Hurricane Ian, a Category 4 hurricane that made landfall in Florida in 2022, presented a unique challenge for residents due to the ongoing COVID-19 pandemic. This study examines how the evolving COVID-19 pandemic affected perceptions of hurricane risks and evacuation decisions, particularly the role of COVID-19 concerns in influencing the decision to stay home. The study found that only 4% of respondents cited COVID-19 as the primary reason for staying home, while 50.1% of participants indicated that having a pet played a significant role in their decision. Most survey participants (70.3%) strongly agreed they were provided with timely information to make a good evacuation decision. The most heavily relied upon sources of information included local media, county/city emergency preparedness websites, and electronic media whereas the least relied upon sources of information were social media and print media. The findings of this study suggest that, even in the context of a pandemic, many residents evacuated for Hurricane Ian. However, COVID-19 concerns did play a role in some individuals' decisions, particularly those who were concerned about the crowded conditions in shelters. This study highlights the importance of providing clear, consistent, and timely information to residents during evacuations, particularly during a public health crisis.
]]>Hurricanes claimed the lives of 6,593 individuals in the United States from 1980 to 2020 (NOAA 2021). Targeted evacuations of populations most at risk ahead of these storms is the best method for preventing future deaths. However, the emergence of the pandemic in early 2020 added a potential barrier to evacuation, with households needing to weigh the risks of hurricanes against Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1. Hurricane Ian challenged forecasters, emergency managers, and the public as forecast models for the storm skewed west of the actual track (Bucci et al. 2023). Forecast graphics released 48 hours and 24 hours before landfall have St. Petersburg, Florida, at the center of the tropical cyclone track forecast cone, with the actual site of landfall at Cayo Costa, Florida, located at the southern edge of the cone (see Figure 2). Unfortunately, people tend to misinterpret the forecast cone, placing too much emphasis on the center of the cone (Boone et al. 2018;Broad et al. 2007).
Coastal communities in Lee County experienced the most devastating effects in Florida, with the highest number of fatalities (Bucci et al. 2023;Robles et al. 2022). As a result, Lee County officials were scrutinized for issuing mandatory evacuation orders approximately 32 hours before landfall, providing residents and visitors with limited time to formulate a plan and evacuate (Olivo et al. 2022;Robles et al. 2022).
The decision to evacuate or remain in place is based upon an evaluation of various factors. Senkbeil et al. (2019) and Morss et al. (2016) have proposed that this decision-making process is intricate because people receive diverse information through various forms of communication, Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1.
Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC which interact with their physical and social surroundings. Numerous elements influence these evacuation decisions, including the storm's severity, the projected landfall location, official warnings, residential construction methods, flood vulnerability, expected consequences, housing type, social influences, individual assessments of injury risk, and past experiences hurricane evacuation experiences (Baker 1991;Demuth et al. 2012;Dow & Cutter 2000;Huang et al. 2016Huang et al. , 1995;;Gladwin et al. 2001;Goldberg et al. 2020;Sherman-Morris et al. 2020;Whitehead et al. 2000).When local authorities issue evacuation orders, complying with these orders and evacuating is a protective measure designed to safeguard individuals from hurricane-related hazards, particularly flooding and storm surge. Nevertheless, many people resist evacuation orders because they believe the downsides of evacuating outweigh the perceived threat. For example, evacuations can be expensive, stressful, inconvenient, and frustrating. Previous evacuation decision-making research by Goldberg et al. (2020) indicated that individuals with prior hurricane experience are likely to make similar decisions in the future. For instance, if an individual has stayed at home during previous hurricanes, they are likely to shelter in place in the future. People who have evacuated in the past are more likely to evacuate. This prior research offers valuable insights into household planning even before the pandemic. Typical responses to hurricanes, such as mass evacuations and opening shelters for evacuees, went directly against newly formed policies and mitigation measures meant to stop the spread of a virus, such as extra precautions in terms of hygiene, social distancing, and isolation (Whytlaw et al. 2021). Introducing health concerns related to a pandemic into the already complex landscape of hurricane evacuation decision-making adds additional challenges and planning requirements. For instance, a study by Botzen et al. (2022) examined individuals residing in flood-prone regions of Florida and found that COVID-19 concerns often took precedence over perceptions of flood risk, ultimately making people threatened by flooding less likely to evacuate.
This research is strengthened by incorporating established disaster response theories. Three prominent frameworks inform this analysis: the Protective Action Decision Model (PADM) (Lindell 2018), the Warning Response Model (WRM) (Mileti & Sorensen, 1990), and the Time-Phased Model of Warning and Response (Sorensen et al. 2018). These models offer valuable insights into individual decision-making and behavioral patterns during emergencies.
Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1.
Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC By incorporating established disaster response theories like the PADM, WRM, and Time-Phased Model, this review provides a robust theoretical foundation for analyzing the evacuation decisions made during Hurricane Ian. These frameworks highlight the multifaceted nature of evacuation decisions, influenced by individual characteristics, social cues, warning messages, and the evolving disaster context. Understanding these factors through the lens of these models can inform future research on evacuation behavior and guide the development of more effective communication strategies and preparedness measures to enhance community resilience during hurricanes.
Prior research has improved the understanding of how individuals assess and respond to hurricane risks, but the COVID-19 pandemic has introduced uncertainty regarding changes in people's threat calculations and evacuation decisions. Previous research on diseases has suggested potential outcomes. Bissell (1983) observed an association between hurricanes and higher rates of endemic disease infection due to the congregation of large numbers of people in shelters, facilitating disease transmission through respiratory droplets. Shultz et al. (2005) conducted a comprehensive analysis of how tropical cyclones can spread diseases. They identified six key factors contributing to infectious disease outbreaks, including how the risk of exposure and infection rose as the population density increased, especially in crowded shelters. Additionally, Ivers & Ryan (2006) reported that disease transmission increased in the aftermath of a disaster as displaced populations gathered in shelters. Lemonick (2011) investigated past disasters and their associations with disease transmission, identifying crowding as an amplifying factor within shelter environments.
Travers (2020) categorized Americans into five distinct "types" based on their risk perception of COVID-19 and trust in the government's response to the pandemic: (1) "Soundingall-alarms," (2) "At risk," (3) "Under-control," (4) "Open-ups," and ( 5) "Dismissers." The largest group, "Sounding-all-alarms," constituting 29% of respondents, expressed concern about the federal response and the potential for worsening conditions. Groups 4 and 5, representing 14%, believed there was an exaggerated response to the threat.
Numerous studies have provided insights into the evolving dynamics of hurricane evacuations during the COVID-19 pandemic. Collins et al. (2021b) surveyed Florida residents, Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1. Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC
revealing that nearly half perceived vulnerability to COVID-19 due to pre-existing health conditions. Additionally, approximately three-quarters expressed greater concern about risks associated with evacuation shelters during the pandemic-surpassing their concerns about hurricanes -emphasizing the pandemic's influential role in evacuation considerations. Similarly, Botzen et al. (2022) surveyed coastal residents in Florida, noting a significant shift with a higher proportion perceiving COVID-19 risks as more alarming than flood risks during hurricane threats, a departure from the predominant concern of flooding in pre-pandemic surveys. The study also found that older individuals were less likely to evacuate under a voluntary order, driven by concerns about COVID-19 infection. Collins et al. (2022b) extended their research to residents of Puerto Rico and the US Virgin Islands at the start of the 2020 hurricane season. Over half of the respondents felt vulnerable to COVID-19, and a substantial portion of those who had previously evacuated to shelters expressed hesitancy due to the perceived infection risk. Zhao et al. (2023) investigated the perspectives of 323 elderly Floridians regarding hurricane evacuations before the 2020 hurricane season, revealing that older adults were willing to evacuate to a shelter during a Category 1 hurricane if social distancing measures were guaranteed. Their findings indicated that COVID-19 risk impacted decisions to evacuate from Category 1 hurricanes but had no discernible influence on evacuation from stronger hurricanes (Zhao et al. 2023).
In contrast, Greer et al. (2023) reported that the perceived risk of COVID-19 did not significantly influence survey respondents' evacuation decisions during Hurricane Laura in 2020.
However, it did affect their expectations regarding the personal impacts of the storm, unveiling a nuanced relationship between pandemic concerns and hurricane evacuation choices. The study also revealed that those with previous hurricane experiences were more likely to evacuate. Wu et al. (2023) delve deeper into evacuation behavior during "dual hazards" (concurrent hurricane and pandemic). They reveal a complex decision-making process driven by perceived threat consequences. Optimistic biases towards both hazards indirectly discourage evacuation, while heightened vulnerability perceptions (measured as personal consequence perceptions) nudge individuals towards escape. Interestingly, hurricane-related vulnerability exerts a stronger influence than COVID-19 vulnerability, emphasizing that hurricane risk remains the dominant factor in evacuation decisions despite pandemic concerns.
Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1. Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC Yusuf et al. (2023) focus on household evacuation behavior during "compound threats" (hurricane and pandemic). They uncover a nuanced decision-making process influenced by vulnerability to both hazards (measured objectively and subjectively). Greater perceived or actual exposure increases evacuation intent, but financial resources significantly moderate this relationship. Households with ample resources readily prioritize evacuation, while those facing financial constraints, particularly racial minorities, often grapple with the potential burden of both hurricane and pandemic, potentially opting to stay put.
Similarly, Diaz et al. (2023) found that vulnerable households, particularly those of nonwhite ethnicity, were more likely to stay put during hurricanes during the COVID-19 pandemic.
Their decisions were heavily influenced by factors like race, risk perception of the hurricane, and fear of contracting COVID-19 in shelters. This study underlines the need for targeted evacuation planning and resource allocation to ensure vulnerable communities have safe and accessible options during hurricane threats. Our initial research phase consisted of a pilot study conducted in June 2020 at the beginning of the hurricane season, during the first year of the COVID-19 pandemic. It consisted of a 40question online survey in both English and Spanish that was distributed via media outlets, regional planning councils, and emergency management agencies. Impressively, responses from 7,102 individuals residing in more than fifty counties were collected in just over two weeks (Collins et al. 2021b). Half of these respondents identified as at risk for COVID-19 due to their health conditions. Additionally, 74.3% of respondents perceived seeking shelter during the COVID-19 pandemic as riskier than facing a hurricane in their homes, leading to many individuals expressing their intention to avoid public shelters during COVID-19. Consequently, there was a 7.7% decrease in those who indicated they would definitely or probably use a shelter.
In 2021, a follow-up survey was disseminated once vaccines were more readily available, targeting coastal residents across the United States-a comparative study to the Florida research.
This study included additional questions about respondents' vaccination status and its impact on hurricane evacuation decisions. Findings revealed that 71.3% of individuals believed their vaccination status would not affect evacuation decision-making (Collins et al. 2023).
Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1.
Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC Data collected during Hurricanes Laura and Sally (Collins et al. 2022a) proved valuable for the current research on Hurricane Ian. The research provides insight into actual hurricane evacuation behaviors for these three events, facilitating the opportunity for comparative analysis.
Key findings from the Hurricane Laura and Sally study showed that 76.7% of individuals chose to stay at home, with 80% citing concerns about COVID-19 as a contributing factor in their decision.
However, among those who evacuated to destinations other than shelters or hotels (constituting 15.8% of the overall sample), the majority (85%) indicated that COVID-19 was not the primary motivation for their evacuation choice. Regarding perceived vulnerability to COVID-19, 39.4% of those affected by Hurricanes Laura or Sally considered themselves vulnerable due to pre-existing health conditions, and two-thirds of the sample expressed that contracting the virus would be a "very serious" concern for them or their loved ones. In terms of perceptions of public shelters, negative sentiments similar to those observed in the pre-evacuation research were evident, with a notable decline in the number of people willing to use public shelters during the pandemic compared to before (Collins et al. 2022a).
In 2021, Hurricane Ida provided another opportunity to gather data on evacuation behaviors. This Category 4 hurricane made landfall in Port Fourchon, Louisiana, on August 29, 2021. Key findings from the study indicated that 44.5% of residents chose to stay at home. Of those who evacuated somewhere other than a hotel or shelter, 80.5% stated that the COVID-19 pandemic did not affect their decision-making (Polen et al. 2023).
As Hurricane Ian approached and local authorities issued evacuation orders, residents grappled with the delicate balance between the need to evacuate and the potential risks of COVID-
Florida residents in the path of Hurricane Ian were administered an anonymous survey, available in both English and Spanish, via Qualtrics from October 9, 2022, to April 10, 2023. This study received approval from the University of South Florida Institutional Review Board (STUDY 001517), and standard consent and data management practices were followed to safeguard the confidentiality and privacy of participants.
The Hurricane Ian Evacuation Decision Making in the Age of COVID-19 survey, employed by Collins et al. (2023a) to investigate evacuation behavior during Hurricane Ian, is provided in the appendix. The design of the survey instrument draws heavily upon the Protective Action Decision Model (PADM) (Lindell 2018). This model examines the factors influencing decision-making processes leading to a potential disaster. It outlines a multi-stage progression, detailing how individuals process information and cues from their environment and media communications (e.g., news reports and social media). Individuals actively seek information from diverse sources (e.g., government agencies, family members, friends) and evaluate the potential threat, ultimately leading to an informed evacuation decision. The survey instrument was initially crafted during the 2020 hurricane season in consultation with government officials, emergency managers, and planners to confirm that the questions aligned with their planning and response efforts. The questions were consistently worded so that comparisons could be made among the hurricanes studied during the pandemic. The instrument was developed through an iterative process, including literature review, expert consultation, and pilot testing with a small sample to refine the questions and ensure clarity. Both the English and Spanish survey instruments were pilot tested for Hurricanes Laura, Sally, and Ida, and a modified version was piloted in person post-Hurricane Ian to ensure question clarity and enhance the survey instruments' reliability.
Respondents were asked to select their evacuation decision from one of four primary options: (1) evacuating to a shelter, (2) evacuating to a location other than a hotel (but not a shelter), (3) evacuating to a hotel, or (4) staying at home. Individuals who initially evacuated to a shelter and were subsequently moved to a hotel were presented with additional questions in the hotel experience section (Figure 4). This research employed a mixed methods design to comprehensively investigate evacuation decisions made during Hurricane Ian. The survey instrument comprised 112 questions, with the length contingent upon participants' specific evacuation experiences. Individuals who evacuated to non-traditional shelters (e.g., residences of friends or family) completed a shorter survey, while those who initially evacuated to a designated shelter followed by relocation to a hotel received the most extensive questionnaire. The average completion time for the survey was 19.97 minutes. The survey incorporated a combination of quantitative and qualitative data collection methods. Quantitative data was gathered on participants' perceptions of hurricane risks, prior evacuation experiences during the COVID-19 pandemic, health status, and associated risk factors. Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1. Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC
To gauge participants' opinions on various topics, we used a 5-point Likert scale. For instance, after showing them a hurricane advisory from the National Weather Service, we asked them to rate their agreement with statements using a scale from "strongly agree" to "strongly disagree."
Additionally, the survey explored perspectives on shelters and how individuals weighed the risks associated with the hurricane against the potential for contracting COVID-19. True/false and yes/no questions were employed to gather factual information and gauge participants' binary preferences on specific issues. To gain deeper insights into these topics, ten open-ended questions were strategically integrated into the survey instrument, allowing respondents to elaborate on their experiences and motivations beyond pre-defined answer choices (Singer & Couper 2017). This mixed methods approach facilitated a more nuanced understanding of the multifaceted factors influencing evacuation decisions during Hurricane Ian.
A combination of convenience and snowball sampling techniques was employed for this study as convenience sampling focuses on a specific location, allowing for the identification of participants beyond the initial contacts in disaster-stricken areas, given the difficulties in reaching individuals in such settings. To enlist participants for this study, the research team capitalized on a diverse and specialized professional network established during previous projects (Collins et al. 2021b) to facilitate the distribution of surveys within the local community. The researchers' existing social connections primarily disseminated the survey instrument via direct email correspondence and social media platforms overseen by a network of contacts. This distribution network, which broadened the survey's reach to various communities in the affected areas, including government officials, specific social media groups dedicated to Hurricane Ian, and various news media outlets. Community partners also shared promotional materials containing survey links with their residents and connections (see Figure 5). A consent form was presented at the start of the digital survey that outlined the minimal risk for participants in the study, meaning the probability of harm is not greater than that encountered in everyday life (National Institute of Mental Health 2023). Participants were then asked to confirm if they were 18 years or older and if they were affected by Hurricane Ian. If participants answered "no" to either question, the survey concluded as the participant did not meet the eligibility criteria for inclusion in the study. d. Data Analyses Following data collection through Qualtrics, the quantitative data was exported to IBM SPSS Version 29 for statistical analysis. The data cleaning process involved transferring the information to Microsoft Excel for initial organization and categorization. Here, questions with brief, descriptive responses (e.g., descriptions of hazards near residences) were grouped thematically to facilitate analysis. Given the primarily categorical or ordinal nature of the survey data, non-parametric statistical methods were predominantly employed. These methods included Chi-Square tests to assess associations between categorical variables and McNemar's tests for paired nominal data. Additionally, straightforward frequency distributions were generated to provide an overview of the response patterns. Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1. Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC
For the open-ended questions, a thematic analysis approach was utilized. This involved coding the responses in Excel to identify recurring themes and patterns. Representative quotes were selected to illustrate these key themes in the results section.
In total, 2,195 unique participants completed at least 75% of the survey, with a majority of respondents (99.7%) completing the survey in English. The respondents in the study are not representative of the broader population of Florida, as reported by the U.S. Census Bureau in 2022.The sample had a higher proportion of women (69.2%, versus the Florida average of 50.8%) and lower racial diversity, with 89.4% identifying as White compared to the Florida average of 76.8%, indicating underrepresentation from minority ethnic groups. Additionally, the sample reported higher annual incomes, with nearly half of the households (43.4%) exceeding an annual income of $90,000 a year, while the average income in Florida was $61,777. Furthermore, the sample exhibited higher rates of educational attainment. Nearly all respondents (99%) had completed high school, compared to the statewide rate of 89% (United States Census Bureau [USCB] 2022). Moreover, 77.0% of the sample held college degrees, in contrast to the statewide rate of 31.5%. Notably, 24.4% of the respondents claimed to possess a master's degree, significantly surpassing the state average of 14.4% for individuals with advanced degrees, including professional and Doctorate degrees. Most of the sample was employed fulltime (56.2%), with a significant proportion (27.7%) identifying as retired (USCB 2022). Given these demographic disparities, the findings of this research may not be readily generalizable to the broader population in the region. For a more comprehensive overview of the surveyed population, please refer to Table 1, which provides additional demographic characteristics. Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1. Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1. Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1. Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC Additionally, twelve individuals (0.5%) required rescue operations during or in the aftermath of the storm. One North Port homeowner who had to be rescued shared, "since all surrounding streets were flooded, we got stuck in the water trying to go to a family member's home AFTER the storm. The North Port Police had to rescue us from our car."
Survey participants were requested to assess their level of concern regarding specific hurricane-related hazards using a Likert-type scale that ranged from "not concerned" (assigned a value of zero) to "extremely concerned" (designated with a value of four). The analysis of responses revealed that respondents exhibited the highest degree of concern for the intensity of the about the following hazards at your residence." Note. "N," "M," and "SD" are used to denote the number of survey responses and standard deviation. The mean score is derived from a scale of 0 to 4, where 0 represents "not concerned," and 4 indicates "extremely concerned."
2) GENERAL EVACUATION DECISIONS
Regarding their actual evacuation choices during Hurricane Ian (Figure 6), 55.9% of the surveyed individuals opted to remain in their homes, while 2.1% sought refuge in public shelters, sharing space with fellow evacuees. A notable 14.9% chose to stay in hotels or motels, with expenses either covered by themselves or by a third party. Furthermore, 27.1% of respondents evacuated to the residence of a friend or family member, remained in their vehicles, or relocated to another site. In terms of evacuation orders, approximately half of the respondents, specifically 46.9%, confirmed that they were under a mandatory evacuation order for Hurricane Ian, while 50.8% reported not being subject to such an order, and a minority, comprising 2.3%, expressed uncertainty regarding their evacuation status.
Accepted Among the sample, 1,028 individuals reported being subject to mandatory evacuation orders. Among these evacuees, 29.8% opted to stay in their homes, 2.9% sought refuge in public shelters, 24.4% secured accommodations in hotels, and 42.9% evacuated to alternative locations.
In contrast, for those not mandated to evacuate, a majority, accounting for 79.5%, chose to remain at home, while 1.3% found shelter in designated facilities, 6.1% lodged in hotels, and 13.1% relocated elsewhere. As anticipated, individuals within mandatory evacuation zones exhibited a higher proportion of evacuations to alternative sites and a lower proportion of those who stayed in their homes. Notably, a small subset of respondents (3.3%) reported the necessity of seeking secondary shelters due to flooding, power outages, or damage to their initial places of refuge.
This survey investigated the alignment between participants' perceptions of their evacuation zone and actual evacuation directives. We achieved this by comparing the county and evacuation zone they reported with official records of mandatory evacuation orders issued by local authorities. The association between evacuation choice and evacuation zone was statistically significant, X2(1) = 210.889, p < 0.001. Nearly half (47.6%, n=1043) resided in a mandatory evacuation zone. Encouragingly, a majority (67.6%, n=705) of those in mandatory evacuation zones heeded the orders and evacuated. However, a concerning minority (32.4%, n=338) chose not to evacuate despite residing in a mandatory zone. Of this group, a significant portion (18.6%, n=63) were unaware of the evacuation order. This highlights a potential gap in communication or public awareness regarding evacuation directives. A fifth (18.8%, n=188) of those residing outside mandatory evacuation zones still opted to evacuate. This suggests a proactive approach by some residents who may have considered potential risks beyond the officially designated zones.
Mobile home residents exhibited the highest evacuation rate at 77.78%, followed by "other" home types at 60.71%. Block and wood frame homes and wood frame homes, which may be perceived as offering greater structural integrity than mobile homes, had similar evacuation rates of 44.73% and 48.64%, respectively. Concrete block homes exhibited the lowest evacuation rate at 38.33%. This observed trend tentatively suggests that individuals may factor in the perceived sturdiness of their homes when making evacuation decisions during hurricanes. A chi- square test of independence confirmed that the association between housing type and evacuation status was statistically significant, X²(4) = 75.106, p < 0.001.
No associations were found between evacuation status and gender, ethnicity, marital status, educational attainment, or income. The survey data revealed a statistically significant association (X 2 (6) = 14.358, p = 0.026) between age and evacuation decisions during Hurricane Ian. Notably, the 65-74 age group had the highest evacuation rate, with 50.6% of respondents in this category indicating they evacuated. Household size was also associated with the decision to evacuate (X 2 (8) = 23.555, p = 0.003). People who lived alone were the only household size where evacuees (50.2%) outnumbered those who stayed home. The number of people 65 and older was also associated with evacuation status (X 2 (1) = 6.052, p = 0.014). Households with at least one member 65 and older were more likely to evacuate, with many individuals highlighting prior experiences and media influencing their decisions. One older adult from Largo shared, "After Hurricane Andrew, I decided I would ALWAYS evacuate." A St. Petersburg, Florida retiree indicated, "I always evacuate. I followed meteorologist Denis Williams, and he emphasized the dangers of storm surge, winds, and falling trees. I don't want to be on the interstate when so many people are trying to evacuate, hence the decision to evacuate to better ground and shelter within the county." One older adult from Dunedin, Florida, expanded upon their decision to evacuate by sharing, "we always evacuate at least five days before projected landfall of any hurricane headed toward the west coast of Florida. If we don't leave, we just become part of the problem."
Most respondents (55.2%) had encountered a hurricane evacuation order in their prior experiences. In those previous instances, approximately one-quarter (26.5%) reported their decision to remain at home, while 6.6% sought refuge in public shelters. Furthermore, a significant portion (69.6%) opted for evacuation to alternative destinations, including the residences of friends and family or lodging in hotels or motels.
Concerning hurricane preparedness, 66.2% of the surveyed individuals had formulated a hurricane evacuation plan, and an overwhelming majority (96.3%) had amassed hurricane supplies sufficient for at least three days before Hurricane Ian. Additionally, roughly one-third (31.2%) possessed a generator. Among those who owned a generator, 52.2% indicated that it played a role in determining whether to evacuate or remain at home. Conversely, for respondents without a generator, 17.8% stated that the absence of this equipment influenced their evacuation choice.
Individuals with generators were more inclined to remain at home than those without generators.
Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1. Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC About one-third (35.4%) of generator owners evacuated, while nearly half (48.9%) of respondents who did not own generators evacuated.
Regarding their thoughts about hurricanes, 27.4% of participants stated that the hurricane category did not influence their decision to evacuate. Of those who did take hurricane categories into account when making their evacuation choices, 39.8% were willing to evacuate for storms as low as Category 3. In contrast, 6.7% set their evacuation threshold at Category 1, 15.5% at Category 2, 28.2% at Category 4, and 9.7% at Category 5 (see Appendix Q38).
To assess perceived vulnerability to COVID-19, respondents were asked, "Do you consider yourself vulnerable to COVID-19 due to existing health conditions/concerns?" This association was found to be statistically significant, X²(2) = 5.271, p = 0.072. Interestingly, a higher proportion of those who answered yes (48.48%) chose to evacuate compared to those who did not perceive such vulnerability (43.13%).
To understand this link further, the survey explored the connection between pre-existing health conditions and perceived COVID-19 vulnerability.
Overall, 62.09% of respondents reported having at least one health condition (see Appendix Q88, Q89). Interestingly, only 42.65% of this group expressed feeling vulnerable to COVID-19 due to these conditions. While a chi-square test of independence suggested an association between having a pre-existing health condition and perceived COVID-19 vulnerability, this association was not statistically significant, X²(2) = 4.714, p = 0.095. This suggests that having a pre-existing health condition does not automatically translate to a heightened sense of vulnerability to COVID-19. Finally, the survey data offered an opportunity to explore potential variations in perceived COVID-19 vulnerability based on the timing of the survey completion. A significant portion (64.79%) completed the survey closer to the evacuation decisions (October-December 2022), while the remaining 35.20% completed it later (January-April 2023). Interestingly, a slightly higher percentage of respondents who completed the survey in 2023 reported feeling vulnerable to COVID-19 (30.68%) compared to those in 2022 (25.99%). However, a chi-square test of independence revealed that this difference was not statistically significant, X²(2) = 5.291, p = 0.71. This could be due to several factors, including evolving perceptions of COVID-19 risk due to variants or personal health experiences during that period. It is also worth noting that a slightly larger proportion of respondents in 2023 reported having health conditions (61.51%) compared to Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1. Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC
). However, this difference is small and may not fully explain the observed increase in perceived vulnerability in 2023.
A small portion of survey respondents (14.9%) chose hotel or motel accommodation. The majority (92.4%) self-financed their stay, while 4.9% received financial support from nonhousehold family or friends, and 0.6% had costs covered by non-profit or government organizations. Regarding transportation to the hotel, the vast majority (95.7%) used their vehicles, 1.2% shared transportation with others, and 3.1% opted for air travel. A substantial proportion (59.3%) of participants were pet owners, with the majority (94.3%) successfully arranging for their pets to accompany them to the hotel (see Appendix Q82, Q83 Q84).
More than half of the participants (55.9%) chose to stay in their homes during Hurricane Ian. To further understand this decision, respondents were asked, "Was COVID-19 the primary reason you stayed at home?" In response to this question, 1.3% strongly agreed, 2.8% somewhat agreed, 12.1% responded neutrally, 6.7% somewhat disagreed, and 77.0% strongly disagreed.
Regarding shelter availability within their county, 53.7% indicated they could find refuge with friends or family, and 72.1% could stay with friends or family from outside their home county (see Appendix Q52, Q54, Q55).
When asked, 'Did you feel that you wanted to evacuate but waited until it was too late?', a subset of respondents (7.7%) acknowledged this sentiment, indicating a desire to evacuate but a delay in their decision until it was no longer feasible. One individual from Sarasota shared how "by the time the storm intensified, it was too late to secure our house and evacuate." At the same time, another resident from Belleair Bluffs indicated that they "waited too long to decide." A homeowner from Venice highlighted how "the storm kept changing direction, and by the time we realized, it was going to hit nearby. It was too late to leave, and we would have been stuck on the road and felt home was safer."
Most of those who chose to stay at home during the hurricane (70.0%) were pet owners.
Approximately half of these pet owners (50.1%) acknowledged that their pets played a significant role in their decision to remain at home (see Appendix Q48, 49). One respondent from Riverview mentioned she had "several pets that [she] was concerned about. Hard to take them all, did not want to leave them alone." Many respondents cited traffic in their decision to stay home (41.4%).
One respondent from Collier County shared how "the memory of sitting in horrible traffic for Hurricane Irma on I75 was a nightmare-crawled to Georgia." A small fraction (2.8%) mentioned transportation limitations as a factor, while 14.2% cited responsibilities for caring for someone who could not or would not evacuate as one of the reasons for their decision. One resident from Largo indicated that "elderly extended family failed to evacuate, and that they did not want to contribute to traffic when those in evac zones needed to go due to the projected storm surge."
Approximately one-fourth (27.1%) of the sample evacuated their residences but opted not to utilize shelters or hotels, instead choosing to stay with friends, coworkers, or family members, remain in their vehicles, or utilize vacation rental accommodations. Most individuals (94.8%) reached their chosen evacuation destination using their personal vehicles (see Appendix Q64). In response to the question, 'Was COVID-19 the primary reason you chose to evacuate somewhere other than a shelter?', most respondents (96.5%) clarified that COVID-19 was not the main factor behind their decision to evacuate. Among those who evacuated to alternative destinations, nearly two-thirds (61.8%) were pet owners, and of these, approximately half (44.7%) acknowledged that their pets had influenced their decision to seek shelter outside of a traditional hurricane shelter (see Appendix Q62, Q63). Various responses highlighted the difficulty in finding available motels or hotels that allowed pets (see Appendix Q66). For example, one respondent from Pinellas County indicated they "could not find a hotel within a reasonable distance that had at least three consecutive days open and would accept pets even though DeSantis had said hotels were to accept pets. Also, the pet fees were too high." At the same time, another resident from Pinellas County stated, "we were ready to go but knew we would have to go to Georgia to find a pet-friendly hotel, so we waited to see if the projected path changed." One individual from St. Petersburg shared how the "motel we had made reservations for canceled, saying they had to evacuate. I had to quickly arrange to go to a friend's house. Also, I have pets, making evacuation/hotel reservations more difficult."
Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1.
Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC
Only 2.1% of those surveyed decided to stay in a public shelter. Most used their personal vehicle to get to the shelter (85.1%), while others walked, used public transport, took a taxi, hired a car-share, or rode in an ambulance (see Appendix Q71). About a quarter of respondents (23.4%) had pets, and of those, 63.6% could bring their pets to the shelter (see Appendix Q72, Q73). Most respondents (93.6%) said that COVID-19 did not affect their decision to evacuate to a shelter (see Appendix Q67).
Nearly one-third (27.6%) of participants indicated the presence of pre-existing health conditions, which elevated their susceptibility to severe illness if infected with COVID-19.
Specifically, 13.9% of all respondents were immunocompromised, and 25.1% were aged 65 or older. Approximately two-thirds (65.3%) mentioned that they or someone in their household had an underlying medical condition associated with a higher risk for severe COVID-19. The analysis revealed that those respondents with health risks for severe COVID-19 were not more inclined to evacuate than those without health risks. However, individuals with at least one person in their household aged 65 or older were more likely to evacuate to a different location (48.2%) (such as a hotel, shelter, or elsewhere) than to stay at home (42.5%).
In response to the question, 'Do you feel you were provided timely information to make a good evacuation decision?', most of the participants believed they had received information in a timely manner to make informed decisions about evacuating. Specifically, 70.3% strongly agreed with this statement, 24.6% somewhat agreed, and only 5.1% did not agree at all. The survey investigated the use of storm surge maps in evacuation decision-making, specifically for Hurricane Ian. A significant majority (88.6%) of respondents indicated having seen a storm surge map similar to the one presented in Figure 7 during Hurricane Ian. Conversely, only 8.3% reported not seeing such a map, and 3.1% remained unsure (see Appendix Q40).
Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1. This high level of awareness regarding storm surge maps during Hurricane Ian is encouraging.
Further analysis revealed a positive association between storm surge map awareness and evacuation decisions. Among respondents who reported seeing a storm surge map during Hurricane Ian (86.5% of all respondents), 60.0% definitively acknowledged considering these forecasted surge maps when deciding whether or not to evacuate. An additional 17.2% indicated probable consideration, suggesting that storm surge maps likely played a role in the evacuation choices of over 77% of the surveyed population. While some respondents (22.8%, including those who answered, "may or may not have," "probably not," and "definitely not") reported that surge maps had little to no influence on their evacuation decisions, the overall trend suggests that these Similarly, the question regarding the cone's size revealed a knowledge gap. The cone is supposed to depict the area of potential track uncertainty (National Hurricane Center 2024).
However, 25.9% of respondents (strongly agreed + agreed) believed the cone represented the storm's physical size. Disagreement with this statement (33.5% + 31.0%) indicated a more accurate understanding among a larger portion of the population. The survey also explored respondents' grasp of the cone's limitations in depicting risk zones. A concerning 6.2% (strongly agreed + agreed) believed only areas within the cone faced hurricane impacts, while nearly 90%
(disagreed + strongly disagreed) grasped that impacts could extend beyond the shaded area.
Finally, a survey question explored respondents' understanding of the cone's ability to depict all potential storm paths. While nearly half (44.4%) believed the cone was entirely comprehensive (strongly agreed + agreed), this indicates a misconception. The cone depicts the most likely path with a 66% chance of containing the center's track (National Hurricane Center 2024). It does not encompass every possible path the storm could take.
In the survey, participants were asked to rate the sources of information they relied on when making decisions related to Hurricane Ian using a Likert scale ranging from 1 (not relied on at all) to 5 (most relied on) (see Table 3 and Figure 9). The most heavily relied upon sources of
N Mean score (± S.D.) Local Media 1936 3.32 (± 1.07) Electronic Media 1798 2.80 (± 1.13) County/City Emergency Preparedness Websites 1772 2.81 (± 1.16) Mobile Weather Apps 1804 2.72 (± 1.20) Government Officials 1743 2.60 (± 1.19) National Media 1777 2.30 (± 1.30) Family and Friends Near You (<50 miles) 1539 1.35 (± 1.35) Distant Family and Friends (>50 miles) 923 1.35 (± 1.33) Neighbors 1462 1.25 (± 1.31) Social Media 1504 1.16 (±1.44) Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1. Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC Radio Broadcasts 1464 1.11 (± 1.39) Print Media 747 0.72 (± 1.04) Table 3. Information Sources That Individuals Used to Make Prior Hurricane Evacuation Decisions. Note. "N" and "SD" denote the number of survey responses and standard deviation.
The mean score is determined on a scale of 1 to 5, where 5 signifies "more likely to rely upon" when making prior hurricane evacuation decisions.
Figure 9. This bar graph depicts the information sources that individuals most relied upon when making evacuation decisions before Hurricane Ian. The lengths of the bars represent the average response on a scale of 1 (least likely) to 5 (most likely) to rely on a particular source.
Across four hurricanes-Laura, Sally, Ida, and Ian-public concern for various hazards remained remarkably consistent (Polen et al. 2023;Collins et al. 2022a). However, a troubling disparity emerged: despite being historically the deadliest threat (Brennan et al. 2023), storm surge consistently ranked as the least concerning hazard among this sample of respondents. This disconnect is particularly concerning, considering 29% of Hurricane Ian respondents resided in 100-year flood zones. While their level of concern (2.25) was slightly higher than the overall average (1.96), it still trailed significantly behind wind, intensity, and gusts, which topped the list
Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1. Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC The utilization of public shelters by study participants accounted for only 2.6% of the sample, consistent with previous research conducted by the same author group (Polen et al. 2023; Collins et al. 2022a) during Hurricanes Ida, Laura, and Sally, where only 0.9% of the overall sample for all three hurricanes sought refuge in public shelters. This aligns with established findings in hurricane evacuation research, which indicate that individuals with greater resources tend to prefer evacuation plans that do not involve public shelters. These shelters have long been viewed unfavorably, even before the pandemic (Lindell & Perry 2012). Other pre-pandemic studies examining shelter usage calculated a mean evacuation rate to shelters of approximately 5%, ranging from 2% to 11% (Wong et al. 2021).
The survey revealed a significant association between pet ownership and evacuation decisions during Hurricane Ian. A substantial proportion (70.0%) of those who chose to shelter in place were pet owners, and for nearly half (50.1%) of these pet owners, their pets played a critical role in the decision to remain at home. This finding is supported by previous research, which suggests pet ownership can be a major factor influencing evacuation behavior (Farmer and DeYoung 2019;Brackenridge et al. 2012;Hunt et al. 2012).
Residents in mobile homes evacuated at the highest rates compared to those in other housing types. This could be because residents weigh the perceived risk of staying in a potentially less secure structure against the challenges of evacuation.
While the immediate influence of COVID-19 on evacuation decisions might be decreasing, its intricate tie to pandemic safety measures creates a clash for some. Wu et al. (2023) propose that concerns about personal COVID-19 risk can even discourage evacuation, potentially due to conflicting safety protocols. This contradicts the idea of a simple link between overall risk tolerance and evacuation (Greer et al. 2023) and Diaz et al.'s (2023) finding that vulnerability directly drives evacuation intent. A deeper understanding of the nuanced interplay between COVID-19 anxieties and evacuation choices is crucial. In this study, 60.5% of respondents stated that COVID-19 did not affect their decision to stay home. In earlier research on Hurricanes Laura and Sally by Collins et al. (2022a), 13.2% indicated that COVID-19 did not affect their evacuation decision. These findings suggest that as vaccines have become more accessible and COVID-19 management has improved, the pandemic has exerted less influence on evacuation decisionmaking within this sample than its earlier stages. This is confirmed by the research conducted Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1. Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC during Hurricane Ida in 2021, where 80.5% of respondents indicated that COVID-19 did not affect their evacuation decision (Polen et al. 2023). A relatively small percentage (7.7%) of people who stayed home during Hurricane Ian indicated they wanted to evacuate but did not have enough time (see Appendix Q56). This is significantly lower than the proportion of respondents who reported wanting to evacuate but running out of time during Hurricanes Ida (22%), Sally (15.7%), and Laura (11.1%) (Polen et al. 2023; Collins et al. 2022a). This discrepancy might be attributed to the improved messaging highlighting the rapid intensification of the storm and the potential for significant storm surge, irrespective of the hurricane's precise path. Furthermore, only 5.1% of participants reported feeling they lacked sufficient information to make a "good evacuation decision" during Hurricane Ian, compared to 13.2% during Hurricane Ida (Polen et al. 2023) (see Appendix Q37). This marked difference further highlights the improved effectiveness of messaging efforts between the two storms. The majority of respondents said they mostly relied on the local media for information about the hurricane (see Appendix Q91). This aligns with similar patterns observed in the aftermath of Hurricanes Laura, Sally, and Ida (Polen et al. 2023; Collins et al. 2022a). Social media had a mean score of 1.16, far lower than those given during Hurricanes Laura and Sally (2.72) and Ida (2.45). This could be due to demographic differences among the surveys, improved communication from local channels, or a heightened awareness of misinformation spread by social media. The survey results revealed a knowledge gap concerning the hurricane cone of uncertainty (see Appendix Q41, Q42, Q43, Q44). A significant portion of respondents (25.9%) held the misconception that the cone represents the storm's physical size, highlighting the need for clearer communication about the cone's purpose. Encouragingly, a larger portion (64.5%) demonstrated a more accurate understanding by disagreeing with this statement. Similarly, the survey identified misconceptions about the cone's limitations in depicting risk zones. A concerning 6.2% of respondents believed only areas within the shaded cone faced hurricane impacts. This underestimation of risk underscores the importance of public education efforts emphasizing the potential for impacts beyond the cone's boundaries. Perhaps most concerning was the finding that nearly half (44.4%) of respondents believed the cone depicted all possible storm paths. The cone represents the most likely path with a 66% chance of containing the center's track. This highlights Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1. Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC a critical knowledge gap that could lead to inadequate preparedness measures in areas outside the cone. The current study's findings on public misconceptions regarding the hurricane cone of uncertainty align with previous research by (Evans et al. 2022), who identified similar misunderstandings about the cone's purpose and limitations. Our findings further emphasize the prevalence of the misconception that the cone depicts the storm's size, as reported in (Broad et al. 2007). This consistency across studies highlights the persistence of this knowledge gap and underscores the need for sustained public education efforts.
This research presents a significant strength in its synergy with the extensive body of work
conducted by et al. (2021a, 2021b, 2022b) on hurricane evacuation decision-making within the context of the ongoing pandemic. This seamless alignment with prior research bolsters the coherence and continuity of the findings, enriching the understanding of this critical issue. While this study provides valuable insights, acknowledging its limitations is crucial for robust interpretation. First, sample sizes vary greatly across hurricanes. The researchers' current investigation of Hurricane Ian boasts 2193 respondents, significantly exceeding the 119 for Ida, 721 for Sally, and 126 for Laura (Polen et al. 2023; Collins et al. 2022a). Furthermore, geographic and demographic discrepancies exist. Surveys for Laura and Ida focused on Louisiana residents, while Sally's sample consisted of Alabamans. In contrast, this Hurricane Ian data is exclusive to Florida. These regional differences, coupled with potential variations in demographics, necessitate caution when drawing direct comparisons across storms. Furthermore, a relatively small number of respondents chose to utilize public shelters, which limited the ability to analyze several questions specific to shelters. As previously mentioned, this limitation is partly attributed to the study sample's composition, which does not reflect the broader population residing in Florida. Consequently, this lack of representativeness impacts the generalizability of the research findings. While not ideal for achieving broad representation, the combined snowball and convenience sampling methods are recognized as effective within the realm of disaster research, as noted by Norris (2006) and Price (2013). However, a drawback is the lack of comprehensive insight into the actual distribution within these affected regions, and the assurance of population representativeness is not guaranteed. Nonetheless, despite its limitations Accepted for publication in Weather, Climate, and Society. DOI 10.1175/WCAS-D-24-0007.1. Brought to you by UNIVERSITY OF UTAH | Unauthenticated | Downloaded 07/31/25 02:46 PM UTC
concerning representational accuracy, this method proved to be an efficient means of reaching out to a community that continues to recover over a year after the hurricane.
The demographic imbalances observed in the sample can be explained by examining prior published research. Studies have found that digital and email-based surveys often exhibit lower representation than alternative survey methods like mail surveys (Dillman 2000;Mackety 2007;Tse 1998). Email samples tend to skew toward individuals from middle and upper socioeconomic classes, characterized by higher levels of education and income, as well as a greater prevalence of white respondents, in contrast to the demographic composition observed in other survey modes (Kehoe et al. 1999;Mehta & Sividas 1995;Sheehan & Hoy 2000;Witte et al. 2000). Additionally, it is well-documented in the literature that women and individuals of white ethnicity tend to exhibit higher response rates across all survey methodologies when compared to their counterparts (Green & Stager 1986;Kehoe et al. 1999;McCabe et al. 2002).
The sampling method used remains a matter of concern, particularly considering that various studies emphasize the vulnerability of minority groups during disasters. Racial and ethnic minorities face heightened risks during pandemics due to their limited capacity to implement preparedness strategies and endure the resulting impacts, considering inequalities in their underlying health status and social determinants. These disparities encompass cultural, educational, and linguistic barriers, socioeconomic differences, and limited access to healthcare, as highlighted by Hutchins et al. (2009) and Yusuf et al. (2023).
While the developed instrument allowed for a more tailored assessment of the research variables, the lack of prior validation other than the pilot test introduces potential limitations in the reliability and generalizability of the findings. Also, while the survey's extensive length was essential for addressing the research questions and ensuring the survey instrument's reliability, the researchers acknowledge that the survey's extended length likely affected the quantity of completed surveys. Furthermore, respondents may have difficulty accurately remembering specific details about their evacuation experience due to the stressful and chaotic nature of the event. This potential for recall bias should be considered when interpreting the findings.