1. Introduction
In the United States, the National Weather Service (NWS) operates a network of approximately 155 ground-based weather radars, with dozens of additional ground-based weather radars owned and operated by the Federal Aviation Administration and television stations (Weber et al. 2007). Weather radar is an essential remote sensing tool for meteorologists to observe and analyze evolving local and regional weather phenomena and to communicate with members of the public when areas of precipitation and associated weather hazards occur. Radar maps and visual displays began to appear in television broadcasts in the 1960s, and by the 1980s it was commonplace for U.S. broadcast meteorologists to display and explain weather radar images and animations to their audiences (Henson 2010; Whiton et al. 1998). The explosion of Internet and mobile technologies since the 1990s provides opportunities for individuals to easily access and view weather radar data and displays, often without the benefit of a meteorologist to aid in interpreting the output. There are now hundreds of meteorological and broadcast news websites and applications that include weather radar displays, allowing individuals to perform their own personal nowcasts (Mass 2012; Zabini 2016). The NWS also provides an interactive web radar display (www.weather.gov/Radar) allowing users to access all 155 ground-based NWS radars.
There has been little research, however, to understand the extent to which the public uses radar, what they use it for, how they interpret it, and the quality of outcomes after consulting radar data. A recent National Academies of Sciences, Engineering, and Medicine (2018) report on the importance of social and behavioral science for the weather enterprise noted the need for research to understand how people are interested in, access, and interpret weather information, as well as how new technologies and ways of communication impact people’s weather interpretations and decisions. We aim to fill this knowledge gap, complementing recent research efforts focused on more effective presentation and communication of forecast and weather hazard information via the web and mobile devices in the United States (Bryant et al. 2014; Casteel 2016; Demuth et al. 2011, 2013; Drost et al. 2016; Klockow 2013; Lazo et al. 2009; Miran et al. 2017; Perreault et al. 2014). We also build upon recent work that explored how professionals—operational meteorologists, emergency managers, and airplane pilots—interpret weather radar data in the course of their respective duties (Baumgart et al. 2008; Heinselman et al. 2015; LaDue et al. 2010; League et al. 2010; Wiggins 2014) by shifting the focus to members of the public. In this study, we focus specifically on users of the NWS website.
The goal of this paper is to uncover several factors that are of significant national and regional importance for the perceived usefulness of the NWS website’s weather radar map display, as rated by website users. We analyze individual, climatological, and infrastructural factors using a large sample of responses from the 2014 NWS customer satisfaction survey and contextualize our findings using the risk information seeking and processing literature. In section 2, we review the relevant factors for this study that potentially influence the usefulness ratings of the NWS website weather radar map display. We describe the NWS survey data and analysis methodologies in section 3, present results of the analyses in section 4, discuss the implications of the findings in section 5, and conclude in section 6 by suggesting potential paths for future research.
2. Factors influencing the usefulness ratings of the NWS website radar display
In this study, we employ a conceptual framework (Fig. 1) that is adapted from Dunwoody and Griffin’s (2015) risk information seeking and processing (RISP) conceptual model, Ryan and Deci’s (2000) self-determination theory, and Earle’s (2010) consensus model of credibility to understand how useful the publicly available radar map display is to NWS website users. The first section of this literature review describes intrinsic and extrinsic factors that may motivate an individual to seek a weather radar map display. The second section discusses the accessibility of weather radar maps, which is conceptualized in terms of personal and infrastructural accessibility factors, while the third section covers the credibility of the source of radar maps including the two components of trust and confidence. Individuals’ cognitive processing and comprehension of risk information is also a key component of the RISP model (Dunwoody and Griffin 2015), but because of limitations inherent to the survey data in this study, we focus only on the motivational, accessibility, and credibility aspects and how they potentially influence usefulness of the NWS website radar map display.
Conceptual framework for the perceived usefulness of the NWS radar display website, adapted from Dunwoody and Griffin’s (2015) RISP conceptual model, Ryan and Deci’s (2000) self-determination theory, and Earle’s (2010) consensus model of credibility.
Citation: Weather, Climate, and Society 10, 4; 10.1175/WCAS-D-17-0108.1
a. Motivation to view weather radar map displays
According to the RISP model (Dunwoody and Griffin 2015), the crux of motivation for risk information seeking is individuals’ perceptions of whether they possess sufficient information on which to decide to take action or not. In this study, we build toward a more comprehensive understanding of motivation in risk information seeking by incorporating self-determination theory (Deci et al. 2017; Ryan and Deci 2000). Self-determination theory encompasses both intrinsic and extrinsic aspects of motivation. Intrinsic factors are associated primarily with a person’s need to feel competent and autonomous. Dunwoody and Griffin’s (2015) key motivational concept of risk information sufficiency can be considered as one aspect of competency, together with an individual’s self-efficacy beliefs. Autonomy is the extent to which a person feels free to choose whether and how to act in a given situation (Deci et al. 2017). Autonomy is also an important concept for motivation in risk information seeking contexts (Pavey and Sparks 2008), including whether people have an internal locus of control, and the feeling that they have the ability to influence outcomes and avoid negative consequences (Trumbo et al. 2016). Intrinsic motivation for risk information is influenced by many factors including sociocultural identities, personality traits, values, and worldviews (Dunwoody and Griffin 2015; Kahan 2012; Leiserowitz 2006; Sjoberg 2000), as well as the salience of the risk context and an individual’s lifestyle and personal interests.
In this study using a secondary survey data source, we do not have explicit measures to capture the full range of important factors related to autonomy, self-efficacy, competency, sociocultural identities, personality traits, or individual values and worldviews. Rather, we use available demographic data as correlates, which offer a glimpse into these complex intrinsic motivational factors. We focus on two particular demographic factors—gender and education level—as well as weather salience, hazard preparedness, and information insufficiency. In the risk communication and perception literature, women are often found to be more proactive in seeking risk information and acting to reduce risks, whereas men are found to be more likely to downplay risks and expend less effort in gathering relevant risk information (e.g., Davidson and Freudenburg 1996; Lindell and Hwang 2008; Peacock et al. 2005; Schumann et al. 2018; Stewart et al. 2012). According to Demuth et al.’s (2011) analyses, women use weather forecasts more frequently than men in the contexts of knowing about the weather, knowing how to dress for the weather, and planning leisure activities. However, studies frequently do not account for evolving gender norms and identities relative to other critical individually intersecting identities such as race/ethnicity or economic class (Enarson et al. 2007; Kahan 2012).
The influence of education level on risk information seeking is likewise tenuous. Whereas some previous research indicated that persons with lower education levels may be motivated to seek risk information because of a larger gap between current knowledge levels and the level of knowledge sufficiency to respond to a risk appropriately (Griffin et al. 1999), other research suggests that higher education levels are associated with greater risk information seeking (Gutteling and de Vries 2017). Recent research suggests lower education levels are associated with greater importance for temperature and reduced importance for precipitation (Demuth et al. 2011).
The final intrinsic motivational factors most relevant for this study are weather salience, weather hazard preparedness, and information insufficiency. Weather salience is the degree of importance an individual gives to weather or changes in the weather (Stewart 2009; Stewart et al. 2012). The salience of the weather in a person’s life may vary according to personal interests, values, and experiences. Thus, weather salience captures important factors related to intrinsic motivation to seek information such as risk tolerance and relevant hazard experience (Dunwoody and Griffin 2015). In this study, we infer weather salience from variables capturing the frequency of visiting the NWS website and whether a respondent has a safety plan for hazardous weather. The latter variable includes the concept of hazard preparedness. Typically, preparedness is conceptualized as an outcome of risk information seeking (as in Kellens et al. 2012), but in the context of seeking to view weather radar maps the causal chain will be reversed. Our reasoning for this reversal is that seeking to view weather radar maps does not indicate to a person how to formulate a safety plan, but may be a tool to help decide whether or not to implement an existing safety plan and when to do so. This conceptualization is similar to how information flow impacts decision-making in Lindell and Perry’s (2012) protective action decision model (PADM). Information insufficiency may also serve as a motivating factor for seeking risk information via radar maps. While there is not an exact instrument to measure information insufficiency, in this study it is partially represented by whether respondents report being confused about the difference between a watch and a warning. Radar displays often include watch and/or warning information that can serve to remind or inform users of their respective meanings and implications for decision-making related to weather hazards. This behavior is consistent with how respondents should attempt to remedy an information gap as theorized in the RISP model (Dunwoody and Griffin 2015). Because of the highly specific context and wording of this variable, the relationship will likely not be as strong as if other information insufficiency variables were available, such as trying to supplement visual, local observations with mapped data to better comprehend spatial–temporal characteristics of precipitation and related hazards, or to gain more information on the intensity or type of nearby precipitation events than can be inferred from a simple probabilistic forecast (e.g., there is a 40% of rain this afternoon).
Extrinsic motivational factors vary from those that have become integrated into a person’s intrinsic motivation to those that compel action through regulation or material incentive (Deci et al. 2017). Examples of extrinsic motivational aspects relevant for weather risk information seeking are social cues (Bean et al. 2015; Wood et al. 2018), environmental cues (Dewitt et al. 2015; Lindell and Perry 2012), and hazard or disaster subcultures (Bankoff 2017; Burnside et al. 2007; Schumann et al. 2018). Extrinsic motivational aspects are particularly influential for risk information seeking when they appeal to the need for a person to feel inclusion as part of a group by following informational subjective norms (Deci et al. 2017; Dunwoody and Griffin 2015). In this study, we focus on hazard or disaster subcultures as potentially influential in seeking information about precipitation and precipitation-related weather hazards. These subcultures arise as part of regional and community identities from the collective experiences and cultural interpretations of certain types of hazards and disaster events as they occur over time in different places (Bankoff 2017). While we do not have an exact instrument to measure the importance of hazard subcultures, we can use climatological data to differentiate between persons living in locations that experience, on average, more or less frequent and intense precipitation events and other related hazards such as lightning and damaging winds. We assume in this study that there will be greater motivation to seek weather information via radar maps in places where it is considered normal to frequently be concerned and informed about precipitation and associated weather hazards as indicated by climatological data and the frequency of weather warnings.
b. Accessibility of weather radar maps
Following Dunwoody and Griffin’s (2015) RISP model, an additional component that acts in concert with motivational aspects to influence weather risk information seeking via radar maps is an individual’s beliefs about her/his capacity to access and gather the relevant risk information. Accessibility of online weather radar maps may be facilitated or hindered by both infrastructural and personal factors (Fig. 1). In terms of infrastructure, radar data are inaccessible or of reduced quality in some portions of the United States because of gaps in the weather radar network or because of blockage of radar signals due to topographic or anthropogenic features (Diederich et al. 2015; Maddox et al. 2002). Likewise, access to telecommunication technologies—particularly home and mobile Internet—is generally much more limited in rural than in urban locations (Cutter et al. 2016; Salemink et al. 2017).
In addition to the geographic limitations of infrastructure, there are individual factors that can limit or deny access to weather information via radar maps. Despite the overall increase in telecommunications technologies over the past 30 years, there remain persons in the United States who do not have home or mobile Internet subscriptions or do not have a computer or mobile device, often because of cost considerations and occasionally because of personal choices (Campos-Castillo 2015; Warf 2013). Access also pertains to whether one knows how to operate computers and mobile devices and is aware of the existence of risk information. Even if people have an Internet-connected device, they may not realize that radar data can be freely viewed online courtesy of the NWS and television stations. This lack of awareness may be related to lack of salience of precipitation information or to language barriers for citizens who are not proficient English speakers. Individuals who are able to frequently and repeatedly view radar maps are likely to be more familiar with display options and over time may gain knowledge and experience to better utilize radar data. There are several methods by which individuals may access weather information. Because our current study focuses on the usefulness of online weather radar maps, we naturally expect those who tend to access all types of weather information via computer or mobile devices will rate online radar maps as more useful. Lazo et al.’s (2009) survey indicated that few of their participants used NWS web pages or mobile devices to obtain weather information. The newer data used in our research will show an increased role of computers and mobile devices in accessing weather information, as these were the channels of interest in the NWS survey. However, the importance of television as a method for obtaining weather information, and the medium’s relationship with the usefulness of online radar maps, should not be dismissed. Broadcast meteorologists serve an important role as interpreters of weather information (Demuth et al. 2009), and in places that frequently experience thunderstorm hazards, local television stations are a critical source of information including weather radar data (Sherman-Morris 2013; Sutter 2013). Therefore, even persons who more often access weather information via television may find online radar maps to be familiar and useful.
c. Credibility of the National Weather Service as a source of radar maps
The third component from Dunwoody and Griffin’s (2015) RISP model that works together with motivation and accessibility to influence risk information seeking is composed of the beliefs held by an individual about the sources or channels through which they are receiving or will receive the risk information. One of the most important beliefs about sources/channels for communication of risk information is the level of credibility or trustworthiness (Dunwoody and Griffin 2015; Peters et al. 1997). Similar to Renn and Levine (1991), we conceptualize credibility or trustworthiness as a sociocultural form of currency that may accumulate or diminish over time. We also apply the consensus model of credibility or trustworthiness (Earle 2010) as being a function of both relational trust and confidence. Relational trust is based primarily on social heuristics and affect (Earle 2010). For example, individuals often place more trust in other individuals or organizations with whom they perceive to share closely held values or beliefs (Siegrist et al. 2000). Similarly, a simple affective notion such as a feeling of familiarity can enhance relational trust (Earle 2010). Confidence, on the other hand, primarily stems from systematic evaluation (Earle 2010). Positive past experiences of risk communication in which an individual or organization has proven to be accurate and reliable will bolster confidence. Risk information will therefore be seen as credible to the extent that it is perceived to be communicated with honorable intentions and professional competency.
Despite the persistent cultural trope that weather forecasts are unreliable, users of weather risk information obtained from the NWS in the United States generally rate its credibility quite high. Lazo et al.’s (2009) online survey of NWS forecast users demonstrated that a very high percentage (over 95%) of people in the United States seek weather forecast information on a daily basis, and many obtain information multiple times per day. They concluded that weather forecasts and information are perhaps the most consumed and consistently familiar forms of scientific information among the American public. Furthermore, their survey showed very high satisfaction ratings for NWS forecasts in general, and users also expressed high confidence in short-range weather forecasts out to 2 days. In their study of risk communication across five wildfire events, Steelman et al. (2015) noted that official maps were rated as one of the most trustworthy and useful among all sources of information available to the public. A similar high level of credibility and usefulness would therefore be expected for online maps and displays of weather radar provided by the NWS, as radar displays may serve to temper cynical or distrustful skepticism (as in Poortinga and Pidgeon 2003) about precipitation forecasts through the updating of expectations in real-time by viewing weather radar data.
3. Data and methods
a. Survey data
The data analyzed in this study are from a national customer satisfaction survey conducted by Claes Fornell International (CFI) Group in 2014 for NOAA and the NWS.1 It was administered to users of the NWS website (weather.gov), with the primary goal of improving the usefulness of NWS web services and products (forecasts, observations, radar, satellite, etc.). The survey consisted of 41 questions and covered topics such as familiarity with the NWS, hazardous weather services, decision-support services, and weather education and outreach. Participants clicked on a link provided by the NWS on its national, regional, and local websites. A total of 31 307 people took the survey.
For this analysis, we are primarily interested in survey respondents who use the NWS website for personal reasons. The survey included a question in which respondents indicated how they use information provided by the NWS. Thus, using this survey question, we created a subset of respondents who indicated they use information from the NWS for personal reasons (such as determining weather-appropriate attire), for outdoor recreation, or as weather enthusiasts. We excluded respondents who indicated that they used NWS information solely for purposes related to their work responsibilities. However, respondents who indicated multiple uses in addition to personal, recreational, and as weather enthusiasts were included. We also limited the analysis to the contiguous United States. After these filtering procedures, the total number of respondents in the dataset was 24 379, with 8341 in the central NWS region, 6645 in the eastern region, 5141 in the southern region, and 4252 in the western region (see Fig. 2 for map of NWS regions). Though the sample used in this study is large and geographically diverse, it is subject to selection bias and does not represent a random sample representative of the general population of the contiguous United States.
1) Dependent variable
The dependent variable is the respondents’ usefulness ratings of the NWS website Doppler radar display. The survey question asked participants, “Using a 1 to 10 scale, where 1 means ‘not at all useful’ and 10 means ‘very useful,’ please rate the usefulness of the NWS weather.gov website on Doppler radar display.” Across the contiguous United States, 58.4% of respondents found the radar display to be very useful (rating of 10) (Fig. 3). Respondents who rated the display either a 9 or 10 make up 72% of the sample. The dependent variable is therefore skewed to the left and does not approximate a normal distribution. The distribution of the dependent variable is similar when broken down by the NWS regions as well (Fig. 3). Radar usefulness was not explicitly defined within the NWS customer satisfaction survey.
Distribution of respondents in each radar usefulness category (%) for the contiguous United States and each NWS region.
Citation: Weather, Climate, and Society 10, 4; 10.1175/WCAS-D-17-0108.1
2) Independent variables
Based on the relevant literature, we chose several variables from the NWS survey to test their relationships with the usefulness ratings of the NWS website radar display. In our regression models, there are 21 independent variables (Table 1), 12 of which are directly from the 2014 NWS customer satisfaction survey. The remaining nine independent variables were derived from a variety of sources (Table 1). Several variables were also calculated by the authors using the software ArcGIS. The variables are organized into three categories from our conceptual model (Fig. 1)—motivational factors, accessibility factors, and credibility factors—to facilitate presentation of the results and subsequent discussion.
Variables utilized in this study, including data sources, measurement levels, and original units, response levels, and hypothesized relationship with the dependent variable.
Descriptive statistics for the independent variables provide context for interpretation of the results presented in section 4 (Table 2). In terms of demographics, approximately 30% of respondents in the analysis marked their gender as female and 70% self-identified as male. Survey respondents reported high levels of education with about 57% having earned at least a bachelor’s degree. Only about 9% reported having never attended college. While race/ethnicity data were collected in the NWS customer satisfaction survey, this variable is not used because there is insufficient prior evidence to hypothesize about statistical relationships with usefulness of the NWS website radar display. Additional demographic factors such as age and income were either not collected or were not made available from the NWS survey.
Independent variable mean values, with standard deviations in parentheses, for the contiguous United States and each NWS region. Values other than means are indicated in column 1.
b. Analysis methodologies
1) Geospatial analysis
The individual survey responses were geocoded to the centroid of each respondent’s reported zip code. The responses were then overlaid with a lattice of 351 hexagons with 100-km side lengths, and the average radar display usefulness was computed using all geocoded responses that fell within each respective hexagon. In addition to mapping the average radar usefulness ratings using the hexagon aggregation, we also performed an optimized Getis-Ord Gi* hot-spot analysis using the software ArcGIS. The hot-spot analysis identified locations characterized by statistically significant geographic clustering of high and/or low average radar usefulness ratings (Getis and Ord 1992; Ord and Getis 1995). The optimization functionality suggested a fixed distance band of 411.7 km for the analysis; this means that the average usefulness rating for each hexagon was compared to the average value for all nearby hexagons within 411.7 km. The number of hexagons reported as statistically significant was adjusted to account for the multiple testing problem and for spatial dependence (Caldas de Castro and Singer 2006).
2) Regression analyses
We employed ordinal logistic regression models for our data analyses for the contiguous United States and for each of the four NWS regions (Fig. 2). Using the Statistical Package for the Social Sciences (SPSS), version 23, we fit ordinal logistic proportional odds models using a cumulative logit link function. When modeling interval measurement-level data (such as our dependent variable), researchers may choose to employ either linear or ordinal models (Agresti 2012). In preliminary analyses, the ordinal regression models outperformed a linear regression model (results not shown). Therefore, we chose to treat our dependent variable as ordinal for better model fit. Because the dependent variable was also heavily skewed to the left, we utilized a generalized linear model methodology that is appropriate for nonnormal data (Agresti 2012). Except for the four binary variables, all independent variables were standardized to z scores prior to input into each model in order to facilitate comparison of the model beta coefficients and odds ratios. Model significance is reported using the likelihood ratio chi-square statistic, although our primary interest in the regression models is to investigate the significance of the independent variables and interpret their relationships with respect to the dependent variable. We present and discuss regional results in addition to the national results due to sociocultural, infrastructural, and climatological variations across the contiguous United States.
4. Results
a. Geospatial analysis of radar map display usefulness ratings
The dependent variable was mapped in order to visualize geographic variation within the contiguous United States (Fig. 4). Overall, usefulness ratings for the NWS Doppler radar display web page were very high. However, respondents in some areas of the United States found it more useful than others. Using a Getis-Ord hot-spot analysis, locations where higher usefulness ratings were significantly clustered (hot spots) appeared in the southeastern United States with a secondary significant cluster in parts of Kansas and Oklahoma (Fig. 5). In addition to the hot spots, several areas where lower usefulness ratings were significantly clustered (cold spots) exist across the Pacific Northwest, Northern California, and parts of Idaho, Montana, and Wyoming.
Average usefulness ratings of the NWS website radar display across the contiguous United States.
Citation: Weather, Climate, and Society 10, 4; 10.1175/WCAS-D-17-0108.1
Hot- and cold-spot analysis of the NWS website radar display usefulness ratings for the contiguous United States.
Citation: Weather, Climate, and Society 10, 4; 10.1175/WCAS-D-17-0108.1
b. Contiguous U.S. regression results
1) Motivational factors
According to the likelihood ratio chi-square test, the contiguous U.S. model was statistically significant with a p value of <0.001 [χ2 = 3460.11, with 21 degrees of freedom (df); Table 3]. The most significant intrinsic motivational factor in the model was the frequency in which respondents reported visiting the weather.gov website; the relationship was positive such that a greater frequency of visits to the website corresponded to higher ratings of radar map display usefulness. Both demographic variables were also statistically significant. Survey participants who reported a higher level of education were likely to rate the NWS radar display to be slightly less useful. Women were almost 18% more likely than men to rate the NWS radar display as less useful. In terms of preparedness, those who reported having a safety plan for hazardous weather were 15% more likely to give a higher usefulness rating for the NWS radar display. In terms of information insufficiency, respondents who reported being confused about the difference between a watch and a warning were 18% more likely to give a higher rating for radar map display usefulness.
Contiguous U.S. ordinal logistic regression results with dependent variable of NWS website radar display usefulness. All variables have been standardized to z scores except for those noted in parentheses as binary variables. Statistical significance of the Wald chi-square value is given by asterisks: * indicates p < 0.1, ** indicates p < 0.05, and *** indicates p < 0.01. Model significance is as follows: χ2 = 3460.11, 21 df, p value < 0.001, and N = 24 379.
Of the six climatological variables representing extrinsic motivation via the amount of precipitation and frequency of related hazards, only half were statistically significant in the contiguous U.S. model (Table 3). The average annual lightning flash rate at each respondents’ zip code centroid had a significant and strong positive relationship with NWS radar display usefulness. However, significant negative relationships were observed when considering the average annual amount of precipitation and frequency of flash flood warnings for the respondents’ zip code centroids. Though these two precipitation-related variables were statistically significant, their effect sizes were quite small compared to lightning flash rate and the intrinsic motivational factors. The remaining three climatological variables were not found to be statistically significant across the contiguous United States.
2) Accessibility factors
All six accessibility variables—three personal and three infrastructural—were significant predictors of radar map display usefulness. Of the three technological methods respondents reported on regarding their personal access to general weather information, two were highly significant predictors of radar usefulness ratings (Table 3). Respondents who reported using a computer, television, or mobile device more frequently to access weather information were likely to rate the NWS radar map display as more useful. However, the effect size for using a mobile device was quite small compared to the effect sizes for computer and television. The infrastructural accessibility factors were all significant in the model but were less important predictors of NWS radar map display usefulness than the personal access methods of computer and television. Survey respondents who lived in zip codes that are more urbanized were likely to rate the NWS radar display as less useful. In relation to the geography of the NWS Doppler radar network, a greater number of radars within 150 km of a respondent’s zip code centroid—accounting for topographic obstruction of the radar beam—was associated with a greater likelihood of a higher overall usefulness rating. However, NWS radar map display ratings were likely to be lower if the distance from the individual’s zip code centroid was greater to the nearest physical NWS radar site.
3) Credibility factors
The three most significant variables in the contiguous U.S. model were related to credibility of the NWS as the source of radar data (Table 3). Respondents who were more likely to take action based on information they received from the NWS were likely to rate the NWS radar display more useful, and this variable explained the most variance within the contiguous U.S. model. Those who were satisfied with the NWS efforts to explain the difference between an advisory, a watch, and a warning were also more likely to rate the NWS radar display more useful. The level of familiarity a respondent reported with the NWS was positively related to radar map display usefulness ratings. However, those who reported that they knew that warnings are issued by the federal government were likely to give lower ratings for radar map display usefulness.
c. Regional ordinal regression results
1) Motivational factors
The ordinal regression models are statistically significant in each of the four NWS regions according to the likelihood ratio chi-square tests (Tables 4 and 5). Consistent with the findings from the contiguous U.S. model, the level of school completed was associated with lower NWS radar display usefulness ratings in all regions. This negative relationship was somewhat stronger in the eastern and central regions, as indicated by the odds ratios. Gender was statistically significant in every region except the southern region, still with a negative relationship. Frequency of visiting weather.gov was statistically significant with a positive relationship with radar map usefulness in all regions, although it appeared to be of somewhat less importance in the eastern and western regions. Also in the eastern and western regions, respondents with a safety plan for hazardous weather were likely to rate the NWS radar map display as more useful than those without a plan; this preparedness indicator was of lesser importance in the southern and central regions. Finally, positive significant relationships between confusion about the difference between a watch and warning and radar map web page usefulness were observed in the southern and central regions.
Ordinal logistic regression results for eastern and southern NWS regions. Table formatting and reporting conventions as in Table 3. For the eastern region, model significance is as follows: χ2 = 942.26, 21 df, p value < 0.001, and N = 6645. For the southern region, model significance is as follows: χ2 = 683.47, 21 df, p value < 0.001, and N = 5141.
Ordinal logistic regression results for central and western NWS regions. Table formatting and reporting conventions as in Table 3. For the central region, model significance is as follows: χ2 = 1135.28, 21 df, p value <0.001, and N = 8341. For the western region, model significance is as follows: χ2 = 543.49, 21 df, p value <0.001, and N = 4252.
Regional results for the extrinsic motivational factors exhibited marked differences from the results for the contiguous United States. Whereas the average annual lightning flash rate was the most significant climatological variable in the national analysis, it was significant only in the western region and had a positive relationship with map usefulness. The average annual frequency of flash flood warnings was significant only in the eastern and central regions and had a negative relationship with radar map usefulness. The average annual amount of precipitation was statistically significant in the central and western regions only, but with opposite relationships with the dependent variable. In the central region, results indicated that more precipitation received on an annual basis corresponded to greater usefulness of the NWS radar display. Higher average annual precipitation in the western region was associated with lower radar map usefulness ratings. The remaining climatological variables were not significant in any of the four regions, just as in the contiguous U.S. model.
2) Accessibility factors
The frequencies of accessing weather information using a computer and television were positively associated with NWS radar display ratings in every region (Tables 4 and 5). In the central and eastern regions, computer was more significant than television, but the order of significance was reversed in the western and southern regions. Results for the infrastructural accessibility factors were less consistent. Respondents who resided in more urbanized zip codes were likely to rate the NWS radar map display as less useful than those who lived in less urbanized zip codes in all regions except the western region. The number of radars not suffering from beam blockage within 150 km of respondents’ zip codes corresponded to higher radar usefulness ratings in the eastern and western regions, yet that variable was not significant in the southern and central regions. Only survey respondents in the western region were likely to rate NWS radar maps less useful when the nearest radar site was farther from their home zip code.
3) Credibility factors
As in the contiguous U.S. model, three of the variables pertaining to credibility factors of the NWS as a source of radar data were among the best predictors of NWS radar map display usefulness ratings in every region (Tables 4 and 5). Respondents who were more likely to take action based on information they receive from the NWS were more likely to find the NWS radar map display useful in each region, and this variable was the most important predictor in each model. Greater familiarity with the NWS and greater satisfaction with NWS efforts to explain the difference between watches and warnings were both significantly associated with higher radar map display usefulness ratings. Finally, respondents in the eastern and southern regions who reported understanding the role of the federal government as the source of weather warnings rated the NWS radar map display as less useful; however, the effect sizes were small.
5. Discussion
This study reveals several significant factors for understanding the perceived usefulness of the NWS radar map display that relate to motivation to view radar map displays, their accessibility, and the credibility of the NWS as source of the radar maps. In terms of intrinsic motivational factors, all five variables were significant in the regression model for the contiguous United States. Our proxy variables to broadly capture respondents’ level of weather salience—the frequency with which the NWS website was visited, and whether or not they reported having a safety plan for hazardous weather—were highly positively associated with NWS website radar map usefulness ratings in the national regression model. The frequency variable was of greater significance in the central and southern NWS regions, whereas the preparedness variable was significant only in the eastern and western regions. The results suggest that individual weather salience was a key motivational factor for viewing the NWS website radar map display. One caveat, however, is that the frequency of visiting the NWS website may also be indicative of greater accessibility and credibility. Future work should seek to clarify these relationships with more focused survey items. Interestingly, respondents who self-reported being confused about the differences between a watch and a warning tended to rate the radar displays as more useful. This may be explained by the presence of warning polygons on the NWS radar display web page, which can assist in remedying information insufficiencies by clarifying the spatial and temporal attributes of warnings for those who might find the watch–warning nomenclature confusing. This variable was not significant in the eastern and western regions, where the watch–warning nomenclature is relevant less often than in the central and southern regions.
The results for the two demographic variables were more difficult to interpret. Respondents who identified as female were more likely to rate the usefulness of the radar displays as lower than male respondents in the national model and in all NWS regions except the southern region. This result is at odds with our expectations from previous literature where women typically are more proactive in seeking risk information in general and for weather forecasts in particular (Demuth et al. 2011; Schumann et al. 2018; Stewart et al. 2012). It is unclear why, in this study, women rated the usefulness of the radar display lower than men did. More targeted research is needed to fully grasp whether there are differences in the use of radar according to gender and why these differences might exist. We also note that while gender was statistically significant in the model, it had relatively low explanatory power, which is in accordance with recent findings from the risk information literature (Nelson 2015; Yang et al. 2014). Education level likewise was found to exhibit a negative relationship with the usefulness ratings of the NWS radar map display in all regression models. This corroborates the findings of Demuth et al. (2011) that lower education levels are associated with greater use of weather information. However, in this study the context is precipitation information, which does not match the breakdown from Demuth et al. (2011) showing lower education levels associated with more frequent seeking of temperature forecast information and less frequent seeking for precipitation. As with the conclusions about gender, future research should seek to clarify the relationship between education level and use of radar among members of the public. Also, the negative relationship found in this study may not generalize to the entire population of the United States since the NWS customer satisfaction survey respondents were more highly educated than would be a nationally representative sample.
Of the six climatological and warning variables we used as proxies for extrinsic motivation in the form of hazard subcultures, only three were significant in the regression model for the contiguous United States. Respondents whose zip codes experienced a higher average annual lightning flash rate were more likely to give higher usefulness ratings for the NWS radar map display. Within the NWS regions, lightning flash rate was not significant with the exception of the western region. Thus, while the national variation in lightning flash rates lends understanding to the usefulness of radar map displays, the variations are not great enough within the other regions to be discriminant. We interpret this result—together with the mapped results of the hot-spot analysis—as broadly indicative of the presence of weather hazard subcultures in portions of the central, southern, and eastern United States in which people are apt to use radar as an information source when there is precipitation associated with thunderstorms. Within the western region, a weather subculture that uses radar maps for information about thunderstorms may exist in Arizona and Utah where thunderstorms occur during the monsoon but likely does not exist in the Pacific Northwest where thunderstorms are relatively infrequent. Given that no official NWS lightning advisory product exists, additional research is required to understand the usefulness of radar map displays for situations in which people are seeking real-time information about lightning.
A somewhat counterintuitive result from the national analysis is that higher average annual precipitation and numbers of flash flood warnings were indicative of lower radar map display usefulness. The former result likely arises because of the higher annual averages in the northwestern United States where precipitation occurs regularly and in less intense bursts than in the central, southern, and eastern United States. The regional results support this interpretation; the western region displayed the same negative relationship between precipitation and NWS radar map display usefulness. The eastern and southern regions showed no significant association between precipitation climatology and radar usefulness, while the central region showed a significant positive relationship. The latter result is likely due to the sharp difference in average annual precipitation between the eastern and western portions of the central region (Arguez et al. 2012; their Fig. 2b on p. 1694). Further evidence for this interpretation is the positive association in the central region between NWS radar usefulness ratings and longitude with greater radar usefulness ratings moving from west to east (Kendall’s tau-b value of 0.045, p value <0.001); this relationship can also be seen visually in the central-region states in Fig. 4. The national result that the number of flash flood warnings was negatively associated with radar usefulness was unexpected. The regional results suggested that the negative relationship was most pronounced in the eastern and central NWS regions, yet there is not a clear causal explanation for this finding at the time of writing.
The second component from our conceptual framework, accessibility of radar map displays, is represented in the analyses by both individual and infrastructural factors. Whereas Lazo et al. (2009) reported relatively low levels of Internet use for seeking forecast information, this study, by the nature of the recruitment via the NWS website, is more illustrative of online information seeking. The individual accessibility variables for computer and television were more significant predictors of NWS radar map usefulness nationally than the motivational variables, with the exception of frequency visiting the NWS website. The frequencies with which individuals accessed weather information via computer and television were both positively associated with NWS radar map display usefulness ratings nationally and by region. The result for computers is expected, as those who frequently access weather information via computer can easily access the NWS radar web page. The latter result for television is likely related to the frequent use and explanation of radar maps on television by broadcast meteorologists (Sutter 2013), contributing to the transformation of radar maps into an accessible and useful tool even for nonexperts. Although the frequency of obtaining weather information via a mobile device was also significant, this was one of the least important variables in the national model. These findings suggest that accessing the mobile version of the NWS radar map web page may be far less useful than when accessed via a computer.
Each of the infrastructural accessibility variables were significant in the national model, but results varied for the regional models. Respondents who lived where a greater number of radar sites provide coverage were likely to give higher usefulness ratings, particularly within the western and eastern regions. Additionally, those who lived farther from the nearest radar site were likely to give lower usefulness ratings, though this finding pertained mainly to the western region. These results suggest—though they exhibited relatively weak (but significant) statistical effects—that the radar network infrastructure does influence lower usefulness ratings in places with lower-quality coverage. Surprisingly, respondents who lived in more urbanized zip codes nationally and across all regions rated the NWS radar map website usefulness lower. It may be that the hypothesized positive infrastructural relationship for radar usefulness in urban areas was captured by the radar network variables, leading to a residual negative urban relationship, which would be explained by an unknown causal mechanism perhaps related to more socioeconomic and cultural motivational factors.
Credibility of the NWS as a source of radar map displays, the third component of risk information seeking considered in this study, was represented by four variables derived from the NWS customer satisfaction survey that pertained to respondents’ familiarity, trust, and confidence in NWS information. Three of the four credibility variables had the most explanatory power of all variables in the national model, with similar results in the regional models. By far the best predictor of higher NWS radar map display usefulness ratings was the likelihood of taking action based on NWS information. This is indicative of the great importance of NWS credibility, and how high confidence in information obtained from the NWS in general spills over to the usefulness of its radar maps as a specific type of information. Two of the other highly significant credibility factors were more indicative of the importance of relational trust. Those who reported being more satisfied with NWS efforts to explain advisory–watch–warning terminology, as well as those who expressed greater familiarity with the NWS, were more likely to give higher ratings for the radar display’s usefulness. Following Earle’s (2010) model of credibility, these variables represent respondents’ relational trust in the NWS, with efforts to explain watch/warning terminology demonstrating care for the public as NWS customers to understand its products, and with familiarity representing an affective dimension of trust. Finally, when participants recognized that weather warnings emanate from the U.S. federal government, they gave lower usefulness ratings for the NWS radar map website. The variable was of small importance compared to most of the others in the regression models. Still, the results may be explained by lower credibility levels that some persons assign to federal government entities in association with political views. Though we do not have data to explicitly demonstrate this causal mechanism, as political views were not measured on the survey, it is consistent with recent findings in similar research contexts (Myers et al. 2017; Steelman et al. 2015).
6. Conclusions
Using the results of our analyses, we can confidently highlight several of the most important conclusions. Geographically, the NWS website radar map display was most useful to those living in the southeastern and south-central United States, and least useful to those living in the northwestern United States. The three most significant variables in the contiguous U.S. model were related to credibility of the NWS as a source of weather information. Specifically, respondents who were more likely to take action based on information they received from the NWS were likely to rate the NWS radar map display more useful, which explained the most variance within the contiguous U.S. model. Those who were satisfied with the NWS efforts to explain differences in terminology, as well as respondents who reported a high level of familiarity with the NWS, were also more likely to rate the NWS radar display as more useful. The frequency of visiting the weather.gov website also corresponded to higher usefulness ratings of the NWS radar map display. The average annual lightning flash rate was the most significant climatological variable. Finally, all six accessibility variables were significant predictors of radar map display usefulness, especially those respondents who obtained weather information via computer or television.
We acknowledge several limitations inherent to our analyses. First, the dependent variable is derived from one question taken from the 2014 NWS customer satisfaction survey. Unfortunately, the NWS survey was not designed to fully understand all of the factors that influence perceived usefulness of a weather radar map display in the context of this study. Also the NWS survey data did not allow for a full quantitative assessment of our conceptual model, as there were no variables to measure comprehension of weather radar data. Another limitation was that radar usefulness was not defined within the NWS customer satisfaction survey, which made analyzing specific aspects of the display that respondents found useful difficult. Furthermore, conclusions about the potential influence of geographic and climatological factors aggregated to the zip code level would be more reliable if they were based on direct answers given about the perceptions of these factors by individual respondents.
As mentioned, a recent National Academies of Sciences, Engineering, and Medicine (2018) report on the importance of social and behavioral science for the weather enterprise noted the need for research to understand how people are interested in, access, and interpret weather information, as well as how new technologies and ways of communication impact people’s weather interpretations and decisions. Future work will look into the perceived value of using weather radar for members of the public and will expand on the conceptual model used in this study. Given the large number of applications that display weather radar, it is imperative for the weather enterprise to fully investigate the scope of public use of weather radar and understand how its value can be maximized to enhance public safety and well-being.
Acknowledgments
The authors wish to thank Rebecca Morss for her helpful feedback on the conceptual framework outlined in this article. They would also like to thank all three anonymous reviewers for their constructive comments and suggestions, as well as the National Weather Service for making this survey data available online. The National Center for Atmospheric Research is sponsored by the National Science Foundation.
REFERENCES
Agresti, A., 2012: Categorical Data Analysis. Wiley, 744 pp.
Arguez, A., I. Durre, S. Applequist, R. S. Vose, M. F. Squires, X. Yin, R. R. Heim Jr., and T. W. Owen, 2012: NOAA’s 1981–2010 U.S. Climate Normals: An overview. Bull. Amer. Meteor. Soc., 93, 1687–1697, https://doi.org/10.1175/BAMS-D-11-00197.1.
Bankoff, G., 2017: Living with hazard: Disaster subcultures, disaster cultures and risk-mitigating strategies. Historical Disaster Experiences: Towards a Comparative and Transcultural History of Disasters across Asia and Europe, G. J. Schenk, Ed., Springer, 45–59, https://doi.org/10.1007/978-3-319-49163-9_2.
Baumgart, L. A., E. J. Bass, B. Philips, and K. Kloesel, 2008: Emergency management decision making during severe weather. Wea. Forecasting, 23, 1268–1279, https://doi.org/10.1175/2008WAF2007092.1.
Bean, H., J. Sutton, B. F. Liu, S. Madden, M. M. Wood, and D. S. Mileti, 2015: The study of mobile public warning messages: A research review and agenda. Rev. Commun., 15, 60–80, https://doi.org/10.1080/15358593.2015.1014402.
Bryant, B., M. Holiner, R. Kroot, K. Sherman-Morris, W. B. Smylie, L. Stryjewski, M. Thomas, and C. I. Williams, 2014: Usage of color scales on radar maps. J. Oper. Meteor., 2, 169–179, http://doi.org/10.15191/nwajom.2014.0214.
Burnside, R., D. S. Miller, and J. D. Rivera, 2007: The impact of information and risk perception on the hurricane evacuation decision-making of greater New Orleans residents. Sociol. Spectrum, 27, 727–740, https://doi.org/10.1080/02732170701534226.
Caldas de Castro, M., and B. H. Singer, 2006: Controlling the false discovery rate: A new application to account for multiple and dependent tests in local statistics of spatial association. Geogr. Anal., 38, 180–208, https://doi.org/10.1111/j.0016-7363.2006.00682.x.
Campos-Castillo, C., 2015: Revisiting the first-level digital divide in the United States. Soc. Sci. Comput. Rev., 33, 423–439, https://doi.org/10.1177/0894439314547617.
Casteel, M. A., 2016: Communicating increased risk: An empirical investigation of the National Weather Service’s impact-based warnings. Wea. Climate Soc., 8, 219–232, https://doi.org/10.1175/WCAS-D-15-0044.1.
Cecil, D. J., D. Buechler, and R. Blakeslee, 2014: Gridded lightning climatology from TRMM LIS & OTD: Dataset description. Atmos. Res., 135–136, 404–414, https://doi.org/10.1016/j.atmosres.2012.06.028.
Cutter, S. L., K. D. Ash, and C. T. Emrich, 2016: Urban–rural differences in disaster resilience. Ann. Amer. Assoc. Geogr., 106, 1236–1252, https://doi.org/10.1080/24694452.2016.1194740.
Davidson, D. J., and W. R. Freudenburg, 1996: Gender and environmental risk concerns. Environ. Behav., 28, 302–339, https://doi.org/10.1177/0013916596283003.
Deci, E. L., A. H. Olafsen, and R. M. Ryan, 2017: Self-determination theory in work organizations: The state of a science. Annu. Rev. Organ. Psychol. Organ. Behav., 4, 19–43, https://doi.org/10.1146/annurev-orgpsych-032516-113108.
Demuth, J. L., B. H. Morrow, and J. K. Lazo, 2009: Weather forecast uncertainty information: An exploratory study with broadcast meteorologists. Bull. Amer. Meteor. Soc., 90, 1614–1618, https://doi.org/10.1175/2009BAMS2787.1.
Demuth, J. L., J. K. Lazo, and R. E. Morss, 2011: Exploring variations in people’s sources, uses, and perceptions of weather forecasts. Wea. Climate Soc., 3, 177–192, https://doi.org/10.1175/2011WCAS1061.1.
Demuth, J. L., R. E. Morss, J. K. Lazo, and D. C. Hilderbrand, 2013: Improving effectiveness of weather risk communication on the NWS point-and-click web page. Wea. Forecasting, 28, 711–726, https://doi.org/10.1175/WAF-D-12-00118.1.
Dewitt, B., B. Fischhoff, A. Davis, and S. B. Broomell, 2015: Environmental risk perception from visual cues: The psychophysics of tornado risk perception. Environ. Res. Lett., 10, 124009, https://doi.org/10.1088/1748-9326/10/12/124009.
Diederich, M., A. Ryzhkov, C. Simmer, P. Zhang, and S. Trömel, 2015: Use of specific attenuation for rainfall measurement at X-band radar wavelengths. Part I: Radar calibration and partial beam blockage estimation. J. Hydrometeor., 16, 487–502, https://doi.org/10.1175/JHM-D-14-0066.1.
Drost, R., M. Casteel, J. Libarkin, S. Thomas, and M. Meister, 2016: Severe weather warning communication: Factors impacting audience attention and retention of information during tornado warnings. Wea. Climate Soc., 8, 361–372, https://doi.org/10.1175/WCAS-D-15-0035.1.
Dunwoody, S., and R. J. Griffin, 2015: Risk information seeking and processing model. The SAGE Handbook of Risk Communication, H. Cho, T. Reimer, and K. A. McComas, Eds., Sage Publications, 102–116.
Earle, T. C., 2010: Trust in risk management: A model-based review of empirical research. Risk Anal., 30, 541–574, https://doi.org/10.1111/j.1539-6924.2010.01398.x.
Enarson, E., A. Fothergill, and L. Peek, 2007: Gender and disaster: Foundations and directions. Handbook of Disaster Research, H. Rodríguez, E. L. Quarantelli, and R. R. Dynes, Eds., Springer, 130–146, https://doi.org/10.1007/978-0-387-32353-4_8.
Getis, A., and J. K. Ord, 1992: The analysis of spatial association by use of distance statistics. Geogr. Anal., 24, 189–206, https://doi.org/10.1111/j.1538-4632.1992.tb00261.x.
Griffin, R. J., S. Dunwoody, and K. Neuwirth, 1999: Proposed model of the relationship of risk information seeking and processing to the development of preventive behaviors. Environ. Res., 80, S230–S245, https://doi.org/10.1006/enrs.1998.3940.
Gutteling, J. M., and P. W. de Vries, 2017: Determinants of seeking and avoiding risk-related information in times of crisis. Risk Anal., 37, 27–39, https://doi.org/10.1111/risa.12632.
Heinselman, P., D. LaDue, D. M. Kingfield, and R. Hoffman, 2015: Tornado warning decisions using phased-array radar data. Wea. Forecasting, 30, 57–78, https://doi.org/10.1175/WAF-D-14-00042.1.
Henson, R., 2010: Weather on the Air: A History of Broadcast Meteorology. American Meteorological Society, 241 pp.
Iowa Environmental Mesonet, 2017: Archived NWS Watch/Warnings. Iowa State University, accessed 12 May 2017, https://mesonet.agron.iastate.edu/request/gis/watchwarn.phtml.
Kahan, D. M., 2012: Cultural cognition as a conception of the cultural theory of risk. Handbook of Risk Theory: Epistemology, Decision Theory, Ethics, and Social Implications of Risk, S. Roeser, et al., Eds., Springer, 725–759, https://doi.org10.1007/978-94-007-1433-5_28.
Kellens, W., R. Zaalberg, and P. De Maeyer, 2012: The informed society: An analysis of the public’s information-seeking behavior regarding coastal flood risks. Risk Anal., 32, 1369–1381, https://doi.org/10.1111/j.1539-6924.2011.01743.x.
Klockow, K. E., 2013: Spatializing tornado warning lead-time: Risk perception and response in a spatio-temporal framework. Doctoral dissertation, University of Oklahoma, 150 pp.
LaDue, D. S., P. L. Heinselman, and J. F. Newman, 2010: Strengths and limitations of current radar systems for two stakeholder groups in the southern plains. Bull. Amer. Meteor. Soc., 91, 899–910, https://doi.org/10.1175/2009BAMS2830.1.
Lazo, J. K., R. E. Morss, and J. L. Demuth, 2009: 300 billion served: Sources, perceptions, uses, and values of weather forecasts. Bull. Amer. Meteor. Soc., 90, 785–798, https://doi.org/10.1175/2008BAMS2604.1.
League, C. E., W. Diaz, B. Philips, E. J. Bass, K. Kloesel, E. Gruntfest, and A. Gessner, 2010: Emergency manager decision-making and tornado warning communication. Meteor. Appl., 17, 163–172, https://doi.org/10.1002/met.201.
Leiserowitz, A., 2006: Climate change risk perception and policy preferences: The role of affect, imagery, and values. Climatic Change, 77, 45–72, https://doi.org/10.1007/s10584-006-9059-9.
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.
Lindell, M. K., and R. W. Perry, 2012: The protective action decision model: Theoretical modifications and additional evidence. Risk Anal., 32, 616–632, https://doi.org/10.1111/j.1539-6924.2011.01647.x.
Maddox, R. A., J. Zhang, J. J. Gourley, and K. W. Howard, 2002: Weather radar coverage over the contiguous United States. Wea. Forecasting, 17, 927–934, https://doi.org/10.1175/1520-0434(2002)017<0927:WRCOTC>2.0.CO;2.
Mass, C., 2012: Nowcasting: The promise of new technologies of communication, modeling, and observation. Bull. Amer. Meteor. Soc., 93, 797–809, https://doi.org/10.1175/BAMS-D-11-00153.1.
Miran, S. M., C. Ling, J. J. James, A. Gerard, and L. Rothfusz, 2017: User perception and interpretation of tornado probabilistic hazard information: Comparison of four graphical designs. Appl. Ergon., 65, 277–285, https://doi.org/10.1016/j.apergo.2017.06.016.
Myers, T. A., J. Kotcher, N. Stenhouse, A. A. Anderson, E. Maibach, L. Beall, and A. Leiserowitz, 2017: Predictors of trust in the general science and climate science research of US federal agencies. Public Understanding Sci., 26, 843–860, https://doi.org/10.1177/0963662516636040.
National Academies of Sciences, Engineering, and Medicine, 2018: Integrating Social and Behavioral Sciences Within the Weather Enterprise. The National Academies Press, 201 pp., https://doi.org/10.17226/24865.
Nelson, J. A., 2015: Are women really more risk-averse than men? A re-analysis of the literature using expanded methods. J. Econ. Surv., 29, 566–585, https://doi.org/10.1111/joes.12069.
Ord, J. K., and A. Getis, 1995: Local spatial autocorrelation statistics: Distributional issues and an application. Geogr. Anal., 27, 286–306, https://doi.org/10.1111/j.1538-4632.1995.tb00912.x.
Pavey, L. J., and P. Sparks, 2008: Threats to autonomy: Motivational responses to risk information. Eur. J. Soc. Psychol., 38, 852–865, https://doi.org/10.1002/ejsp.477.
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.
Perreault, M. F., J. B. Houston, and L. Wilkins, 2014: Does scary matter? Testing the effectiveness of new National Weather Service tornado warning messages. Commun. Stud., 65, 484–499, https://doi.org/10.1080/10510974.2014.956942.
Peters, R. G., V. Covello, and D. B. McCallum, 1997: The determinants of trust and credibility in environmental risk communication: An empirical study. Risk Anal., 17, 43–54, https://doi.org/10.1111/j.1539-6924.1997.tb00842.x.
Poortinga, W., and N. F. Pidgeon, 2003: Exploring the dimensionality of trust in risk regulation. Risk Anal., 23, 961–972, https://doi.org/10.1111/1539-6924.00373.
Renn, O., and D. Levine, 1991: Credibility and trust in risk communication. Communicating Risks to the Public: International Perspectives, R. E. Kasperson and P. J. M. Stallen, Eds., Technology, Risk, and Society, Vol. 4, Kluwer Academic Publishers, 175–218, https://doi.org/10.1007/978-94-009-1952-5_10.
Ryan, R. M., and E. L. Deci, 2000: Self-determination theory and the facilitation of intrinsic motivation, social development, and well-being. Amer. Psychol., 55, 68–78, https://doi.org/10.1037/0003-066X.55.1.68.
Salemink, K., D. Strijker, and G. Bosworth, 2017: Rural development in the digital age: A systematic literature review on unequal ICT availability, adoption, and use in rural areas. J. Rural Stud., 54, 360–371, https://doi.org/10.1016/j.jrurstud.2015.09.001.
Schumann, R. L., III, K. D. Ash, and G. C. Bowser, 2018: Tornado warning perception and response: Integrating the roles of visual design, demographics, and hazard experience. Risk Anal., 38, 311–332, https://doi.org/10.1111/risa.12837.
Sherman-Morris, K., 2013: The public response to hazardous weather events: 25 years of research. Geogr. Compass, 7, 669–685, https://doi.org/10.1111/gec3.12076.
Siegrist, M., G. Cvetkovich, and C. Roth, 2000: Salient value similarity, social trust, and risk/benefit perception. Risk Anal., 20, 353–362, https://doi.org/10.1111/0272-4332.203034.
Sjoberg, L., 2000: Factors in risk perception. Risk Anal., 20, 1–12, https://doi.org/10.1111/0272-4332.00001.
Steelman, T. A., S. M. McCaffrey, A.-L. Knox Velez, and J. A. Briefel, 2015: What information do people use, trust, and find useful during a disaster? Evidence from five large wildfires. Nat. Hazards, 76, 615–634, https://doi.org/10.1007/s11069-014-1512-x.
Stewart, A. E., 2009: Minding the weather: The measurement of weather salience. Bull. Amer. Meteor. Soc., 90, 1833–1841, https://doi.org/10.1175/2009BAMS2794.1.
Stewart, A. E., J. K. Lazo, R. E. Morss, and J. L. Demuth, 2012: The relationship of weather salience with the perceptions and uses of weather information in a nationwide sample of the United States. Wea. Climate Soc., 4, 172–189, https://doi.org/10.1175/WCAS-D-11-00033.1.
Sutter, D., 2013: Broadcast meteorology and the supply of weather forecasts: An exploration. J. Econ. Finance, 37, 463–477, https://doi.org/10.1007/s12197-011-9186-7.
Trumbo, C. W., L. Peek, M. A. Meyer, H. L. Marlatt, E. Gruntfest, B. D. McNoldy, and W. H. Schubert, 2016: A cognitive-affective scale for hurricane risk perception. Risk Anal., 36, 2233–2246, https://doi.org/10.1111/risa.12575.
Warf, B., 2013: Contemporary digital divides in the United States. Tijdschrift voor economische en sociale geografie, 104, 1–17, https://doi.org/10.1111/j.1467-9663.2012.00720.x.
Weber, M. E., J. Y. N. Cho, J. S. Herd, J. M. Flavin, W. E. Benner, and G. S. Torok, 2007: The next-generation multimission U.S. surveillance radar network. Bull. Amer. Meteor. Soc., 88, 1739–1751, https://doi.org/10.1175/BAMS-88-11-1739.
Whiton, R. C., P. L. Smith, S. G. Bigler, K. E. Wilk, and A. C. Harbuck, 1998: History of operational use of weather radar by U.S. weather services. Part I: The pre-NEXRAD era. Wea. Forecasting, 13, 219–243, https://doi.org/10.1175/1520-0434(1998)013<0219:HOOUOW>2.0.CO;2.
Wiggins, M. W., 2014: Differences in situation assessments and prospective diagnoses of simulated weather radar returns amongst experienced pilots. Int. J. Ind. Ergon., 44, 18–23, https://doi.org/10.1016/j.ergon.2013.08.006.
Wood, M. M., D. S. Mileti, H. Bean, B. F. Liu, J. Sutton, and S. Madden, 2018: Milling and public warnings. Environ. Behav., 50, 535–566, https://doi.org/10.1177/0013916517709561.
Yang, Z. J., A. M. Aloe, and T. H. Feeley, 2014: Risk information seeking and processing model: A meta-analysis. J. Commun., 64, 20–41, https://doi.org/10.1111/jcom.12071.
Zabini, F., 2016: Mobile weather apps or the illusion of certainty. Meteor. Appl., 23, 663–670, https://doi.org/10.1002/met.1589.
We accessed and downloaded the survey data in 2015 via a link published in a winter 2014/15 NWS newsletter (https://verification.nws.noaa.gov/content/pm/pubs/peak/2014/2014_Winter_Newsletter.pdf). The 2014 survey data are no longer available via the CFI portal, but the authors will share the data used for this study upon request.
