health consequences than other groups. Indeed, in countries like Germany (Kemen et al. 2021), the United Kingdom (Abrahamson et al. 2009), Australia (Banwell et al. 2012), and the United States (Lane et al. 2014), prior work shows that older adults do not perceive themselves as more vulnerable than others.
Similar findings emerge for other vulnerable groups. For example, Williams and Grundstein (2018) interviewed 25 parents and caregivers about leaving and/or forgetting children in hot cars, an act that can result in the child’s death. Here, many participants believed they would never forget their child in a hot car, indicating “I would never forget my grandkids in the car” and “how can you forget your child in the car?” (Williams and Grundstein 2018, p. 5). Instead, it would be other parents and caregivers, such as single and/or low-income parents, that would be most likely to forget their children in the car.
Given the importance for message recipients to identify their own vulnerability to heat impacts, we pose the following RQ intended to capture Mayrhuber et al.’s (2018) second assumption regarding vulnerable populations:
-
RQ4—How frequently do NWS WFO tweets include specific vulnerable populations?
Next, we discuss Mayrhuber et al.’s (2018) third assumption, which focuses on protective action decision-making.
c. Assumption 3: The benefits of heat advice are commonly understood and taken seriously
Protective action statements embedded within a message are referred to as guidance (Mileti and Peek 2000; Mileti and Sorensen 1990). Mayrhuber et al.’s (2018) third assumption relates to the idea that individuals inherently know why certain types of guidance are recommended. If this connection is unknown, this can negatively impact one’s decision-making or deciding (Mileti and Peek 2000; Mileti and Sorensen 1990). Deciding can be defined as message recipients contemplating what to do about a risk (Wood et al. 2018). At this stage in decision-making, individuals often ask themselves “what can be done to achieve protection?” (Lindell and Perry 2012, p. 622) and if a behavior is an effective way to protect themselves (Witte 1992). Here, message recipients consider what action(s) to perform and if they are effective and feasible given their situation (Lindell and Perry 2012).
Providing detailed information about why a behavior is recommended helps connect the effectiveness of the action to one’s safety (Lindell and Perry 2012). However, Mayrhuber et al. (2018) find that of the 23 studies sampled, communication interventions “provide merely descriptions on heat advice (“stay hydrated,” “avoid heat,” and “check on vulnerable people in your social network”) but the mechanisms of how exactly behavior can be changed and what models could be used largely remain unexplained” (Mayrhuber et al. 2018, p. 51).
Thus, providing the reason as to why a behavior is recommended is a key component of heat messaging. This type of information has also been found to increase self-efficacy or one’s confidence to act (Frisby et al. 2013), which is an important factor in motivating response (Sutton et al. 2021). Self-efficacy is especially important for heat risks, whereby one must feel they can effectively mitigate the impacts of extreme heat through performing certain behavior(s) (see McLoughlin et al. 2023). Therefore, we pose the following RQ intended to capture Mayrhuber et al.’s (2018) third assumption related to guidance information:
-
RQ5—How frequently do NWS WFO tweets include guidance information?
Last, we discuss Mayrhuber et al.’s (2018) fourth assumption, which relates to how guidance information is directed to heat vulnerable populations and those responsible for their care.
d. Assumption 4: Caretakers of vulnerable groups have the capacity and training to engage in heat reduction measures
Recommendations from the Centers for Disease Control and Prevention and other organizations that address the impacts of extreme heat advise individuals to check on neighbors, especially older adults, to assess their hydration levels, cooling mechanisms, and signs of heat stress (Centers for Disease Control and Prevention 2021). However, many caretakers do not have the training, resources, and/or skills necessary to effectively help older adults. For example, Malmquist et al. (2022) interviewed 19 Swedish nursing home residents aged 61–92. These participants reported that their nursing home staff either lacked the time or capacity to engage in heat reduction measures for themselves and other residents.
If individuals are advised to check on or assist their neighbors or family members but have limited knowledge, training, time, and/or resources, or are unaware of the added dangers of extreme heat, messaging interventions that only advise one to “check on neighbors and/or older residents” may be insufficient to reduce heat consequences for this vulnerable population. Specifically, if there are steps required to successfully “check on others,” these should be (i) included in the message, (ii) logically ordered, and (iii) actionable to increase one’s perceived effectiveness of performing the recommended behavior (Frisby et al. 2013, 2014).
Thus, to capture Mayrhuber et al.’s (2018) fourth assumption regarding how guidance is presented to vulnerable populations and those responsible for their care, we pose the following RQ:
-
RQ6—How frequently do NWS WFO tweets include guidance directed at caregivers of specific vulnerable populations?
e. Summary
The purpose of this study is to evaluate the language contained in heat communication interventions of NWS WFOs on Twitter. We pose six research questions intended to quantify certain message characteristics that capture the frequency to which NWS WFOs embed Mayrhuber et al.’s (2018) four assumptions in their heat risk messaging via Twitter. This assessment helps evaluate heat information effectiveness overall.
3. Method
a. Identifying data sources and data collection
First, we identified a geographically representative sample of cities and/or regions around the United States that experienced excessive heat events in the summer of 2021. These locations included Albany, New York; New York (NYC), New York; Huntsville, Alabama; New Orleans, Louisiana; Raleigh, North Carolina; Portland, Oregon; and Seattle, Washington(Table 1). Then, we identified the Twitter username for the NWS WFO that would be responsible for communicating about high-impact weather events in each location. Then, we narrowed our search to focus on the tweets sent during the first excessive heat event of the season (with the exception of NWS Albany and NWS NYC, where we collected tweets from a second heat event for additional data and to make more targeted recommendations for the region in which the authors reside). The first heat event of the season has been described as the most dangerous because people have not yet acclimated to high temperatures or have not prepared their homes/environments for hot weather (Anderson and Bell 2011).
WFO location, dates, and number of tweets collected.
We identified the first heat event of the season via the Iowa State Mesonet database by identifying the dates when NWS heat products (i.e., heat advisory, excessive heat watch, and excessive heat warnings) were first issued. Then, using the “Advanced Search” functionality on Twitter, we searched the terms “heat,” “hot,” and “temperature” within the messages sent by WFOs identified in each location. We narrowed the date of our search to 5 days prior to the first NWS watch, warning, or advisory to ensure all heat-related messages were captured. We ended data collection for a particular location when no more heat content emerged. This resulted in a total of 250 tweets to analyze. Table 1 provides the date ranges and number of tweets for each location.
Each individual original tweet (i.e., nonreply) was downloaded and all images embedded within a tweet were saved separately.
b. Data coding
We coded both the tweet text and image text using a deductive content analysis approach. First, we used the concepts in the warning response model to create overarching categories about the hazard, including its description and impact, the location of impact, protective action guidance, timing of event, and message source (Kuligowski et al. 2023; Mileti and Peek 2000; Mileti and Sorensen 1990; Sutton and Kuligowski 2019). Based on prior research that applies the warning response model to short messaging (e.g., Bean et al. 2022; Kuligowski et al. 2023), we define these categories, and provide examples, in Table 2.
Coding category, definition, and example.
Within each category, we then developed subcategories based on the exact language used in the tweets (see Table 3). We used this binary and nonthematic system of coding to indicate whether a content category was present in each tweet. We coded the textual content in the message and the image separately for each tweet, and then combined the codes to inform the overall message content frequency. Codes were not mutually exclusive, thereby any one tweet can contain multiple categories and subcategories. Any category or aspect not accounted for in our categories was noted as “other.” Initial coding was conducted by the third author and then discussed with the lead author to confirm agreement.
Coding categories (highlighted in boldface type), subcategories, and frequencies.
4. Results
Table 3 provides the overall number of tweets coded for each category in a single tweet (i.e., either text or image). We provide a more in-depth discussion of these categories and how they were communicated and coupled with other types of information below.
a. Signal words
Signal words include heat advisory, excessive heat watch, and excessive heat warning. Approximately 23.6% (n = 59) of tweets included a heat signal word. Heat advisory was the most common signal word (20.8%; n = 52). Heat advisories are the most common heat product issued by the NWS (Hondula et al. 2022), and many tweets state that a heat advisory is in effect and/or is expanded, supporting this assertion with an image of the geographic area(s) at risk (15.2%; n = 38). However, only one tweet explained and defined what a heat advisory was and the criteria for their issuance in their location (see Fig. 1).
One tweet defined what a heat product means for their area (https://twitter.com/NWSNewOrleans/status/1418632462939435009).
Citation: Weather, Climate, and Society 15, 4; 10.1175/WCAS-D-23-0033.1
b. Hazard description
A hazard description was included in approximately 96% of tweets (n = 239). Most frequently, tweets referenced temperature (word or numeric value; n = 202; 80.8%) and “heat index” (n = 97; 38.8%).
Heat advisories are issued based on heat index values (Hawkins et al. 2017). Heat index accounts for temperature and humidity (Hondula et al. 2022)—or what a temperature “feels like” given humid conditions (National Weather Service 2022). However, the connection between heat advisory and heat index was not explained in our sample. Specifically, heat advisory messages did not indicate the conditions that triggered an advisory and how heat index plays a role. Instead, 15.2% of tweets (n = 38) imply the link between heat advisory (a signal word) and heat index (a description of the hazard) by presenting both pieces of information within a single message (through tweet text, image, or both; see Fig. 2) but did not describe the connection between the two.
An example tweet that provides a heat warning/advisory map and associated heat index values side by side, yet the connection between these two types of information is not included (https://twitter.com/NWSHuntsville/status/1421187058127802374).
Citation: Weather, Climate, and Society 15, 4; 10.1175/WCAS-D-23-0033.1
Furthermore, when heat index and/or heat index values were mentioned, these terms were not explained nor defined for the audience (see Fig. 3 for an example). However, in some instances, WFOs indirectly indicate what heat index is by mentioning heat index and humidity within the same tweet (9.2%; n = 23).
An example tweet mentioning “heat index” (annotated with red boxes by authors), yet heat index is not defined (https://twitter.com/NWSNewOrleans/status/1420478054917357570).
Citation: Weather, Climate, and Society 15, 4; 10.1175/WCAS-D-23-0033.1
Some tweets also reference “apparent temperature” (6%; n = 15), which is another term that was undefined for the audience. Apparent temperature is the “feels like” temperature (Steadman 1984) and is a common way heat index is referenced by the meteorological community.
Other hazard descriptions that might affect personal safety were also mentioned, including “(high) temperature(s)” (80.8%; n = 202), as well as terms that reference the severity of the heat event, including “excessive heat” (10.4%; n = 26), “extreme heat” (6.0%; n = 15), and “dangerous heat” (6.0%; n = 15). “Heat wave” was also mentioned in 12% of tweets (n = 30); however, “heat wave” lacks a common or universal definition within the meteorological community (Conti et al. 2022) and was also not defined in these tweets.
c. Hazard impacts
We next look at the frequency and use of heat impact information. Contents about heat impacts were included in approximately 30.8% (n = 77) of tweets.
Heat illness was the most frequently mentioned specific heat impact (12.4%; n = 31). However, “heat illness” represents a catchall term for many specific types of heat impacts, including heat cramps, heat exhaustion, heat stroke, sunburn, and heat rash (Centers for Disease Control and Prevention 2017). When heat illness was mentioned, the specific illnesses and their associated symptoms were not described (see Fig. 4).
An example tweet indicating that message receivers should drink water and avoid time outside to “prevent heat-related illness,” yet what heat-related illness is and its associated symptoms are not defined (https://twitter.com/NWSRaleigh/status/1420808514814070788).
Citation: Weather, Climate, and Society 15, 4; 10.1175/WCAS-D-23-0033.1
Death—or the most severe consequence of exposure to extreme heat—was the second most common specific impact (8%; n = 20). In a few instances, specific illnesses, such as heat stroke (6.8%; n = 17) and heat exhaustion (6.4%; n = 16) were mentioned, but their symptoms were often not included, except in one image that was included in nine tweets overall (3.6%; see Fig. 5).
Nine tweets included this image that specifies the symptoms of heat exhaustion and heat stroke (e.g., see https://twitter.com/NWSRaleigh/status/1420834158968987651).
Citation: Weather, Climate, and Society 15, 4; 10.1175/WCAS-D-23-0033.1
Additional types of impacts that were mentioned only once (and thus were classified as “other”) include those related to pets (e.g., excessive panting, pavement burning pet paws), pregnant women and children (e.g., low birth weight, preterm birth, infant mortality), and other types of heat health impacts, such as heat cramps and heat rash.
d. Vulnerable populations
Approximately 36% of tweets (n = 91) mentioned an at-risk or vulnerable population. Of these groups, those engaging in outdoor recreational activities (22.8%; n = 57) and outdoor workers (22.4%; n = 56) were most frequently included, followed by pets and children (16.4%; n = 41). The “elderly” were mentioned much less frequently (5.6%; n = 14) and at the same rate as those with access and functional needs (A&FN).
Furthermore, why these groups are vulnerable to heat impacts is rarely included, except for one type of image, which is available via the NWS WRN Program and was included in 2.8% (n = 7) of tweets sent by four WFOs (see Fig. 6). The information in this image explains why specific groups are vulnerable by indicating that “age and certain conditions make the body less able to regulate temperature.” However, given the infrequent use of this image, tweets did not generally mention why specific groups are vulnerable to negative heat impacts.
Seven tweets included this image that specifies conditions that make someone vulnerable to heat impacts (e.g., see https://twitter.com/NWSSeattle/status/1407420578588291073).
Citation: Weather, Climate, and Society 15, 4; 10.1175/WCAS-D-23-0033.1
e. Guidance
Of the 250 tweets sampled, 57.2% (n = 143) included some form of guidance information. The most recommended specific protective actions include drinking water/hydrating (40%; n = 100), limiting outdoor activity (26%; n = 65), wearing light/loose fitting clothing (16.8%; n = 42), “keeping cool” (12.4%; n = 31), and taking breaks if outdoors (12%; n = 30).
Other protective actions that were mentioned but did not fit into these categories (and thus are defined as “other”) include actions such as altering one’s diet (e.g., “eat foods with lots of water in them,” “eat light,” or “avoid alcohol”), checking for additional information (e.g., “check the weather forecast”; check for tips at an external source or website), and seeking emergency services if needed (e.g., “call 911 if experiencing heat symptoms”; “call 911 if you see a pet or child in a hot car”).
The hazard, its impacts, and what to do in response was mentioned in 25.6% of tweets (n = 64); thus, why recommended protective actions help mitigate the impacts of extreme heat (e.g., the link between heat, heat exhaustion, and hydration) was not commonly included.
f. Guidance for caregivers
Approximately 8% of tweets advise message receivers to check on others (n = 20). In looking more closely at these tweets, seven (2.8%) specified to check on friends, family members/relatives, “loved ones,” neighbors, and/or pets. In addition, eight of the tweets in our dataset (3.2%) posted by five separate WFOs included the image in Fig. 7, which indicated to “check up on the elderly, sick, and those without AC.” One tweet mentioned “check on those who are sensitive to extreme heat,” but did not define who these groups are. Importantly, none of the tweets instructing one to “check on others” or another population indicated (i) what to check for (i.e., symptoms), and (ii) what to do in response.
Eight tweets included this image that asks message receivers to “check up on the elderly, sick, and those without AC” (e.g., see https://twitter.com/NWSAlbany/status/1408349284932718595).
Citation: Weather, Climate, and Society 15, 4; 10.1175/WCAS-D-23-0033.1
5. Discussion
Although individual NWS WFOs have their own standards for communication (Olson et al. 2019), general patterns emerge that reveal important themes about NWS extreme heat communication on social media. We intertwine a discussion of our findings with the assumptions found in Mayrhuber et al. (2018). Then, based on our findings as to where messaging gaps exist in current NWS WFO heat communication, we draw upon the best practices for warning message design and other bodies of allied communication literature to craft recommendations for heat messaging improvements related to each research question in Table 4.
Recommended message design strategies for heat risk communication.
a. Research questions 1, 2, and 3: Informing people about heat and heat dangers will lead to behavioral change
Hazard description information was, by far, the most frequent type of heat information included, with approximately 96% of tweets describing the hazard in some capacity. Other investigations of NWS heat-related social media messages found a similar pattern, whereby words used to convey the intensity of the hazard itself (e.g., temperature) are overemphasized relative to all other types of information (e.g., protective action statements; Li et al. 2018). One of the most common ways WFOs described heat was by noting the “heat index.” However, what heat index is and why high values are dangerous were not mentioned.
Furthermore, attempting to contextualize heat risk through communicating heat impacts (i.e., the consequences of extreme heat on the individual) was only included in approximately 31% of tweets. When heat impacts were mentioned, they were frequently described as “heat illness.” However, what constitutes heat illness, as well as its symptoms, was not commonly included. Thus, our results indicate that NWS WFOs disproportionately describe the severity of the hazard itself, rather than how the hazard will personally affect those at risk. They also convey this information using vague or undefined terms.
NWS WFOs commonly used undefined signal words (i.e., watch, warning, advisory) to alert people of a heat risk. Indeed, our results show that these heat signal words were a common way that heat risk was communicated, but these terms were rarely defined for message recipients.
As specialized technical terms, words like “heat advisory,” “heat index,” and “heat illness” can be classified as jargon—or terms that have meaning to NWS WFOs but may not be readily understood outside that group. These findings also suggest that NWS WFOs ascribe to the same assumptions about heat communication that Mayrhuber et al. (2018) found in their review of heat communication interventions—that informing people about the danger of heat and particular risks will make them aware and adapt their behavior according to advice. Informing or alerting people to danger via signal words and coupling these terms to their vague conditions and undefined impacts will not persuade message recipients to engage in adaptive behavior(s) (Mileti and Peek 2000).
Experts’ tendency to conflate informing and persuading speaks to a noted phenomenon in science communication called the “knowledge gap” or the “information deficit model,” which assumes that “people take action when they learn of a problem” (Snyder 2007, p. S34). This model assumes that bringing awareness of an issue via providing information will lead to acceptance of message recommendations and behavior change. However, scholars argue that providing information alone will not, in most cases, lead to adoption of risk-mitigating behaviors (Seethaler et al. 2019; Simis et al. 2016; Trench 2006). Instead, messages must be carefully crafted to include convincing arguments as to why an audience should engage in or stop certain behaviors (Silk et al. 2011).
Therefore, risk communicators need to first determine if the goal of their communication intervention is to inform or persuade (Oxman et al. 2022). Those who wish to inform their audience about a particular issue (i.e., increase understanding) should use plain language rather than technical terms and jargon, which can negatively affect message processing and understanding (Sivle and Aamodt 2019). However, if the goal of a message is to change perceptions (i.e., increase feelings of danger) or motivate one to perform protective action(s), message senders must move from a model of informing to a model of persuasion.
Persuasion first depends on individuals recognizing the severity of the situation (Witte 1992). Based on our finding that hazard impact information is included at a much lower rate than hazard description information, and this information was typically conveyed using jargon, we recommend that the consequences of heat to one’s health not only be included, but also clearly defined. In Table 4, we offer these message design recommendations, as well as additional studies that can be referenced to help develop these statements.
However, message recipients must also feel they are at risk before behavior change can occur (Witte 1992). This relates to the second assumption found by Mayrhuber et al. (2018) and our fourth research question.
b. Research question 4: At-risk individuals will recognize their own vulnerability, which will lead to concern and awareness
Our findings indicate that approximately 64% of tweets sent by NWS WFOs did not mention specific vulnerable populations. Similar findings have emerged from prior research by Li et al. (2018), which found that, relative to severity information, vulnerability information is less emphasized in NWS heat-related social media messaging. Furthermore, only 5.6% of messages focused on the aging, who are considered one of the most vulnerable groups to negative heat impacts (Cheng et al. 2018; Conti et al. 2022).
Overall, messages did not frequently include why certain populations are particularly at risk. For example, one image appended to approximately 3% of tweets indicate the factors that make someone vulnerable to heat consequences (see Fig. 6). This means that the connection between heat and vulnerability is overwhelmingly implied, and NWS WFOs are assuming that at-risk individuals will recognize their own vulnerability and feel concern (Mayrhuber et al. 2018).
Based on these findings, we recommend including vulnerability information at the same rate as severity information to help people personalize their risk (Mileti and Peek 2000). Furthermore, vulnerability information should also connect the hazard to the conditions that make someone vulnerable. In other words, if vulnerable populations are identified in a message, risk communicators must explain why they are vulnerable, which provides justification for why certain actions should be taken (see Table 4), which is discussed next.
c. Research question 5: The benefits of heat advice are commonly understood and taken seriously
Our findings indicate that the connection between the hazard and the recommended protective action was not frequently included, whereby approximately 57% of tweets included guidance information but only 25.6% of tweets provided an implied connection between the hazard, its impacts, and the recommended behavior(s). Although we do not know if these messages are understood or taken seriously, results show that the benefits of heat mitigation behaviors are not clearly communicated.
Heat communication messages should provide instructions about the recommended actions and justify why they should be performed, as these types of information can increase perceived response efficacy (i.e., the belief that the behavior is effective in mitigating a threat) and self-efficacy (i.e., confidence in their ability to act; Frisby et al. 2013, 2014; Murray-Johnson and Witte 2011). Both perceived response efficacy and self-efficacy have a powerful effect on behavior change (Floyd et al. 2000; Milne et al. 2000). Therefore, the association between the hazard and the protective action needs to be included in a message, as individuals should not have to infer why the behavior is recommended (see Table 4; Mileti and Peek 2000; Witte 1992).
Note that, even if one understands the connection between the hazard and protective action(s) and their willingness to act is high, barriers to action can inhibit adoption of risk-mitigating behavior(s) (Grundstein and Williams 2018). These barriers can include one’s capability to act (e.g., physical mobility, literacy) and/or opportunities and access to interventions (e.g., vehicle to drive to cooling center, financial resources to invest in or run air conditioning (AC); Hayden et al. 2017; Madrigano et al. 2018; Mayrhuber et al. 2018). These barriers are also present in Mayrhuber et al.’s (2018) fourth assumption, assessed by research question 6.
d. Research question 6: Caretakers of vulnerable groups have the capacity and training to engage in heat reduction measures
Mayrhuber et al. (2018) note that it is often assumed that caretakers of vulnerable groups possess the capacity to intensify care during extreme heat events and are sufficiently trained in thermoregulation and heat reduction measures. We measured this assumption by looking at the number of tweets that tell people to “check on others.”
When examining these messages in detail, they primarily indicate who to check on, not how to do so. They do not indicate what symptoms to look for, how to help, and if resources are available. These messages assume that others (i) truly understand dire conditions that excessive heat will cause, (ii) have an awareness of their neighbors and good relations, and (iii) have the resources and skills necessary to care for others.
“Check on others” also assumes a safety net composed of family and neighbors rather than other societal resources. One potential, but unintended, consequence of this approach is that it shifts the lack of extreme heat mitigation and response to one of individual neglect. As Klinenberg (2015) argues, those with the greatest needs and who are entitled to social protection are the least likely to receive help, with their caregivers most likely to be blamed.
Furthermore, the majority of public health and communication campaigns focus on educating individuals to change behavior. As Dorfman and Wallack (2012) argue “[public communication campaigns] present knowledge as the ultimate power—armed with the right information, people can control their health destinies” (p. 566). Yet these campaigns neglect the social, economic, environmental, and policy factors that influence our health outcomes (i.e., the structural determinants of health). By focusing on individual change, rather than the underlying systemic constraints that affect our health, public communication campaigns rarely lead to improved population-level health outcomes (Viswanath et al. 2021).
e. Limitations
We identified the following limitations of our quantitative content analysis of heat-related tweets sent by seven NWS WFOs during the summer of 2021.
First, we assessed content sent by seven WFOs across the country for one heat event each (except for NWS Albany and NWS NYC, which included two events each). Although we attempt to identify a geographically representative sample, assessing more NWS WFO heat Twitter content within each region may be necessary to further refine the frequency of message content types and provide messaging improvements. Furthermore, a more nuanced analysis may be needed to describe differences in heat-related message content in the western parts of the United States, where heat products, as well as humidity, are less common (National Weather Service 2021).
Second, we coded WFO tweet text and the text contained in the images, rather than the composition of the images themselves. Therefore, we cannot offer any visual design or presentation suggestions beyond image text. Furthermore, we only analyzed the initial tweet rather than connected tweets or replies.
Third, because our focus is on the language and content of heat risk communication, we do not determine if these tweets were received and attended to, and by whom. However, prior research has found that the demographics of Twitter users have implications for how organizations should target specific audiences on this platform. For example, those aged 65 and older account for only 7% of total Twitter users (Dinesh and Odabas 2023), suggesting that those who have been identified as being among the most vulnerable will not be easily reached directly on Twitter. However, Twitter may be an appropriate place to reach younger caregivers for those who are aging.
We do not know if these tweets lead to behavior change, because we did not conduct any outcome evaluation. Although we suggest additions to tweet content to increase the likelihood of behavior change overall, we also recognize that there is a character limit on this channel and thus, not all suggestions can be incorporated into a single tweet. However, images and “threaded” or connected messages can help add additional message content beyond the 280-character limit for non–Twitter Blue users (Twitter 2022).
f. Future research
The most logical next step is to evaluate the messages we examined in this study to determine the extent to which they are (i) understood, (ii) believed, and (iii) personalized and the extent to which they influence behavior change (Mileti and Peek 2000; Mileti and Sorensen 1990). However, we identify other areas of research needed for informational and persuasive heat communication interventions beyond testing these preexisting messages. These next steps rely on risk and health communication experts to conduct systematic research that blends theoretical health communication principles with warning message design. Ultimately, heat risk communicators need messaging strategies that move beyond providing information, suggesting more persuasive and theoretical work is needed to understand the drivers of behavior in this area (e.g., Valois et al. 2020).
To develop more persuasive heat communication interventions, researchers should first engage with the target audience(s) via interviews or focus groups. Arguably the most important activity in developing heat messaging interventions is to gain a better understanding of audience(s) prior to message development, as we cannot assume we know the underlying problems to address without first interacting with an audience (Zhao 2020).
Then, messaging interventions can be developed for both the generalized public and specific vulnerable populations based on audience engagement (Grundstein and Williams 2018). Prior research has found that generalized advice regarding heat-related behaviors and heat advice have a higher chance of being ignored and perceived as irrelevant (Li and Howe 2023; Lowe et al. 2011); thus, targeting specific vulnerable populations is more likely to have an effect than messages developed for everyone. Targeted messaging entails engaging in some form of audience segmentation, which creates homogeneous groups by “[dividing] a population, market, or audience into groups whose members are more like each other than members of other segments” (Grunig 1989, p. 202). Demographic factors are a common way to segment audiences. For example, by developing messaging interventions for older adults who are “aging into” vulnerability, the persuasive effects of an intervention will be improved by addressing differences in message processing (Guido et al. 2021) and how they identify themselves (Grundstein and Williams 2018). However, other principles may also be used to segment audiences (e.g., stages of change, self-efficacy, social context, media usage, belief in climate change; Silk et al. 2011; Zhao 2020).
Future research should consider broadening this type of investigation beyond a single governmental agency within the United States. First, the social media messages of other organizations responsible for communicating heat risks and protective actions, like the Centers for Disease Control and Prevention, should be examined. In addition, heat information presented on other social media platforms and channels, such as governmental websites (e.g., the National Integrated Heat Health Information System; https://www.heat.gov/) should also be assessed for the extent to which Mayrhuber et al.’s (2018) assumptions are present. How heat alerts and warnings are communicated on social media outside the United States should also be evaluated to assess HHWS effectiveness (e.g., Watson and Finn 2014).
6. Conclusions
NWS WFOs communicate about heat with several assumptions about how people understand and react to heat-related information (Mayrhuber et al. 2018). On social media specifically, these organizations communicate to the public in a way that could ultimately hinder behavior change. First, they commonly include technical terms or “jargon” that may not be readily understood by message receivers. Second, they do not indicate why vulnerable populations are particularly at risk for negative heat impacts. Third, they do not connect why certain protective actions are necessary to reduce extreme heat consequences. Fourth, they ask message receivers to “check on others” but do not provide instructions on how to effectively do so. Without including these types of information, the persuasiveness of NWS heat communication interventions on social media will be greatly reduced. By identifying these NWS messaging gaps and assumptions, we provide the foundational knowledge needed to develop improved NWS heat risk communication interventions on social media.
Acknowledgments.
This project is supported by the NOAA WPO Observations Grant NA21OAR4590360. The content is solely the responsibility of the authors and does not necessarily represent the official views of NOAA. We thank Elisabeth Dubois for her work on collecting the initial data used in this study.
Data availability statement.
Data may be made available upon request.
REFERENCES
Abrahamson, V., J. Wolf, I. Lorenzoni, B. Fenn, S. Kovats, P. Wilkinson, W. N. Adger, and R. Raine, 2009: Perceptions of heatwave risks to health: Interview-based study of older people in London and Norwich, UK. J. Public Health, 31, 119–126, https://doi.org/10.1093/pubmed/fdn102.
Anderson, G. B., and M. L. Bell, 2011: Heat waves in the United States: Mortality risk during heat waves and effect modification by heat wave characteristics in 43 US communities. Environ. Health Perspect., 119, 210–218, https://doi.org/10.1289/ehp.1002313.
Ban, J., W. Shi, L. Cui, X. Liu, C. Jiang, L. Han, R. Wang, and T. Li, 2019: Health-risk perception and its mediating effect on protective behavioral adaptation to heat waves. Environ. Res., 172, 27–33, https://doi.org/10.1016/j.envres.2019.01.006.
Banwell, C., J. Dixon, H. Bambrick, F. Edwards, and T. Kjellström, 2012: Socio-cultural reflections on heat in Australia with implications for health and climate change adaptation. Global Health Action, 5, 19277, https://doi.org/10.3402/gha.v5i0.19277.
Bassil, K. L., and D. C. Cole, 2010: Effectiveness of public health interventions in reducing morbidity and mortality during heat episodes: A structured review. Int. J. Environ. Res. Public Health, 7, 991–1001, https://doi.org/10.3390/ijerph7030991.
Bean, H., N. Grevstad, A. Meyer, and A. Koutsoukos, 2022: Exploring whether wireless emergency alerts can help impede the spread of Covid‐19. J. Contingencies Crisis Manage., 30, 185–203, https://doi.org/10.1111/1468-5973.12376.
Benmarhnia, T., L. Schwarz, A. Nori-Sarma, and M. L. Bell, 2019: Quantifying the impact of changing the threshold of New York City heat emergency plan in reducing heat-related illnesses. Environ. Res. Lett., 14, 114006, https://doi.org/10.1088/1748-9326/ab402e.
Casanueva, A., and Coauthors, 2019: Overview of existing heat-health warning systems in Europe. Int. J. Environ. Res. Public Health, 16, 2657, https://doi.org/10.3390/ijerph16152657.
Centers for Disease Control and Prevention, 2017: Warning signs and symptoms of heat-related illness. CDC, https://www.cdc.gov/disasters/extremeheat/warning.html.
Centers for Disease Control and Prevention, 2021: Older adults and extreme heat. CDC, https://www.cdc.gov/aging/emergency-preparedness/older-adults-extreme-heat/index.html.
Cheng, J., Z. Xu, H. Bambrick, H. Su, S. Tong, and W. Hu, 2018: Heatwave and elderly mortality: An evaluation of death burden and health costs considering short-term mortality displacement. Environ. Int., 115, 334–342, https://doi.org/10.1016/j.envint.2018.03.041.
Conti, A., M. Valente, M. Paganini, M. Farsoni, L. Ragazzoni, and F. Barone-Adesi, 2022: Knowledge gaps and research priorities on the health effects of heatwaves: A systematic review of reviews. Int. J. Environ. Res. Public Health, 19, 5887, https://doi.org/10.3390/ijerph19105887.
Di Liberto, T., 2021: Astounding heat obliterates all-time records across the Pacific northwest and western Canada in June 2021. NOAA, https://www.climate.gov/news-features/event-tracker/astounding-heat-obliterates-all-time-records-across-pacific-northwest.
Dinesh, S., and M. Odabas, 2023: 10 facts about Americans and Twitter as it rebrands to X. Pew Research Center, https://www.pewresearch.org/fact-tank/2022/05/05/10-facts-about-americans-and-twitter/.
Dorfman, L., and L. Wallack, 2012: Putting policy into health communication: The role of media advocacy. Public Communication Campaigns, R. E. Rice and C. K. Atkin, Eds., Vol. 3, SAGE Publications, 337–348.
Floyd, D. L., S. Prentice‐Dunn, and R. W. Rogers, 2000: A meta‐analysis of research on protection motivation theory. J. Appl. Soc. Psychol., 30, 407–429, https://doi.org/10.1111/j.1559-1816.2000.tb02323.x.
Frisby, B. N., D. D. Sellnow, D. R. Lane, S. R. Veil, and T. L. Sellnow, 2013: Instruction in crisis situations: Targeting learning preferences and self-efficacy. Risk Manage., 15, 250–271, https://doi.org/10.1057/rm.2013.7.
Frisby, B. N., S. R. Veil, and T. L. Sellnow, 2014: Instructional messages during health-related crises: Essential content for self-protection. Health Commun., 29, 347–354, https://doi.org/10.1080/10410236.2012.755604.
Greene, K., and L. S. Brinn, 2003: Messages influencing college women’s tanning bed use: Statistical versus narrative evidence format and a self-assessment to increase perceived susceptibility. J. Health Commun., 8, 443–461, https://doi.org/10.1080/713852118.
Grundstein, A. J., and C. A. Williams, 2018: Heat exposure and the general public: Health impacts, risk communication, and mitigation measures. Human Health and Physical Activity During Heat Exposure, Y. Hosokawa, Ed., Springer, 29–43.
Grunig, J. E., 1989: Publics, audiences and market segments: Segmentation principles for campaigns. Information Campaigns: Balancing Social Values and Social Change, C. T. Salmon, Ed., SAGE Publications, 199–228.
Guido, G., M. Pichierri, C. Rizzo, V. Chieffi, and G. Moschis, 2021: Information processing by elderly consumers: A five-decade review. J. Serv. Mark., 35, 14–28, https://doi.org/10.1108/JSM-09-2019-0368.
Hass, A. L., and K. N. Ellis, 2019: Motivation for heat adaption: How perception and exposure affect individual behaviors during hot weather in Knoxville, Tennessee. Atmosphere, 10, 591, https://doi.org/10.3390/atmos10100591.
Hawkins, M. D., V. Brown, and J. Ferrell, 2017: Assessment of NOAA National Weather Service methods to warn for extreme heat events. Wea. Climate Soc., 9, 5–13, https://doi.org/10.1175/WCAS-D-15-0037.1.
Hayden, M. H., and Coauthors, 2017: Adaptive capacity to extreme heat: Results from a household survey in Houston, Texas. Wea. Climate Soc., 9, 787–799, https://doi.org/10.1175/WCAS-D-16-0125.1.
Hondula, D. M., S. Meltzer, R. C. Balling Jr., and P. Iñiguez, 2022: Spatial analysis of United States National Weather Service excessive heat warnings and heat advisories. Bull. Amer. Meteor. Soc., 103, E2017–E2031, https://doi.org/10.1175/BAMS-D-21-0069.1.
Kalkstein, A. J., and S. C. Sheridan, 2007: The social impacts of the heat–health watch/warning system in Phoenix, Arizona: Assessing the perceived risk and response of the public. Int. J. Biometeor., 52, 43–55, https://doi.org/10.1007/s00484-006-0073-4.
Kemen, J., S. Schäffer-Gemein, J. Grünewald, and T. Kistemann, 2021: Heat perception and coping strategies: A structured interview-based study of elderly people in Cologne, Germany. Int. J. Environ. Res. Public Health, 18, 7495, https://doi.org/10.3390/ijerph18147495.
Klinenberg, E., 2015: Heat Wave: A Social Autopsy of Disaster in Chicago. University of Chicago Press, 320 pp.
Kotharkar, R., and A. Ghosh, 2022: Progress in extreme heat management and warning systems: A systematic review of heat-health action plans (1995–2020). Sustainable Cities Soc., 76, 103487, https://doi.org/10.1016/j.scs.2021.103487.
Kuligowski, E. D., N. A. Waugh, J. Sutton, and T. J. Cova, 2023: Ember alerts: Assessing wireless emergency alert messages in wildfires using the warning response model. Nat. Hazards Rev., 24, 04023009, https://doi.org/10.1061/NHREFO.NHENG-1724.
Lambrecht, K., B. J. Hatchett, K. VanderMolen, and B. Feldkircher, 2021: Identifying community values related to heat: Recommendations for forecast and health risk communication. Geosci. Commun., 4, 517–525, https://doi.org/10.5194/gc-4-517-2021.
Lane, K., and Coauthors, 2014: Extreme heat awareness and protective behaviors in New York City. J. Urban Health, 91, 403–414, https://doi.org/10.1007/s11524-013-9850-7.
Lehman, S., 2020: Drinking more water during hot weather. Verywell Fit, https://www.verywellfit.com/drink-more-water-during-hot-weather-2506918.
Li, Y., and P. D. Howe, 2023: Universal or targeted approaches? An experiment about heat risk messaging. Nat. Hazards, 117, 381–398, https://doi.org/10.1007/s11069-023-05864-8.
Li, Y., A. L. Hughes, and P. D. Howe, 2018: Communicating crisis with persuasion: Examining official Twitter messages on heat hazards. Proc. 15th ISCRAM Conf., Rochester, NY, Information Systems for Crisis Response and Management, https://idl.iscram.org/files/yajieli/2018/2124_YajieLi_etal2018.pdf.
Lindell, M. K., and R. W. Perry, 2012: The protective action decision model: Theoretical modifications and additional evidence. Risk Anal., 32, 616–632, https://doi.org/10.1111/j.1539-6924.2011.01647.x.
Lowe, D., K. L. Ebi, and B. Forsberg, 2011: Heatwave early warning systems and adaptation advice to reduce human health consequences of heatwaves. Int. J. Environ. Res. Public Health, 8, 4623–4648, https://doi.org/10.3390/ijerph8124623.
Madrigano, J., K. Lane, N. Petrovic, M. Ahmed, M. Blum, and T. Matte, 2018: Awareness, risk perception, and protective behaviors for extreme heat and climate change in New York City. Int. J. Environ. Res. Public Health, 15, 1433, https://doi.org/10.3390/ijerph15071433.
Malmquist, A., M. Hjerpe, E. Glaas, H. Karlsson, and T. Lassi, 2022: Elderly people’s perceptions of heat stress and adaptation to heat: An interview study. Int. J. Environ. Res. Public Health, 19, 3775, https://doi.org/10.3390/ijerph19073775.
Matzarakis, A., G. Laschewski, and S. Muthers, 2020: The heat health warning system in Germany—Application and warnings for 2005 to 2019. Atmosphere, 11, 170, https://doi.org/10.3390/atmos11020170.
Mayrhuber, E. A.-S., and Coauthors, 2018: Vulnerability to heatwaves and implications for public health interventions: A scoping review. Environ. Res., 166, 42–54, https://doi.org/10.1016/j.envres.2018.05.021.
McLoughlin, N., C. Howarth, and G. Shreedhar, 2023: Changing behavioral responses to heat risk in a warming world: How can communication approaches be improved? Wiley Interdiscip. Rev.: Climate Change, 14, e819, https://doi.org/10.1002/wcc.819.
Mileti, D. S., 2018: PrepTalks: Dr. Dennis Mileti “Modernizing public warning messaging.” U.S. Federal Emergency Management Agency, YouTube, https://www.youtube.com/watch?v=oYya009bc2M.
Mileti, D. S., and J. H. Sorensen, 1990: Communication of emergency public warnings: A Social science perspective and state-of-the-art assessment. Oak Ridge National Laboratories, 160 pp., https://doi.org/10.2172/6137387.
Mileti, D. S., and L. Peek, 2000: The social psychology of public response to warnings of a nuclear power plant accident. J. Hazard. Mater., 75, 181–194, https://doi.org/10.1016/S0304-3894(00)00179-5.
Milne, S., P. Sheeran, and S. Orbell, 2000: Prediction and intervention in health‐related behavior: A meta‐analytic review of protection motivation theory. J. Appl. Soc. Psychol., 30, 106–143, https://doi.org/10.1111/j.1559-1816.2000.tb02308.x.
Morss, R. E., C. L. Cuite, J. L. Demuth, W. K. Hallman, and R. L. Shwom, 2018: Is storm surge scary? The influence of hazard, impact, and fear-based messages and individual differences on responses to hurricane risks in the USA. Int. J. Disaster Risk Reduct., 30, 44–58, https://doi.org/10.1016/j.ijdrr.2018.01.023.
Murray-Johnson, L., and K. Witte, 2011: Looking toward the future: Health message design strategies. The Routledge Handbook of Health Communication, T. L. Thompson et al., Eds., Lawrence Erlbaum Associates, Inc., 473–495.
National Weather Service, 2021: Weather related fatality and injury statistics. NOAA, accessed 22 June 2022, https://www.weather.gov/hazstat/.
National Weather Service, 2022: What is the heat index? NOAA, accessed 20 July 2023, https://www.weather.gov/ama/heatindex.
Olson, M. K., J. Sutton, S. C. Vos, R. Prestley, S. L. Renshaw, and C. T. Butts, 2019: Build community before the storm: The National Weather Service’s social media engagement. J. Contingencies Crisis Manage., 27, 359–373, https://doi.org/10.1111/1468-5973.12267.
Oxman, A. D., and Coauthors, 2022: Health communication in and out of public health emergencies: To persuade or to inform? Health Res. Policy Syst., 20, 28, https://doi.org/10.1186/s12961-022-00828-z.
Potter, S. H., P. V. Kreft, P. Milojev, C. Noble, B. Montz, A. Dhellemmes, R. J. Woods, and S. Gauden-Ing, 2018: The influence of impact-based severe weather warnings on risk perceptions and intended protective actions. Int. J. Disaster Risk Reduct., 30, 34–43, https://doi.org/10.1016/j.ijdrr.2018.03.031.
Reynolds, B., and M. W. Seeger, 2005: Crisis and emergency risk communication as an integrative model. J. Health Commun., 10, 43–55, https://doi.org/10.1080/10810730590904571.
Ripberger, J. T., C. L. Silva, H. C. Jenkins-Smith, and M. James, 2015: The influence of consequence-based messages on public responses to tornado warnings. Bull. Amer. Meteor. Soc., 96, 577–590, https://doi.org/10.1175/BAMS-D-13-00213.1.
Rossi, M. G., and F. Macagno, 2021: The communicative functions of metaphors between explanation and persuasion. Inquiries in Philosophical Pragmatics, F. Macagno and A. Capone, Eds., Springer, 171–191.
Seethaler, S., J. H. Evans, C. Gere, and R. M. Rajagopalan, 2019: Science, values, and science communication: Competencies for pushing beyond the deficit model. Sci. Commun., 41, 378–388, https://doi.org/10.1177/1075547019847484.
Seneviratne, S. I., and Coauthors, 2021: Weather and climate extreme events in a changing climate. Climate Change 2021: The Physical Science Basis, V. Masson-Delmotte et al., Eds., Cambridge University Press, 1513–1766.
Sheridan, S. C., 2007: A survey of public perception and response to heat warnings across four North American cities: An evaluation of municipal effectiveness. Int. J. Biometeor., 52, 3–15, https://doi.org/10.1007/s00484-006-0052-9.
Silk, K. J., C. K. Atkin, and C. T. Salmon, 2011: Developing effective media campaigns for health promotion. The Routledge Handbook of Health Communication. T. L. Thompson, R. Parrott, and J. F. Nussbaum, Eds., Routledge, 230–246.
Silver, A., and J. Andrey, 2019: Public attention to extreme weather as reflected by social media activity. J. Contingencies Crisis Manage., 27, 346–358, https://doi.org/10.1111/1468-5973.12265.
Simis, M. J., H. Madden, M. A. Cacciatore, and S. K. Yeo, 2016: The lure of rationality: Why does the deficit model persist in science communication? Public Understanding Sci., 25, 400–414, https://doi.org/10.1177/0963662516629749.
Sivle, A. D., and T. Aamodt, 2019: A dialogue‐based weather forecast: Adapting language to end‐users to improve communication. Weather, 74, 436–441, https://doi.org/10.1002/wea.3439.
Snyder, L. B., 2007: Health communication campaigns and their impact on behavior. J. Nutr. Educ. Behav., 39, S32–S40, https://doi.org/10.1016/j.jneb.2006.09.004.
Steadman, R. G., 1984: A universal scale of apparent temperature. J. Climate Appl. Meteor., 23, 1674–1687, https://doi.org/10.1175/1520-0450(1984)023<1674:AUSOAT>2.0.CO;2.
Sutton, J., and E. D. Kuligowski, 2019: Alerts and warnings on short messaging channels: Guidance from an expert panel process. Nat. Hazards Rev., 20, 04019002, https://doi.org/10.1061/(ASCE)NH.1527-6996.0000324.
Sutton, J., L. Fischer, and M. M. Wood, 2021: Tornado warning guidance and graphics: Implications of the inclusion of protective action information on perceptions and efficacy. Wea. Climate Soc., 13, 1003–1014, https://doi.org/10.1175/WCAS-D-21-0097.1.
Trench, B., 2006: Science communication and citizen science: How dead is the deficit model? Ninth Int. Conf. on Public Communication of Science and Technology (PCST-9), Seoul, South Korea, PCST, WB1-07, https://www.pcst.network/wp-content/uploads/2022/05/PCST-2006-Conference-programme.pdf.
Twitter, 2022: How to create a thread on Twitter. X Help Center, https://help.twitter.com/en/using-twitter/create-a-thread.
Valois, P., D. Talbot, D. Bouchard, J.-S. Renaud, M. Caron, M. Canuel, and N. Arrambourg, 2020: Using the theory of planned behavior to identify key beliefs underlying heat adaptation behaviors in elderly populations. Popul. Environ., 41, 480–506, https://doi.org/10.1007/s11111-020-00347-5.
Viswanath, K., R. F. McCloud, and M. A. Bekalu, 2021: Communication, health, and equity: Structural influences. The Routledge Handbook of Health Communication, T. L. Thompson, R. Parrott, and J. F. Nussbaum Eds., Routledge, 426–440.
Wänke, M., and L. Reutner, 2010: Pragmatic persuasion or the persuasion paradox. The Psychology of Attitudes and Attitude Change, J. P. Forgas, J. Cooper, and W. D. Crano, Eds., Routledge, 183–198.
Washington State Department of Health, 2021: Heat wave 2021. Accessed 21 June 2022, https://doh.wa.gov/emergencies/be-prepared-be-safe/severe-weather-and-natural-disasters/hot-weather-safety/heat-wave-2021.
Watson, H., and R. L. Finn, 2014: Social media and the 2013 UK heat wave: Opportunities and challenges for future events. 11th Int. Conf. on Information Systems for Crisis Response and Management, University Park, PA, ISCRAM, https://idl.iscram.org/files/watson/2014/1069_Watson+Finn2014.pdf.
Williams, C. A., and A. J. Grundstein, 2018: Children forgotten in hot cars: A mental models approach for improving public health messaging. Inj. Prev., 24, 279–287, https://doi.org/10.1136/injuryprev-2016-042261.
Williams, S., and Coauthors, 2022: Evaluating cost benefits from a heat health warning system in Adelaide, South Australia. Aust. N. Z. J. Public Health, 46, 149–154, https://doi.org/10.1111/1753-6405.13194.
Witte, K., 1992: Putting the fear back into fear appeals: The extended parallel process model. Commun. Monogr., 59, 329–349, https://doi.org/10.1080/03637759209376276.
Wogalter, M. S., V. C. Conzola, and T. L. Smith-Jackson, 2002: Research-based guidelines for warning design and evaluation. Appl. Ergon., 33, 219–230, https://doi.org/10.1016/S0003-6870(02)00009-1.
Wood, M. M., D. S. Mileti, H. Bean, B. F. Liu, J. Sutton, and S. Madden, 2018: Milling and public warnings. Environ. Behav., 50, 535–566, https://doi.org/10.1177/0013916517709561.
Wu, Y., X. Wang, J. Wu, R. Wang, and S. Yang, 2020: Performance of heat-health warning systems in Shanghai evaluated by using local heat-related illness data. Sci. Total Environ., 715, 136883, https://doi.org/10.1016/j.scitotenv.2020.136883.
Zhao, X., 2020: Health communication campaigns: A brief introduction and call for dialogue. Int. J. Nurs. Sci., 7, S11–S15, https://doi.org/10.1016/j.ijnss.2020.04.009.