1. Introduction
Extreme heat events are one of the deadliest natural hazards in the United States (Berko et al. 2014; Gasparrini et al. 2015; U.S. EPA and CDC 2016) and pose deadly threats to people worldwide (Mora et al. 2017; Franzke and Torelló i Sentelles 2020). Extreme heat is projected to increase in frequency and severity in response to increasing atmospheric concentrations of greenhouse gases driven by human activity (Jeon et al. 2016; U.S. EPA and CDC 2016; Angélil et al. 2017; Vose et al. 2017; IPCC 2021). Urbanization is also increasing the number of people exposed to deadly heatwaves (Tuholske et al. 2021). Furthermore, there is demonstrated influence of human activity on the severity of heat-health impacts (Vicedo-Cabrera et al. 2021), and individual behavior and risk judgements can lead to different impacts across similarly exposed populations (Semenza et al. 2008; White-Newsome et al. 2011; Lefevre et al. 2015; Wilhelmi and Hayden 2010). Increasing physical exposure to extreme heat and its complex interaction with social sensitivity factors associated with social inequities in hazard impacts and responses (such as gender, age, and race/ethnicity) create varying risk environments for different subpopulations across the country. This underscores the need for decision-makers and risk managers to develop strategies and define priorities to mitigate the negative impacts of extreme heat, since heat mortality and morbidity are often preventable if appropriate individual and collective actions are taken.
In this study, we examine how sociodemographic indicators associated with health disparities in the impacts of extreme heat also influence risk perceptions across the contiguous United States. Using georeferenced survey data and multilevel regression modeling, we report the associations of individual-level factors (e.g., gender, age, race/ethnicity, and work status) with risk perceptions, while also estimating risk perceptions among different subpopulations. These results provide decision-makers with valuable information about which vulnerable subpopulation tends to perceive (or not) the threat of extreme heat, which informs targeted risk communication and hazard preparedness campaigns.
2. Literature review
a. Extreme heat risk
While there is no universal definition of an extreme heat event or heatwave, these events are commonly understood as periods characterized by excessively high levels of temperature and/or humidity that jeopardize human health due to severity of exposure or duration (Robinson 2001; Smith et al. 2013; White-Newsome et al. 2014; U.S. EPA and CDC 2016; Hawkins et al. 2017; Liss et al. 2017). Mora et al. (2017) found that about 30% of the global population is exposed to deadly heat conditions for at least 20 days each year, and this number is expected to increase to between 48% and 74% by 2100 under different global warming scenarios. As temperatures continue to rise, a greater proportion of U.S. citizens will be exposed to extreme heat conditions in the future (Jones et al. 2015).
Extreme heat is a commonly experienced hazard with both immediate and delayed negative health impacts that can result in illness and fatalities during pronounced heatwaves. For example, in July 1995, during a 5-day extreme heat event in Chicago, Illinois, over 700 deaths were recorded in excess of historical norms, representing an increase of 85% from the previous year (Semenza et al. 1996; Klinenberg 2003). In May 2015, record temperatures throughout southern India led to at least 2320 confirmed fatalities (Ratnam et al. 2016; Mazdiyasni et al. 2017). In August 2003, a particularly severe heatwave affected much of western Europe claiming more than 70 000 lives (Robine et al. 2008). Despite these high numbers, heat deaths are likely underreported due to heat’s tendency to exacerbate existing medical conditions (Åström et al. 2011; Liss et al. 2017; Mora et al. 2017). Some negative heat-health impacts such as dizziness and fatigue are experienced by a broader segment of the population (Khare et al. 2015; Hayden et al. 2017). For example, a study in England found that more than one-half of the younger adults reported experiencing headache and sunburn during summer 2013 (Khare et al. 2015). The intensity and scope of these impacts are influenced by geographic factors, population dynamics, time, scale, and the efficacy of communities’ adaptive policies (Semenza et al. 1996; Klinenberg 2003; U.S. EPA 2006; Anderson and Bell 2011; Reid et al. 2012; IPCC 2014; Tierney 2014).
b. Risk assessment and extreme heat
Assessment of vulnerability and risk is critical to identify priorities and develop management strategies (IPCC 2012). Decision-makers need locally relevant information about the distribution of potential negative impacts to inform mitigation and risk reduction strategies. The risks associated with climate change and natural hazards can be assessed by supplementing physical models of hazard exposure (Gill and Malamud 2014; Hawkins et al. 2017; Mora et al. 2017) with analyses that seek to incorporate dynamic human vulnerability factors that affect sensitivity and adaptive capacity (Reid et al. 2009; Tomlinson et al. 2011; Buscail et al. 2012; Wolf and McGregor 2013; Weber et al. 2015). Vulnerability is a key determinant of potential impacts of hazardous events, and sensitivity and lack of adaptive capacities are in turn causes of vulnerability (IPCC 2012). Sensitivity refers to the potential of being negatively affected by hazards due to personal, household, and contextual factors (such as social, economic, political, or cultural factors) that magnify the impact of a hazard event (Grothmann and Reusswig 2006; Johnson et al. 2012; Reid et al. 2012; IPCC 2014; Tierney 2014; Jones et al. 2015). Adaptive capacity is the ability of individuals or a group to take actions that mitigate hazard risks such as social capital (Kalkstein and Sheridan 2007; Bobb et al. 2014; IPCC 2014; Tierney 2014; Jones et al. 2015). While the ability to predict climatic changes and the occurrence of heat events on a global scale by better understanding the dynamic properties and interactions of Earth’s natural systems has improved (Schellnhuber 1999; Famiglietti et al. 2015), the dynamic properties of human systems remain difficult to capture in comprehensive risk assessments.
In the context of extreme heat, some sociodemographic factors (see Table 1) have been associated with disparities in morbidity and mortality from extreme heat and included in risk assessments as indicators of heat vulnerability (Harlan et al. 2006, 2013; Medina-Ramón et al. 2006; Anderson and Bell 2009, 2011; Reid et al. 2009, 2012; Buscail et al. 2012; Johnson et al. 2012; Wolf and McGregor 2013; Gronlund et al. 2014; Weber et al. 2015). Age is a demographic factor of heat vulnerability because older individuals are statistically more likely to be negatively impacted by extreme heat exposure as they tend to be more physiologically susceptible to heat, more limited in their ability to access health services due to mobility constraints, and more prone to social isolation (Semenza et al. 1996; Stafoggia et al. 2006; Reid et al. 2009; Uejio et al. 2011; Wolf and McGregor 2013; Gronlund et al. 2014; Liss et al. 2017). In the United States, epidemiological studies have found that men have higher rates of heat-related mortality and morbidity than women during extreme heat events (Semenza et al. 1996; Whitman et al. 1997; Choudhary and Vaidyanathan 2014; Hess et al. 2014; Schmeltz et al. 2015). Being active in the heat and lower social contact may contribute to higher heat vulnerability among men, although women face socioeconomic inequities in the United States that may also increase risk (Kovats and Hajat 2008). People with lower educational attainment tend to face greater natural hazard risks in general due to difficulties they face in accessing health services and hazard information (Cutter et al. 2003; Reid et al. 2009; Anderson and Bell 2011; Weber et al. 2015). Low-income and socioeconomically disadvantaged people, particularly disabled individuals, are significantly more likely to be negatively affected by natural hazards, including extreme heat, due to a lack of resources required to cope with the hazard (Harlan et al. 2006; Anderson and Bell 2009; Reid et al. 2009). Previous studies have indicated that larger households (with a greater number of residents) tend to have greater access to the social and material resources required to cope with heat hazards (but are more likely to have children more susceptible to negative heat impacts), whereas smaller households are more prone to social isolation, a significant source of vulnerability (Semenza et al. 1996; Cutter et al. 2003; Klinenberg 2003, 80–81; Reid et al. 2009; Weber et al. 2015). Due to social, political, and economic inequities, minoritized racial and ethnic populations often experience greater health impacts from extreme heat (Cutter et al. 2003; Reid et al. 2009; Anderson and Bell 2011; Weber et al. 2015), and they can also be more exposed to extreme heat at the neighborhood level due to historic patterns of discrimination such as redlining (Benz and Burney 2021). These social, economic, and demographic factors can be categorized as “sensitivity” factors, but they may also influence adaptive capacity in shaping overall vulnerability.
Summary of sensitivity factors known to influence extreme heat risk.
c. Risk perception
In addition to these sensitivity factors, risk perception has also been acknowledged as an important factor of heat vulnerability (Wilhelmi and Hayden 2010). Risk perception is a determinant of individual risk decision-making and influences the likelihood of an individual engaging in personal protective behaviors (Slovic 1987; van der Pligt 1996; Brewer et al. 2004). Personal behavior and preparedness can either attenuate or exacerbate vulnerability. The relationship between risk perception and behavior has been studied with respect to certain environmental and health hazards (Wachinger et al. 2013). Previous studies have found that heat risk perceptions positively influence heat-protective behaviors (Lane et al. 2014; Hayden et al. 2017; Madrigano et al. 2018; Ban et al. 2019; Hass and Ellis 2019; Zander et al. 2019; Hass et al. 2021). For example, a recent U.S. national survey found that risk perceptions and subjective experience with health effects of extreme heat predicted heat-protective behaviors (Esplin et al. 2019). Data on risk perceptions provide information on how individuals perceive their own vulnerability and their likelihood of taking protective action (Tierney 2014), which is increasingly sought by government officials and risk managers (Wolf et al. 2010; Reid et al. 2012; White-Newsome et al. 2014).
While sociodemographic sensitivity factors such as age and housing characteristics can be included in risk assessment due to the availability of census data at subnational levels, risk assessment typically lacks data on risk perception (Wilhelmi and Hayden 2010). Furthermore, little is known about what data may be good proxies for heat risk perception due to a lack of knowledge about how key sensitivity factors are associated with risk perception. Existing knowledge is limited to surveys in a small number of cities (Kalkstein and Sheridan 2007; Madrigano et al. 2018; Chakalian et al. 2019). For example, a study conducted in New York City found that low-income individuals were more likely to be concerned about heat, but men—who also have elevated vulnerability to heat—tended to have lower heat risk perceptions (Madrigano et al. 2018).
Risk reduction strategies may be more effective if they account for individual-level social factors related to hazard awareness, risk judgements, and subsequent decision-making behaviors that likely vary at subnational levels (Slovic 1987; Renn 1998; Howe et al. 2019). Failure to account for risk perception in risk assessment can lead to inadequate hazard communication and misguided management priorities. For example, a lack of knowledge about the association of sensitivity factors and risk perception may result in difficulties in identifying communication priorities since little is known whether vulnerable populations perceive their elevated vulnerability. If a certain vulnerable subgroup does not perceive a higher risk of extreme heat events for themselves, their family, and their community, the subgroup should be a priority for practitioners to target risk communication efforts.
To bridge the knowledge gap, this study investigates how sociodemographic factors are associated with heat risk perception, using nationally representative survey data from the contiguous United States. This study asks, how do key social, economic, and demographic factors known to be important indicators of mortality and morbidity from extreme heat (summarized in Table 1) relate to extreme heat risk perceptions? We hypothesize that individual-level factors that have been found to be associated with greater personal risk of heat-related impacts in previous studies will be positively associated with heatwave risk perceptions. This study complements Howe et al. (2019), which describes place-based geographic patterns in heat risk perceptions at multiple scales (census tract, county, and state) across the United States using small-area estimation models. Building on the same dataset, in this paper we focus on understanding how individual sociodemographic factors predict heat risk perceptions and how such factors interact with each other. By focusing on the predictors of heat risk perceptions, this research helps to identify particular subpopulations who face well-documented vulnerability but are less likely to perceive themselves to be at amplified risk from extreme heat. Such information can help decision-makers to define communication priorities and assess hazard vulnerability and risk in a more comprehensive way.
3. Methods
a. Study area and data
This study examines heatwave risk perceptions across the contiguous United States during the warm months of 2015 using nationally representative survey data (Fig. S1 in the online supplemental material). The survey was administered online biweekly over the course of 20 weeks, beginning in May. The survey was conducted on the GfK KnowledgePanel Omnibus, a shared-cost weekly online survey whose respondents are sampled from a probability-based panel. GfK recruited panel members using address-based sampling of all U.S. addresses from the U.S. Postal Service Delivery Sequence File and provided households without internet access with a computer and internet service (in our sample, 20% of respondents lacked home broadband internet access). The overall sample size was n = 10 532. However, because of the panel design of this survey, responses were collected more than once for some individuals. These subsequent responses were filtered from the dataset before analysis and the final sample size was n = 8789 unique respondents. Individual identifiers were removed from the data, and the precise geographic coordinates of respondents were jittered within a radius of 150 m for respondent confidentiality.
A heat wave is a period of unusually and uncomfortably hot weather. If a heat wave were to occur in your local area, how much, if at all, do you think it would harm the following: Your health? Your family’s health? The health of others in your community?
The responses to each of the survey questions, which were collected using a slider bar on a 0–100 scale, were combined to create an overall heatwave risk perception index used as the dependent variable in this study. This index had high internal consistency (Cronbach’s alpha = 0.95). The index represents heatwave risk perception values on a scale of 0–100, with 100 representing the highest degree of perceived risk to heat. The high internal consistency of the heatwave risk perception index suggests that it captures a single construct.
The survey also collected data on the sociodemographic characteristics of each respondent. Seven sociodemographic variables (gender, age, race/ethnicity, income, education, work status, household size) were used in this study’s regression analyses along with geographic data recorded for each response. (The structure of these variables is detailed in Table S1 in the online supplemental material.)
b. Analysis
The scope of this analysis is focused on evaluating the sociodemographic factors associated with risk perceptions, rather than developing an exhaustive model capturing all possible factors. We fit a random intercept (multilevel) regression model to the heatwave risk perception index, parameterized according to statistical best practices for confirmatory hypothesis testing (Hofmann 1997; Gelman and Hill 2007, chapters 11–12; Zuur et al. 2009; Barr et al. 2013). The purpose of the models in this paper is explanatory rather than predictive, and it is designed to test hypotheses about associations between known vulnerability factors and risk perceptions. The same methods and statistical techniques described below for the initial model build were applied to each subsequent model. All analyses were performed using the R programming language and environment using the lme4 package (Bates et al. 2015).
The initial model (Table S2 in the online supplemental material) was composed solely of categorical random effects (Winter 2013; Hofmann 1997; Barr et al. 2013). The model coefficients (effects) associated with these predictors and their sublevels are random effects estimated with partial pooling—also known as linear unbiased prediction (Winter 2013; Goldberger 1962; Gelman and Hill 2007, chapter 12). By treating the extreme heat risk factors addressed in the study hypotheses as random effects, the effect of the levels of each predictor can be assessed in relation to their difference from the overall mean (i.e., the average risk perception score across the U.S. population) (Robinson 1991; Hofmann 1997; Barr et al. 2013).
Multilevel regression models use best linear unbiased predictors (BLUPs) to predict random effect values rather than estimate fixed parameters and establish a hierarchical framework through which meaningful differences between levels can be discerned. The BLUPs are analogous to prediction in the empirical Bayes methodological framework, in which parameters associated with a prespecified prior distribution are estimated from the data, thereby approximating the full hierarchical Bayes model (Hofmann 1997; Gelman and Hill 2007, chapter 11; Barr et al. 2013). By utilizing prediction instead of estimation, the strengths of Bayesian inference can be integrated within a classical statistical framework to support hierarchical linear modeling. Consequently, we employ BLUPs because the primary interest of this study is in making inferences about the distribution of risk perception values, their degree of variance at different levels, and the underlying population more so than in the effects themselves (e.g., fixed effects) or explicitly testing for measurable differences between specific levels (Gelman and Hill 2007, chapter 11).
In a mixed effect model, intercorrelations between fixed effects can quickly be assessed en masse via a correlation matrix; however, random effect models require systematic evaluation of each predictor’s individual contribution to the model. Multilevel modeling best practices (Hofmann 1997; Gelman and Hill 2007) involve starting with a maximal model and using log-likelihood tests to iteratively pare down the number of predictors. Best practices also indicate that in many circumstances, it is more appropriate to retain predictors that would otherwise be eliminated after the log-likelihood test because they are important to the conceptual or theoretical framework adopted across the study—for example, including or excluding the theoretically important random effect “Education” had no quantifiable impact on model output (Table 2).
Model results predicting heatwave risk perception index. Observations = 8789; CI indicates confidence interval. One, two, and three asterisks indicate significance at p < 0.05, p < 0.01, and p < 0.001, respectively. Note the different column meanings for fixed effects (the intercept).
Our model specification includes the following sociodemographic predictors: gender, age, race/ethnicity, income, education, work status, and household size. (Descriptive statistics are available in Table S1 in the online supplemental material.) In addition to these sociodemographic variables, we also include an interaction term for gender by race and ethnicity, since this interaction is supported by previous research on hazard risk perceptions: the “white male effect” found in many risk perception studies (that white males tend to exhibit lower risk perceptions than other demographic groups) indicates that the interaction of gender and race/ethnicity is important to include in models of risk perceptions, since the effects of gender and race/ethnicity alone do not fully capture the effect (Finucane et al. 2000). In addition, by using random effects associated with geographic factors (census region, state), the model was able to account for some degree of spatial autocorrelation and overcome assumptions of independence that would normally be violated if geographically clustered data were to be analyzed using traditional linear regression modeling (Hofmann 1997; Gelman and Hill 2007, chapter 11).
Model results describe intergroup variation across sociodemographic factors hypothesized to influence heatwave risk perceptions. The outcome variable is a risk perception index on a scale of 0–100 with 100 representing the highest degree of perceived risk. Random effects included in this model provide a direct measure of how much of the reported risk perception scores’ variance around this mean is explained by group-level differences.
4. Results
Nationwide, the mean heatwave risk perception index was 39 (n = 8789; std dev = 24) on a 0–100 scale (Fig. S2 in the online supplemental material). Heatwave risk perception was associated with the following statistically significant predictors: race/ethnicity, income, gender, work status, age, state, and region (Table 2).
Income was a statistically significant predictor of individual heatwave risk perceptions with a large effect size [σ = 3.72; χ2(1) = 89.52, with p < 0.001]. Higher-income individuals tend to have lower risk perceptions than lower-income individuals (Fig. 1) and the national average. Holding other predictors constant at their means, respondents earning less than $15,000 per year scored 1.26 times as high on the heatwave risk perception index (47) as respondents earning over $150,000 per year (37).
Effects of model predictors with associated 95% confidence intervals, excluding state and region. Points represent best linear unbiased predictor estimates for random effects in a multilevel model.
Citation: Weather, Climate, and Society 14, 4; 10.1175/WCAS-D-21-0104.1
The race and ethnicity variable was also a strong and significant predictor of heatwave risk perceptions [σ = 3.51; χ2(2) = 103.98, with p < 0.001]. Holding other variables constant, white, non-Hispanic or Latino respondents had the lowest estimated heatwave risk perception index at 37, while Hispanic or Latino (44) and Other, non-Hispanic or Latino respondents (47) had the highest estimated heatwave risk perception index (this category includes non-Hispanic or Latino Asian, American Indian or Alaska Native, and Native Hawaiian or other Pacific Islander U.S. residents). Gender was a statistically significant predictor of heatwave risk perceptions [σ = 2.32; χ2(2) = 80.27, with p < 0.001]. Although the effect was not large, the heatwave risk perception index was higher among women (44) than men (41). While the race/ethnicity by gender interaction did not significantly improve model fit overall [σ = 0.95; χ2(2) = 1.84, with p = 0.17], white non-Hispanic or Latino male respondents tended to have much lower heatwave risk perceptions scores (35) than the mean for all other race by gender groupings (43).
Work status was also a strong and statistically significant predictor of heatwave risk perceptions [σ = 2.99; χ2(1) = 29.77, with p < 0.001]. Across five work status categories, disabled nonworking respondents reported much higher heatwave risk perceptions (48) than those in the remaining four work status categories (not working—seeking a job, 42; working, 41; not working—retired, 41; not working—other, 40).
Age was a small but significant predictor of heatwave risk perceptions [σ = 1.08; χ2(1) = 6.17, with p = 0.0129]. Respondents in the older age categories (65 years and older and 45–54 years) had slightly higher heatwave risk perceptions (44) than those in the category 35–44 years old (41).
The remaining sociodemographic variables did not significantly improve model fit. Heatwave risk perceptions did not show significant variation by educational attainment [σ = 0.37; χ2(1) = 0.09, with p = 0.76] or household size [σ = 0.58; χ2(1) = 1.35, with p = 0.25].
We estimated variation in the heatwave risk perception index across geographic units (state and region) using the same techniques, by specifying geographic units as random effects. Respondents’ state of residence was a statistically significant predictor of heatwave risk perceptions [σ = 2.25; χ2(1) = 24.94, with p < 0.001]. At a broader scale, the U.S. census region in which each state was grouped was also a statistically significant predictor of risk perceptions and explained variation beyond that at the state level [σ = 2.23; χ(1) = 10.62, with p = 0.001]. The Midwest tended to have the lowest heatwave risk perceptions (39.9) while the South had the highest risk perceptions (44.9). (Geographic effects are summarized in Fig. S3 in the online supplemental material.) Howe et al. (2019) provides additional detail on geographic variation in heat risk perceptions at multiple scales.
5. Discussion
The principal objective of this study was to determine how key sociodemographic factors known to be important contributors to overall heat vulnerability (summarized in Table 1) also influence heatwave risk perceptions across the contiguous United States. Several individual-level sociodemographic factors were associated with differences in heatwave risk perceptions—either positively or negatively, as hypothesized—and accounted for a statistically significant proportion of total variance around the national average. Overall, sociodemographic predictors explain a similar amount of individual variation in heatwave risk perceptions as they do risk perceptions of other hazards (Peacock et al. 2005; Lindell and Hwang 2008; Kellens et al. 2011; Knuth et al. 2013).
This study also has several limitations. While our findings are based on a nationally representative survey sample and generalizable to the U.S. population, low-population sociodemographic groups are less represented in our sample, which limits the ability to draw conclusions about their heat risk perceptions. Our survey data were collected during one season (summer 2015), which may limit our ability to generalize to other seasons where heat is a potential hazard (such as late spring or early fall) or other years in which the U.S. population may experience different patterns of weather conditions. A third limitation is that we focus here only on several survey questions on risk perceptions of heat. Resource constraints limited our ability to collect additional survey questions that may provide a fuller picture of impacts, decision-making, and responses to heat among the American public (e.g., Esplin et al. 2019). For example, future surveys should examine how experiences with direct and indirect heat-health impacts may influence risk perceptions.
Heatwave risk perception indices for subpopulations known to be at increased risk tended to deviate from the national average in line with the directionality of their effect on heat vulnerability, as found by previous research, with the notable exception of gender. Gender, a factor that previous studies have identified as an important determinant of extreme heat sensitivity, is an important determinant of risk perception. However, men—who experience more impacts from heat to their health (Semenza et al. 1996; Whitman et al. 1997; Kovats and Hajat 2008; Choudhary and Vaidyanathan 2014; Hess et al. 2014; Schmeltz et al. 2015)—perceive themselves to be at lower risk than women. This finding suggests particular importance for risk communicators to conduct targeted communication efforts to men in the United States. Minoritized racial groups are known to be at increased risk of being negatively impacted by extreme heat (Cutter et al. 2003; Klinenberg 2003, 80–81; Anderson and Bell 2009, 2011; Reid et al. 2009, 2012; Wolf and McGregor 2013; Weber et al. 2015) and also tend to have higher heat risk perceptions. Previous studies have found that working, nondisabled individuals are less sensitive to negative hazard impacts, while disabled persons are more susceptible to negative impacts (Semenza et al. 1996; Cutter et al. 2003; Klinenberg 2003, 80–81; U.S. EPA 2006; IPCC 2014; Ebi et al. 2018; U.S. EPA and CDC 2016). In this study, disabled nonworking respondents reported much higher heatwave risk perceptions. As hypothesized, respondents with higher incomes tended to have much lower heat risk perceptions than the national average and individuals with lower incomes tended to have higher risk perceptions.
The relatively low variance across some subpopulations may be partially a consequence of the conservative nature of mixed effect models, which rely upon partial pooling and combinations of individual-level and contextual-level characteristics that tend to pull subpopulation estimates toward their respective national averages. Despite this, a few at-risk subpopulations tended to have lower risk perceptions than expected (Fig. 1). Some factors known to increase vulnerability, such as age and education, were not associated with substantial differences in risk perception. Although age—a factor that previous studies have identified as an important determinant of extreme heat-health impacts (Klinenberg 2003; Anderson and Bell 2009, 2011; White-Newsome et al. 2014; Gronlund et al. 2014)—was found to be a statistically significant predictor of heat risk perceptions, practically it did not have a pronounced effect on extreme heat risk perception. The most senior subpopulation (≥ 65 years of age) reported only slightly higher risk perceptions than younger subpopulations despite their elevated risk. While we cannot identify whether this pattern is due to younger subpopulations overestimating their risk or older subpopulations underestimating their risk, we would still expect to find larger differences between the two groups if risk perceptions aligned with health risks. Since they do not, the possible underestimation of extreme heat risk by a particularly vulnerable subpopulation indicates that older populations may be less likely to take protective behaviors than would be appropriate given their risk profile. This is particularly significant given that an aging, increasingly urban U.S. population—with an increasing number of individuals considered to be vulnerable to heat (Basu 2009; Ortman et al. 2014; Jones et al. 2015; Lehner and Stocker 2015; Mora et al. 2017)—will likely be exposed to more frequent and intense extreme heat events—particularly in urban heat islands (Tomlinson et al. 2011; Li and Bou-Zeid 2013; U.S. EPA and CDC 2016). This increasing exposure, combined with a tendency to underestimate age-related risk, suggests that risk-reduction programs should also be focused on older individuals, including risk communication efforts.
No relationship was observed between education and heat risk perception despite the fact that individuals with lower educational attainment often face greater difficulty in accessing health services and information about the nature of natural hazards (Cutter et al. 2003; Medina-Ramón et al. 2006; U.S. EPA 2006; Anderson and Bell 2009, 2011; Reid et al. 2009, 2012; Smith 2013, 85–86; Weber et al. 2015; U.S. EPA and CDC 2016).
Additionally, previous research has identified household size as an important predictor of hazard risk, as larger households with more people living together are more likely to have the financial and social resources required to cope with environmental hazards and avoid social isolation (Semenza et al. 1996; Cutter et al. 2003; Klinenberg 2003, 80–81; Reid et al. 2009, 2012; Weber et al. 2015). In our model household size had no effect on heatwave risk perception when also controlling for income.
Overall, we find evidence that the socioeconomic factors associated with health impacts from extreme heat correspond in many ways to the factors associated with heat risk perceptions among the U.S. population. Income tends to be a strong predictor of heat risk perceptions, along with work status, gender, and race/ethnicity. Conceptually, income is directly associated with the ability to protect oneself from the heat through, for example, household adaptations such as installing and using air conditioning. Income is also associated with employment type and location. While our survey did not include detailed questions on employment type, higher-paying occupations tend to be located in indoor climate-controlled environments, while many outdoor occupations are lower paying (such as agricultural and construction labor) and employees in such outdoor occupations are exposed to greater heat risks.
The results of this study and Howe et al. (2019) show that heatwave perceptions do vary spatially and demonstrate statistically significant, nonrandom geographic patterns. People living in regions with histories of greater exposure to extreme heat events tended to have higher risk perceptions (Howe et al. 2019). However, this study indicates that the association of key sociodemographic variables with heatwave risk perceptions persists even after controlling for geography. In addition, our individual-level analysis identifies patterns less clearly visible at the state, community, or neighborhood level. For example, Howe et al. (2019) show that counties with older populations do not, on average, have higher heat risk perceptions than counties with younger populations. Our results, however, show a small but statistically significant positive relationship between age and heat risk perceptions across the population. Furthermore, we demonstrate the meaningful effects of certain key individual predictors (such as gender and work status) that may themselves vary less across communities but more between and within households and remain important factors for understanding how people perceive risks.
Taken together, heatwave risk perceptions demonstrate substantial variation across the U.S. population. For example, the combination of race and ethnicity with income illustrates a wide range of predicted heatwave risk perceptions (Fig. 2). Selected sociodemographic factors including income, race/ethnicity, work status, and gender exhibit similar or greater variance to the broadscale geographic factors of state and region. When combined, demographic and geographic factors are associated with large variation in risk perceptions across the population. Across all possible combinations, we estimate that the group with the highest heatwave risk perceptions (65.1) are Louisiana women 45–54 years old in the “other, non-Hispanic or Latino” race/ethnicity category (which includes Asian, American Indian or Alaska Native, Native Hawaiian or other Pacific Islander U.S. residents) who are disabled and not working with incomes of less than $15,000 per year. By contrast, the group estimated to have the lowest heatwave risk perceptions (22.0) are Minnesota men 35–44 years old in the “white, non-Hispanic or Latino” race/ethnicity category who are not disabled with incomes of greater than $150,000 per year.
Predicted heatwave risk perception index values for each combination of significant sociodemographic predictors. Each dot represents one type of individual based on each possible permutation of income, race/ethnicity, gender, age, and work status. Dots are ordered by estimated heatwave risk perception index and race/ethnicity.
Citation: Weather, Climate, and Society 14, 4; 10.1175/WCAS-D-21-0104.1
Findings of this study inform risk communication strategies and risk reduction management in two ways. First, for men and older adults, our study suggests that these groups tend to underestimate their elevated vulnerability from extreme heat (although we cannot rule out the possibility that comparison groups are relatively overestimating their vulnerability). The underestimation of risks is likely to contribute to maladaptation during extreme heat events (Esplin et al. 2019; Hass and Ellis 2019). This finding highlights the importance to conduct targeted risk communication and to help people in the United States to fully understand their risks. As compared with efficacy statements (e.g., information about the location of cooling centers), communication strategies that emphasize vulnerability (e.g., explanations about why all people are vulnerable to extreme heat) should be prioritized to test in future studies to better communicate heat-health risks (e.g., Li et al. 2021). Second, for low-income, nonwhite, and disabled subpopulations, this study found that these subpopulations have much higher heat risk perceptions than the national average, which is in line with their elevated risk of health impacts from heat. For risk management and communication with these subpopulations, this finding suggests that it is important to allocate resources (such as utility bill relief) to help at-risk populations cope with extreme heat. When communicating with such populations, efficacy statements about how to reduce their risks—as compared with strategies emphasizing vulnerability—might be more effective to help them overcome barriers to taking protective actions.
6. Conclusions
Using national survey data, we used hierarchical linear models to examine how sociodemographic and geographic variables relate to heatwave risk perception in the United States. The direction of heatwave risk perception predictors across the contiguous United States generally reflects trends identified in health impacts for many sociodemographic factors, with the notable exceptions of gender and, to some extent, age. Highlighting the distribution of perceived risk can help set priorities for subpopulation-specific risk communication strategies. Our results allow estimates of risk perceptions for specific subpopulations in relation to overall national trends. Variation in specific subpopulations, especially at the extremes, may be of particular interest for risk reduction efforts, including targeted risk communication.
Low risk perception increases vulnerability because people are less likely to respond to the hazards they do not perceive. In other words, what we believe to be real shapes our behavior—reactively or proactively. When vulnerable subpopulations, such as men and the elderly, do not perceive themselves to be at greater risk from heat, this presents barriers to risk reduction. This study found that age did not substantively influence heat risk perception, suggesting that older people may underestimate their elevated risk. In addition, men may also underestimate their increased risk from extreme heat events. These findings can inform risk communication programs to target these populations who may not currently fully understand their vulnerability. Effective risk communication strategies can reduce sensitivity to heat and enhance adaptive capacity by promoting protective behavior at the individual and community levels. For example, the protective behaviors promoted by risk communication campaigns might include risk awareness, avoiding unnecessary exposure, and developing personal heat-safety plans. The first steps in designing effective risk communication programs are identifying vulnerable subpopulations, studying their distribution, and evaluating their unique circumstances; data on risk perception and its association with sociodemographic factors help accomplish these goals.
Heat risk is increasing around the world due to global warming caused by anthropogenic greenhouse gas emissions and urbanization, but total hazard risk can be reduced by targeted interventions aimed at strengthening adaptive capacity and addressing human vulnerability factors (Adger 2006; Smit and Wandel 2006; Noble et al. 2014, 847–849). To do this, researchers, risk managers, and community members will need to work together to identify vulnerability factors (Mimura et al. 2014, 871–877; Ebi et al. 2018). This study details a new systematic approach for understanding risk perceptions across subpopulations using nationally representative survey data that is generalizable to the U.S. population. Leveraging advances in both the natural and social sciences to understand the drivers and distribution of heat vulnerability is vital to minimizing future loss in the face of rising exposure. Studying the landscapes of beliefs, risk perceptions, and behaviors can inform policy as well as our understanding of vulnerability at a range of temporal and spatial scales.
Acknowledgments.
This work was partially supported by the National Science Foundation (SES-1459903, SES-1459872, and BCS-1753082). Thanks are given to E. Helen Berry, Claudia Radel, and Beth Shirley for comments and suggestions.
Data availability statement.
Data that support the findings of the paper are available on the Open Science Framework (https://doi.org/10.17605/OSF.IO/RZ3VF).
REFERENCES
Adger, W. N., 2006: Vulnerability. Global Environ. Change, 16, 268–281, https://doi.org/10.1016/j.gloenvcha.2006.02.006.
Anderson, B. G., and M. L. Bell, 2009: Weather-related mortality: How heat, cold, and heat waves affect mortality in the United States. Epidemiology, 20, 205–213, https://doi.org/10.1097/EDE.0b013e318190ee08.
Anderson, B. G., 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 U.S. communities. Environ. Health Perspect., 119, 210–218, https://doi.org/10.1289/ehp.1002313.
Angélil, O., D. Stone, M. Wehner, C. J. Paciorek, H. Krishnan, and W. Collins, 2017: An independent assessment of anthropogenic attribution statements for recent extreme temperature and rainfall events. J. Climate, 30, 5–16, https://doi.org/10.1175/JCLI-D-16-0077.1.
Åström, D. O., F. Bertil, and R. Joacim, 2011: Heat wave impact on morbidity and mortality in the elderly population: A review of recent studies. Maturitas, 69, 99–105, https://doi.org/10.1016/j.maturitas.2011.03.008.
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.
Barr, D. J., L. Roger, C. Scheepers, and H. J. Tily, 2013: Random effects structure for confirmatory hypothesis testing: Keep it maximal. J. Mem. Lang., 68, 255–278, https://doi.org/10.1016/j.jml.2012.11.001.
Basu, R., 2009: High ambient temperature and mortality: A review of epidemiologic studies from 2001 to 2008. Environ. Health, 8, 40, https://doi.org/10.1186/1476-069X-8-40.
Bates, D., M. Mächler, B. Bolker, and S. Walker, 2015: Fitting linear mixed-effects models using lme4. J. Stat. Software, 67, 1–48, https://doi.org/10.18637/jss.v067.i01.
Benz, S. A., and J. A. Burney, 2021: Widespread race and class disparities in surface urban heat extremes across the United States. Earth’s Future, 9, e2021EF002016, https://doi.org/10.1029/2021EF002016.
Berko, J., D. D. Ingram, S. Saha, and J. D. Parker, 2014: Deaths attributed to heat, cold, and other weather events in the United States, 2006–2010 . National Health Statistics Reports, No. 76, U.S. Department of Health and Human Services, Washington, DC, 1–16, https://permanent.fdlp.gov/gpo60470/nhsr076.pdf.
Bobb, J. F., R. D. Peng, M. L. Bell, and F. Dominici, 2014: Heat-related mortality and adaptation to heat in the United States. Environ. Health Perspect., 122, 811–816, https://doi.org/10.1289/ehp.1307392.
Brewer, N. T., N. D. Weinstein, C. L. Cuite, and J. E. Herrington, 2004: Risk perceptions and their relation to risk behavior. Ann. Behav. Med., 27, 125–130, https://doi.org/10.1207/s15324796abm2702_7.
Buscail, C., E. Upegui, and J.-F. Viel, 2012: Mapping heatwave health risk at the community level for public health action. Int. J. Health Geogr., 11, 38, https://doi.org/10.1186/1476-072X-11-38.
Chakalian, P. M., L. Kurtz, S. L. Harlan, D. White, C. J. Gronlund, and D. M. Hondula, 2019: Exploring the social, psychological, and behavioral mechanisms of heat vulnerability in the city of Phoenix, AZ. J. Extreme Events, 06, 2050006, https://doi.org/10.1142/S2345737620500062.
Choudhary, E., and A. Vaidyanathan, 2014: Heat stress illness hospitalizations—Environmental public health tracking program, 20 states, 2001–2010. Morb. Mortal. Wkly. Rep., 63, 1–10.
Curriero, F. C., K. S. Heiner, J. M. Samet, S. L. Zeger, L. Strug, and J. A. Patz, 2002: Temperature and mortality in 11 cities of the eastern United States. Amer. J. Epidemiol., 155, 80–87, https://doi.org/10.1093/aje/155.1.80.
Cutter, S. L., B. J. Boruff, and W. L. Shirley, 2003: Social vulnerability to environmental hazards. Soc. Sci. Quart., 84, 242–261, https://doi.org/10.1111/1540-6237.8402002.
Ebi, K. L., and Coauthors, 2018: Human health. Impacts, Risks, and Adaptation in the United States: Fourth National Climate Assessment, Vol. II, D. R. Reidmiller et al., Eds., U.S. Global Change Research Program, 539–571, https://doi.org/10.7930/NCA4.2018.CH14.
Esplin, E. D., J. R. Marlon, A. Leiserowitz, and P. D. Howe, 2019: “Can you take the heat?” Heat-induced health symptoms are associated with protective behaviors. Wea. Climate Soc., 11, 401–417, https://doi.org/10.1175/WCAS-D-18-0035.1.
Famiglietti, J. S., A. Cazenave, A. Eicker, J. T. Reager, M. Rodell, and I. Velicogna, 2015: Satellites provide the big picture. Science, 349, 684–685, https://doi.org/10.1126/science.aac9238.
Finucane, M. L., P. Slovic, C. K. Mertz, J. Flynn, and T. A. Satterfield, 2000: Gender, race, and perceived risk: The “white male” effect. Health Risk Soc., 2, 159–172, https://doi.org/10.1080/713670162.
Franzke, C. L. E., and H. Torelló i Sentelles, 2020: Risk of extreme high fatalities due to weather and climate hazards and its connection to large-scale climate variability. Climatic Change, 162, 507–525, https://doi.org/10.1007/s10584-020-02825-z.
Gasparrini, A., and Coauthors, 2015: Mortality risk attributable to high and low ambient temperature: A multicountry observational study. Lancet, 386, 369–375, https://doi.org/10.1016/S0140-6736(14)62114-0.
Gelman, A., and J. Hill, 2007: Data Analysis Using Regression and Multilevel/Hierarchical Models. Cambridge University Press, 648 pp.
Gill, J. C., and B. D. Malamud, 2014: Reviewing and visualizing the interactions of natural hazards. Rev. Geophys., 52, 680–722, https://doi.org/10.1002/2013RG000445.
Goldberger, A. S., 1962: Best linear unbiased prediction in the generalized linear regression model. J. Amer. Stat. Assoc., 57, 369–375, https://doi.org/10.1080/01621459.1962.10480665.
Gronlund, C. J., A. Zanobetti, J. D. Schwartz, G. A. Wellenius, M. S. O’Neill, and M. S. O’Neill, 2014: Heat, heat waves, and hospital admissions among the elderly in the United States, 1992–2006. Environ. Health Perspect., 122, 1187–1192, https://doi.org/10.1289/ehp.1206132.
Grothmann, T., and F. Reusswig, 2006: People at risk of flooding: Why some residents take precautionary action while others do not. Nat. Hazards, 38, 101–120, https://doi.org/10.1007/s11069-005-8604-6.
Harlan, S. L., A. J. Brazel, L. Prashad, W. L. Stefanov, and L. Larsen, 2006: Neighborhood microclimates and vulnerability to heat stress. Soc. Sci. Med., 63, 2847–2863, https://doi.org/10.1016/j.socscimed.2006.07.030.
Harlan, S. L., J. H. Declet-Barreto, W. L. Stefanov, and D. B. Petitti, 2013: Neighborhood effects on heat deaths: Social and environmental predictors of vulnerability in Maricopa County, Arizona. Environ. Health Perspect., 121, 197–204, https://doi.org/10.1289/ehp.1104625.
Hass, A. L., and K. N. Ellis, 2019: Using wearable sensors to assess how a heatwave affects individual heat exposure, perceptions, and adaption methods. Int. J. Biometeor., 63, 1585–1595, https://doi.org/10.1007/s00484-019-01770-6.
Hass, A. L., J. D. Runkle, and M. M. Sugg, 2021: The driving influences of human perception to extreme heat: A scoping review. Environ. Res., 197, 111173, https://doi.org/10.1016/j.envres.2021.111173.
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.
Hess, J. J., S. Saha, and G. Luber, 2014: Summertime acute heat illness in U.S. emergency departments from 2006 through 2010: Analysis of a nationally representative sample. Environ. Health. Perspect., 122, 1209–1215, https://doi.org/10.1289/ehp.1306796.
Hofmann, D. A., 1997: An overview of the logic and rationale of hierarchical linear models. J. Manage., 23, 723–744, https://doi.org/10.1177/014920639702300602.
Howe, P. D., J. R. Marlon, X. Wang, and A. Leiserowitz, 2019: Public perceptions of the health risks of extreme heat across US states, counties, and neighborhoods. Proc. Natl. Acad. Sci. USA., 116, 6743–6748, https://doi.org/10.1073/pnas.1813145116.
IPCC, 2012: Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation. C. B. Field et al., Eds., Cambridge University Press, 582 pp., https://www.ipcc.ch/report/managing-the-risks-of-extreme-events-and-disasters-to-advance-climate-change-adaptation/.
IPCC, 2014: Summary for policymakers. Climate Change 2014: Impacts, Adaptation, and Vulnerability, Part A: Global and Sectoral Aspects, C. B. Field et al., Eds., Cambridge University Press, 32 pp., https://www.ipcc.ch/site/assets/uploads/2018/02/ar5_wgII_spm_en.pdf.
IPCC, 2021: Climate Change 2021: The Physical Science Basis. Cambridge University Press, in press.
Jeon, S., C. J. Paciorek, and M. F. Wehner, 2016: Quantile-based bias correction and uncertainty quantification of extreme event attribution statements. Weather Climate Extremes, 12, 24–32, https://doi.org/10.1016/j.wace.2016.02.001.
Johnson, D. P., J. S. Wilson, and G. C. Luber, 2009: Socioeconomic indicators of heat-related health risk supplemented with remotely sensed data. Int. J. Health Geogr., 8, 57, https://doi.org/10.1186/1476-072X-8-57.
Johnson, D. P., A. Stanforth, V. Lulla, and G. Luber, 2012: Developing an applied extreme heat vulnerability index utilizing socioeconomic and environmental data. Appl. Geogr., 35, 23–31, https://doi.org/10.1016/j.apgeog.2012.04.006.
Jones, B., B. C. O’Neill, L. McDaniel, S. McGinnis, L. O. Mearns, and C. Tebaldi, 2015: Future population exposure to US heat extremes. Nat. Climate Change, 5, 652–655, https://doi.org/10.1038/nclimate2631.
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.
Kellens, W., R. Zaalberg, T. Neutens, W. Vanneuville, and P. De Maeyer, 2011: An analysis of the public perception of flood risk on the Belgian Coast. Risk Anal., 31, 1055–1068, https://doi.org/10.1111/j.1539-6924.2010.01571.x.
Khare, S., S. Hajat, S. Kovats, C. E. Lefevre, W. B. de Bruin, S. Dessai, and A. Bone, 2015: Heat protection behaviour in the UK: Results of an online survey after the 2013 heatwave. BMC Public Health, 15, 878, https://doi.org/10.1186/s12889-015-2181-8.
Klinenberg, E., 2003: Heat Wave: A Social Autopsy of Disaster in Chicago. 2nd ed. University of Chicago Press, 320 pp.
Knuth, D., D. Kehl, L. Hulse, and S. Schmidt, 2013: Risk perception, experience, and objective risk: A cross-national study with European emergency survivors. Risk Anal., 34, 1286–1298, https://doi.org/10.1111/risa.12157.
Kovats, R. S., and S. Hajat, 2008: Heat stress and public health: A critical review. Annu. Rev. Public Health, 29, 41–55, https://doi.org/10.1146/annurev.publhealth.29.020907.090843.
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.
Lefevre, C. E., W. B. de Bruin, A. L. Taylor, S. Dessai, S. Kovats, and B. Fischhoff, 2015: Heat protection behaviors and positive affect about heat during the 2013 heat wave in the United Kingdom. Soc. Sci. Med., 128, 282–289, https://doi.org/10.1016/j.socscimed.2015.01.029.
Lehner, F., and T. F. Stocker, 2015: From local perception to global perspective. Nat. Climate Change, 5, 731–734, https://doi.org/10.1038/nclimate2660.
Li, D., and E. Bou-Zeid, 2013: Synergistic interactions between urban heat islands and heat waves: The impact in cities is larger than the sum of its parts. J. Appl. Meteor. Climatol., 52, 2051–2064, https://doi.org/10.1175/JAMC-D-13-02.1.
Li, Y., A. L. Hughes, and P. D. Howe, 2021: Toward win-win message strategies: The effects of persuasive message content on retweet counts during natural hazard events. Wea. Climate Soc., 13, 487–502, https://doi.org/10.1175/WCAS-D-20-0039.1.
Lindell, M. K., and S. N. Hwang, 2008: Households’ perceived personal risk and responses in a multihazard environment. Risk Anal., 28, 539–556, https://doi.org/10.1111/j.1539-6924.2008.01032.x.
Liss, A., R. Wu, K. K. H. Chui, and E. N. Naumova, 2017: Heat-related hospitalizations in older adults: An amplified effect of the first seasonal heatwave. Sci. Rep., 7, 39581, https://doi.org/10.1038/srep39581.
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.
Mazdiyasni, O., and Coauthors, 2017: Increasing probability of mortality during Indian heat waves. Sci. Adv., 3, e1700066, https://doi.org/10.1126/sciadv.1700066.
Medina-Ramón, M., A. Zanobetti, D. P. Cavanagh, and J. Schwartz, 2006: Extreme temperatures and mortality: Assessing effect modification by personal characteristics and specific cause of death in a multi-city case-only analysis. Environ. Health Perspect., 114, 1331–1336, https://doi.org/10.1289/ehp.9074.
Mimura, N., R. S. Pulwarty, I. Elshinnawy, M. H. Redsteer, H. Q. Huang, J. N. Nkem, and R. A. Sanchez Rodriguez, 2014: Adaptation planning and implementation. Climate Change 2014: Impacts, Adaptation, and Vulnerability, Part A: Global and Sectoral Aspects, C. B. Field et al., Eds., Cambridge University Press, 869–898.
Mora, C., and Coauthors, 2017: Global risk of deadly heat. Nat. Climate Change, 7, 501–506, https://doi.org/10.1038/nclimate3322.
Noble, I. R., S. Huq, Y. A. Anokhin, J. A. Carmin, D. Goudou, F. P. Lansigan, B. Osman-Elasha, and A. Villamizar, 2014: Adaptation needs and options. Climate Change 2014: Impacts, Adaptation, and Vulnerability, Part A: Global and Sectoral Aspects, C. B. Field et al., Eds., Cambridge University Press, 833–868.
Ortman, J. M., V. A. Velkoff, and H. Hogan, 2014: An aging nation: The older population in the United States . U.S. Census Bureau Rep. P25-1140, 28 pp., https://www.census.gov/content/dam/Census/library/publications/2014/demo/p25-1140.pdf.
Peacock, W. G., S. D. Brody, and W. Highfield, 2005: Hurricane risk perceptions among Florida’s single family homeowners. Landscape Urban Plann., 73, 120–135, https://doi.org/10.1016/j.landurbplan.2004.11.004.
Ratnam, J. V., S. K. Behera, S. B. Ratna, M. Rajeevan, and T. Yamagata, 2016: Anatomy of Indian heatwaves. Sci. Rep., 6, 24395, https://doi.org/10.1038/srep24395.
Reid, C. E., M. S. O’Neill, C. J. Gronlund, S. J. Brines, D. G. Brown, A. V. Diez-Roux, and J. Schwartz, 2009: Mapping community determinants of heat vulnerability. Environ. Health Perspect., 117, 1730–1736, https://doi.org/10.1289/ehp.0900683.
Reid, C. E., and Coauthors, 2012: Evaluation of a heat vulnerability index on abnormally hot days: An environmental public health tracking study. Environ. Health Perspect., 120, 715–720, https://doi.org/10.1289/ehp.1103766.
Renn, O., 1998: The role of risk perception for risk management. Reliab. Eng. Syst. Saf., 59, 49–62, https://doi.org/10.1016/S0951-8320(97)00119-1.
Robine, J.-M., S. L. K. Cheung, S. Le Roy, H. Van Oyen, C. Griffiths, J.-P. Michel, and F. R. Herrmann, 2008: Death toll exceeded 70 000 in Europe during the summer of 2003. C. R. Biol., 331, 171–178, https://doi.org/10.1016/j.crvi.2007.12.001.
Robinson, G. K., 1991: That BLUP is a good thing: The estimation of random effects. Stat. Sci., 6, 15–32, https://doi.org/10.1214/ss/1177011926.
Robinson, P. J., 2001: On the definition of a heat wave. J. Appl. Meteor., 40, 762–775, https://doi.org/10.1175/1520-0450(2001)040<0762:OTDOAH>2.0.CO;2.
Schellnhuber, H. J., 1999: ‘Earth system’ analysis and the second Copernican revolution. Nature, 402, C19–C23, https://doi.org/10.1038/35011515.
Schmeltz, M. T., G. Sembajwe, P. J. Marcotullio, J. A. Grassman, D. U. Himmelstein, and S. Woolhandler, 2015: Identifying individual risk factors and documenting the pattern of heat-related illness through analyses of hospitalization and patterns of household cooling. PLOS ONE, 10, e0118958, https://doi.org/10.1371/journal.pone.0118958.
Semenza, J. C., C. H. Rubin, K. H. Falter, J. D. Selanikio, W. D. Flanders, H. L. Howe, and J. L. Wilhelm, 1996: Heat-related deaths during the July 1995 heat wave in Chicago. N. Engl. J. Med., 335, 84–90, https://doi.org/10.1056/NEJM199607113350203.
Semenza, J. C., D. J. Wilson, J. Parra, B. D. Bontempo, M. Hart, D. J. Sailor, and L. A. George, 2008: Public perception and behavior change in relationship to hot weather and air pollution. Environ. Res., 107, 401–411, https://doi.org/10.1016/j.envres.2008.03.005.
Slovic, P., 1987: Perception of risk. Science, 236, 280–285, https://doi.org/10.1126/science.3563507.
Smit, B., and J. Wandel, 2006: Adaptation, adaptive capacity and vulnerability. Global Environ. Change, 16, 282–292, https://doi.org/10.1016/j.gloenvcha.2006.03.008.
Smith, K., 2013: Environmental Hazards: Assessing Risk and Reducing Disaster. 6th ed. Routledge, 504 pp.
Smith, T. T., B. F. Zaitchik, and J. M. Gohlke, 2013: Heat waves in the United States: definitions, patterns and trends. Climatic Change, 118, 811–825, https://doi.org/10.1007/s10584-012-0659-2.
Stafoggia, M., and Coauthors, 2006: Vulnerability to heat-related mortality: A multicity, population-based, case-crossover analysis. Epidemiology, 17, 315–323, https://doi.org/10.1097/01.ede.0000208477.36665.34.
Tierney, K., 2014: The Social Roots of Risk: Producing Disasters, Promoting Resilience. Stanford University Press, 318 pp.
Tomlinson, C. J., L. Chapman, J. E. Thornes, and C. J. Baker, 2011: Including the urban heat island in spatial heat health risk assessment strategies: A case study for Birmingham, UK. Int. J. Health Geogr., 10, 42, https://doi.org/10.1186/1476-072X-10-42.
Tuholske, C., K. Caylor, C. Funk, A. Verdin, S. Sweeney, K. Grace, P. Peterson, and T. Evans, 2021: Global urban population exposure to extreme heat. Proc. Natl. Acad. Sci. USA., 118, e2024792118, https://doi.org/10.1073/pnas.2024792118.
Uejio, C. K., O. V. Wilhelmi, J. S. Golden, D. M. Mills, S. P. Gulino, and J. P. Samenow, 2011: Intra-urban societal vulnerability to extreme heat: The role of heat exposure and the built environment, socioeconomics, and neighborhood stability. Health Place, 17, 498–507, https://doi.org/10.1016/j.healthplace.2010.12.005.
U.S. EPA, 2006: Excessive heat events guidebook. EPA Rep. 430-B-06-005, 52 pp., https://www.epa.gov/sites/default/files/2016-03/documents/eheguide_final.pdf.
U.S. EPA and CDC, 2016: Climate change and extreme heat: What you can do to prepare . EPA Rep. 430-R-16-061, 20 pp., https://www.epa.gov/sites/default/files/2016-10/documents/extreme-heat-guidebook.pdf.
van der Pligt, J., 1996: Risk perception and self-protective behavior. Eur. Psychol., 1, 34–43, https://doi.org/10.1027/1016-9040.1.1.34.
Vicedo-Cabrera, A. M., and Coauthors, 2021: The burden of heat-related mortality attributable to recent human-induced climate change. Nat. Climate Change, 11, 492–500, https://doi.org/10.1038/s41558-021-01058-x.
Vose, R., D. R. Easterling, K. Kunkel, and M. Wehner, 2017: Temperature changes in the United States. Climate Science Special Report: Fourth National Climate Assessment, Vol. I, D. J. Wuebbles et al., Eds., U.S. Global Change Research Program, 185–206.
Wachinger, G., O. Renn, C. Begg, and C. Kuhlicke, 2013: The risk perception paradox–implications for governance and communication of natural hazards. Risk Anal., 33, 1049–1065, https://doi.org/10.1111/j.1539-6924.2012.01942.x.
Weber, S., N. Sadoff, E. Zell, and A. de Sherbinin, 2015: Policy-relevant indicators for mapping the vulnerability of urban populations to extreme heat events: A case study of Philadelphia. Appl. Geogr., 63, 231–243, https://doi.org/10.1016/j.apgeog.2015.07.006.
White-Newsome, J. L., B. N. Sánchez, E. A. Parker, J. T. Dvonch, Z. Zhang, and M. S. O’Neill, 2011: Assessing heat-adaptive behaviors among older, urban-dwelling adults. Maturitas, 70, 85–91, https://doi.org/10.1016/j.maturitas.2011.06.015.
White-Newsome, J. L., B. Ekwurzel, M. Baer-Schultz, K. L. Ebi, M. S. O’Neill, and G. B. Anderson, 2014: Survey of county-level heat preparedness and response to the 2011 summer heat in 30 U.S. States. Environ. Health Perspect., 122, 573–579, https://doi.org/10.1289/ehp.1306693.