The practice of naming winter storms has generated a large amount of discussion within the meteorology community of late. While storm naming has typically been reserved for tropical systems, some media organizations in the United States recently began naming winter storms but oftentimes using differing criteria. Anecdotal comments have labeled this practice as a marketing initiative and other forecasting organizations have criticized The Weather Channel for naming storms (Palmer), but little to no research has investigated whether naming winter storms serves useful to forecasters, practitioners, and the general public. The purpose of this study is twofold. First, the hope is to further the discussion and investigation of naming winter storms. This study provides empirical evidence that suggests that little difference exists between individual perceptions dependent on whether a name is used or the type of name used. The results indicate that individuals do not differ in levels of perceived severity or susceptibility toward a fictional winter storm dependent on the type of name used. Similarly, perceptions of the credibility of media organizations do not change dependent on the storm name. Second, this study discusses the implications of the results with respect to the current storm naming process and provide future areas of exploration, which can further an understanding of the practice.
The process of naming storm systems is not novel to the meteorology community. Naming hurricanes and tropical systems in the North Atlantic and eastern Pacific basins is a practice that the National Hurricane Center (NHC) and the National Weather Service (NWS)—previously the Weather Bureau—have implemented for more than half a century. One of six unique lists of names, managed by the World Meteorological Organization (WMO), is used in rotation for a given Atlantic hurricane season and recycled 6 years later (National Hurricane Center 2017). Storms are named in alphabetical order, starting with the letter A, with respect to time of formation in the respective basin. Initially, storms were named for easier communication between ships and other coastal entities requiring the information. Prior to this, latitude and longitude information was used as an identifier for which storm was being discussed.
Similarly, the Institute for Meteorology of the Free University Berlin began naming both low-pressure and high-pressure systems in 1954, using female names for lows and male names for highs, to track synoptic weather systems more easily (Institut für Meteorologie 2017). The Buffalo NWS forecast office has also named past storms after constellations, famous scientists, and breeds of cow (see Lake Storm Aphid; Poloncarz 2010). More recently, NWS has named storms in a post hoc fashion, and media organizations have even created online polls (Freedman 2010), which has resulted in a sitting president acknowledging the name (Snowmageddon; AP, 2010). Unfortunately, little social science research has investigated the practicality, usefulness, acceptability, or perceptions regarding using naming systems for winter storms.
After a 2011 Halloween storm impacted the East Coast and was dubbed “Snowtober” by numerous media outlets, The Weather Channel (TWC) began informally reviewing the public safety impact and awareness of storm naming. In 2012, TWC began to officially name winter storms in hopes of generating more public awareness about the storms. New technology platforms, such as Twitter, allow for quick access to storm information (e.g., searching using a Twitter hashtag) and can aid in providing winter storm warning coverage to impacted populations (Palmer 2013). Criteria used for naming a winter storm is implemented based on the Integrated Meteorological Population and Area Calculation Tool (IMPACT). IMPACT calculates the population and area to be impacted by winter storm forecast, based on thresholds set by the NWS for winter storm warnings. The IMPACT dictates that storm names are given when winter storm warnings associated with the event will cover at least 2 million people or 400 000 km2 [see Nizol (2014) for a full explanation of the naming criterion].
While the United Kingdom’s Met Office and Ireland’s Met Eireann have also decided to name midlatitude cyclones, the NWS has refrained from acknowledging winter storm names because “a winter storm’s impact can vary from one location to another, and storms can weaken and redevelop, making it difficult to define where one ends and another begins” (Palmer 2013). This could ultimately create confusion when storms are referred to using different naming strategies. Additionally, a common anecdotal critique of the naming system by meteorologists is the trivial nature of the names, which differ from common European names such as the WMO lists. Some have called for more organizational synergy in the naming process to help it work most effectively (Fritz 2015).
There is much debate in the field of meteorology as to whether naming winter storms is a marketing initiative or if the idea helps individuals find and utilize information about a weather event. The common critique of the naming system TWC employs is how it chooses to use unique sounding names that people rarely encounter on a day-to-day basis (Palmer 2013; Fritz 2015; Greenfield 2013). Some argue that this strategy trivializes the naming process, as winter storm names should mimic those in the NHC list that are familiar to people in hopes of encouraging storm preparation rather than distracting them from it. Previous literature has shown that users will tweet in a humorous manner more than tweeting pertinent information, when using a hashtag including a winter storm name such as “#Nemo,” compared to a localized additional hashtag such as “#BoSnow” (Lachlan Spence Lin Najarian and Greco 2014).
The field of meteorology has progressed to incorporate social science insights into forecasting and communicating whenever possible. A recent example of incorporating social science into the field of meteorology is the Storm Surge Risk Communication Project by the NHC, which underwent years of empirical testing (via survey and experimental methodologies) before it was made available for public access. More recently, researchers have called for forecasters to test how their messages are received and interpreted—especially in risk situations (Morrow et al. 2015)—to better understand the cognitive and emotional responses that can facilitate preventative behavioral intentions prior to a risk event [see Witte (1992) for types of responses]. A similar appeal for using risk communication research to develop and examine weather risk messages has also been voiced in the literature (Demuth et al. 2012; Lindell and Brooks 2013; National Research Council 2006, 2010; UCAR 2012).
This research evaluates the effectiveness of storm naming strategies from a social science perspective. It examines whether the type of storm name can influence how individuals perceive the severity of the storm, individual potential impact, and the credibility of the media organization presenting the information. The study drew from the constructs of the Health Belief Model (HBM) and the construct of source credibility. HBM proposes that numerous factors—such as perceptions of severity, susceptibility, benefits, and barriers to behavior—play a key role in the interpretation of related risk messages as well as the likelihood of desired response to that message (Becker 1974). It is a framework that has been utilized in a variety of health, risk, and environmental communication contexts, including recycling behaviors (Lindsay and Strathman 1997) and perceptions of climate change (Semenza et al. 2011). The present study utilized two HBM constructs, perceived susceptibility and perceived severity, in order to evaluate the effectiveness of storm naming strategies in motivating behavior. Perceived susceptibility is the appraisal of an individual’s vulnerability to a specific risk, such as an impending winter storm; perceived severity is one’s appraisal of the seriousness of a specific risk, such as the negative impacts of an impending winter storm (Becker 1974). Individuals are unlikely to take action toward a weather risk unless they understand the forecast, believe it is personally relevant to them, and puts them at risk (see Baker et al. 2012; Villegas et al. 2013).
Prior research suggests that credibility inductions may be important in terms of motivating people toward taking action in the face of impeding threats (Trumbo and McComas 2003). To date, however, little if any research has examined the impact of specific naming strategies on source credibility and other compliance-related responses. Source credibility is the image or attitude that an individual holds toward another individual or organization in terms of the information stemming from that source (McCroskey and Teven 1999). In terms of evaluating source credibility in social media environments, mock tweets have been commonly used as stimulus materials in research investigating the perceived credibility of risk messages stemming from government agencies, expert individuals, and those retweeted by various entities. While in a more naturalistic environment, Twitter content exists alongside other tweets and as part of a longer timeline, numerous experimental studies have demonstrated that using single tweets as stimuli are effective in capturing immediate responses, such as perceptions of credibility or compliance (see Edwards et al. 2014; Lin et al. 2016; Spence et al. 2016). For example, Spence et al. (2015) took an actual tweet from the Centers for Disease Control and Prevention (CDC) and manipulated the time sent to understand if recency would influence credibility perceptions.
For the current experiment, TWC was the source of the mock tweet, since they are the largest media entity that currently names winter storms. Twitter was chosen as the dissemination platform, since TWC develops storm names with particular care for hashtags and social media communication (Norcross 2016). In addition, approximately two-thirds of Twitter users use the platform to obtain news (Barthel and Shearer 2015), and thus it is reasonable to presume that people seeking weather news would utilize Twitter and encounter a message from TWC, which includes a winter storm name hashtag. To that end, we investigated whether perceived source credibility of TWC will be greater in the control message or in one of the named conditions (RQ1). We also explore whether perceived severity and perceived susceptibility toward winter storms will vary across naming strategies (RQ2). Finally, we explore whether different naming strategies will elicit favorable or unfavorable reactions to the storm names (RQ3).
Undergraduate students (M = 19.2 years old, SD = 1.15) enrolled at a large, northeastern public university in the United States were recruited to participate in April 2015 (N = 407). Participants reviewed and agreed to an Institutional Review Board (IRB)-approved consent sheet before proceeding to the research task. First, participants reported their general media usage and the frequency of checking the weather forecast before being randomly assigned to one of three experimental conditions. These three conditions included a control condition with no storm name, a condition with a common European name (Bill; drawn from a recent NHC Atlantic hurricane name list), and a condition using a winter storm name from a recent TWC name list (Zelus; TWC 2014–15 winter storm name list). These names were chosen because no significant recent events occurred, which used these names. Participants in the control condition saw a mock Tweet posted by TWC, which read, “Up to 1 FOOT of #SNOW for parts of New England as the Winter Storm tracks NE this weekend. Expect travel delays” (Fig. 1). Participants in the two named conditions read a mock Tweet posted by TWC, which read, “Up to 1 FOOT of #SNOW for parts of New England as Winter Storm [#Bill or #Zelus] tracks NE this weekend. Expect travel delays” (Figs. 2 and 3). Mock tweets were modeled off previous TWC tweets for the New England area that were posted during the 2014–15 winter season.
After seeing one of the three tweets, participants were asked to indicate their reaction to the storm name by responding to the statement “I think the name [Bill/Zelus] presents something that is…” using seven paired adjectives (e.g., ridiculous versus reasonable, unrealistic versus realistic) anchored on a semantic differential scale. Since this item was developed for this study, the factor structure of the original seven items was checked by performing a principal components’ exploratory factor analysis. A one factor solution emerged, with two unreliable items that did not load on the primary factor. These items were removed, and the scale was subject to a second factor analysis. In the new 5-item scale, one single factor emerged (eigenvalue = 3.22), and all of the factor loadings were above 0.7. The scores from the five, remaining paired adjectives were averaged into an averaged storm name reaction variable (Cronbach’s alpha = 0.83). A list of all items included in the storm name reaction variable are included in Table 1.
To assess the source credibility of TWC, participants were asked to respond to the statement “please indicate your impression of TWC…” using an 18-item semantic differential scale (e.g., untrustworthy versus trustworthy, unethical versus ethical) that describes three source credibility dimensions—trustworthiness, goodwill, and competence—adapted from McCroskey and Teven (1999). All three credibility dimensions were reliable, with Cronbach’s alpha values ranging from 0.82 to 0.94. All items are listed in Table 1. Participants also indicated how much they disagree or agree with eight winter storm–related questions that reflect their risk perception (on a 7-point scale, adapted from previous HBM research). The eight total items were respectively merged and averaged to create a 4-item perceived severity variable and a 4-item perceived susceptibility variable. Both items had Cronbach’s alpha values at 0.8 or greater. All of the items for the severity and susceptibility variables are presented in Table 1. Participants also answered a set of demographic questions, including age, gender, ethnicity, and political affiliation, before being thanked for their time and completing the survey.
To examine the first two research questions, analysis of variance (ANOVA) tests were conducted across the three experimental conditions. There were no significant differences in any of the three source credibility factors across conditions. Though not statistically significant, those who were in the control condition (M = 2.99, SD = 1.12) seemed to manifest greater trust toward TWC [F (2407) = 1.74, n.s. (where n.s. = not significant)], compared to the Bill (M = 2.83, SD = 1.06) and Zelus conditions (M = 2.93, SD = 1.18). Similar nonsignificant results emerged for goodwill [F (2, 407) = 1.85, n.s.], with the control condition appearing to have the highest goodwill toward TWC (M = 3.66, SD = 0.86) compared to the Bill message (M = 3.50, SD = 0.97) and Zelus message (M = 3.55, SD = 0.96). Finally, the control condition (M = 2.75, SD = 1.10) also reported greater perceived competence in TWC [F (2407) = 1.01, n.s.] compared to the Bill (M = 2.66, SD = 1.07) and Zelus conditions (M = 2.63, SD = 1.11).
Regarding research question two, those who were in the control condition reported without statistical significance the lowest perceived severity (M = 4.66, SD = 0.98) of winter storms [F (2405) = 0.36, n.s.] compared to those who were in the Bill (M = 4.73, SD = 1.08) and Zelus conditions (M = 4.70, SD = 1.12). The susceptibility variable provided different results [F (2405) = 2.04, n.s.], with those in the Bill condition (M = 4.94, SD = 1.09) reporting the strongest perceived susceptibility toward winter storms compared to those in the control (M = 4.85, SD = 1.09) and Zelus conditions (M = 4.77, SD = 1.18).
To analyze reactions to the storm names for the third research question, an independent sample’s Student’s t test was conducted to compare the reaction of those who saw Bill condition to the reaction of those who saw the Zelus condition. Results indicated a slightly less favorable reaction [t (279) = 0.294, n.s.] to the name Zelus (M = 3.59, SD = 0.87) compared to those who were in the Bill condition (M = 3.62, SD = 0.85). No analysis was performed using the control condition for RQ3 because the participants did not have a storm name for which they could react.
Finally, post hoc analyses were performed to analyze any potential differences between the control condition and both named conditions collapsed together (i.e., no name versus using any name). An independent sample’s Student’s t test was again conducted to compare the perceptions of credibility, severity, and susceptibility for these two conditions. Again, storm name reactions were not included as an outcome variable for this analysis, as those who were exposed to the control condition were not asked storm name reaction questions. Similar, nonsignificant differences emerged for the dependent variables of interest. First, those who were in the control condition (M = 2.75, SD = 1.10) felt higher levels of competence toward TWC [t (408) = 0.89, n.s.] than those who were in either of the storm name conditions (M = 2.65, SD = 1.09). Similarly, those in the control condition manifested higher goodwill (M = 3.66, SD = 0.86) toward TWC [t (408) = 1.39, p = 0.09] than those who were in the storm name conditions (M = 3.53, SD = 0.96). An analogous pattern of results exists for perceptions of trustworthiness as well [t (408) = 0.93, n.s.]. Those in the control condition perceived higher trustworthiness of TWC (M = 2.99, SD = 1.12) than those in the storm name conditions (M = 2.88, SD = 1.12). For both severity and susceptibility, the pattern of results flipped. Those who were exposed to the control condition (M = 4.85, SD = 1.09) reported lower perceptions of severity [t (406) = −0.15, n.s.] than those who saw a storm name (M = 4.86, SD = 1.14). Likewise, those who were in the control condition (M = 4.66, SD = 0.98) had lower perceptions of susceptibility [t (406) = −0.49, n.s.] than those who saw a storm name (M = 4.72, SD = 1.10).
4. Study implications
Developing messaging and forecasting communication strategies by incorporating relevant social science research whenever possible is crucial to forecasts or messages being properly considered by the intended audience. According to a recent Pew survey, approximately two-thirds of Twitter users utilize the platform to obtain news (Barthel and Shearer 2015). As communication tools and media technologies continue to advance in the digital age, users have the ability to enter or select a location (or done automatically for them) via an interactive interface (screen) and are provided with a plethora of weather information. They then can supplement this information by easily accessing other information via the web or social media. There exists a need to better understand how individuals seek out information in a timely fashion, such as searching for information via a hashtagged storm name, to better allow forecasters to reach their intended audience. Also, understanding if consumption of weather information affects perceived credibility of the source—as well as weather risk perceptions and risk preparedness behaviors—is also essential to improving weather communication effectiveness for TWC, TV stations, NWS, and the like. This knowledge is especially important when there is an impending severe weather situation—where short-term credibility judgements or other perceptions may be more important in decision-making—and could influence hazards to humans and their communities.
Overall, the results offer an initial exploration into credibility and threat perceptions when experimentally testing a winter storm naming strategy. Regarding research question one, little difference emerged in perceptions of credibility depending on which condition an individual viewed. For all three factors of source credibility, the pattern of means indicated that those exposed to the control condition had higher perceptions of competence, goodwill, and trustworthiness than either of the named conditions. Similarly, the post hoc analysis follows this pattern as well. It should be noted that overall, the responses for source credibility averaged at about the midpoint of the scale (means ranging from 2.75 to 3.66 on a 5-point Likert scale), but participants perceived a higher amount of goodwill in TWC than they felt about the organizations level of competence or trustworthiness. Future research may be interested in teasing out why individuals feel that the organization has goodwill toward the public but is not seen as competent or trustworthy in this case.
Further examining research question 2, participants had overall high levels of susceptibility and severity, as the means were above 4.5 on a 7-point Likert scale. This indicates that there may have been a ceiling effect for these two risk variables. Generally speaking, the audience expects that winter storms can be severe, and they are susceptible to them occurring, which may be an artifact of the sample’s geographic location. It is also noteworthy that the highest means existed in the condition that used a common European storm name, Bill, rather than the less familiar name Zelus. This may suggest that individuals are more likely to perceive a storm to be more severe or themselves more susceptible if the storm is named akin to how other storms are named (i.e., using common names, rather than uncommon names), but additional research is needed to further test this idea. For example, one could compare different styles of names or naming strategies or investigate this phenomenon using a real-world risk by utilizing mock tweets and interactive Twitter pages that post on more than one instance.
Future research should also address the importance of credibility in the warning/naming agency and whether naming decisions influence the credibility of a respected weather organization compared to an unfamiliar or even fictitious weather organization. For example, how might these results change if a government agency like the NWS is the warning agency rather than a media organization? There also exists a need to address the potential confusion caused by the use of multiple names from multiple sources under threat conditions. The degree to which these naming strategies impact responses across weather events of varying magnitude and severity may also be worthy of consideration.
While utilizing a student sample was a limitation of this study, it nevertheless offers some interesting insight in this particular case. Since college-aged individuals and young adults are the largest users of social media—approximately 90% of them use social media (Perrin 2015)—it could be anticipated that they would be the largest group of users who are tweeting about the winter storm names. The lack of significant difference with regard to the variables studied here indicates that the students were not necessarily taken aback by a humorous or uncommon name such as Zelus nor did they associate greater storm risk perception with such names. Additionally, nearly two-thirds of Twitter users are utilizing the platform as a source of news information (Barthel and Shearer 2015), suggesting that the platform may also be utilized as a source for weather information. Future research should seek to address potential sample selection bias by sampling from diverse populations when continuing this line of research. Since placing participants into risky situations and collecting survey responses is oftentimes unethical, exposure is often limited and may differ from information consumption during an actual impending weather risk.
It should also be noted that despite the lack of statistically significant findings, there was some movement in the pattern of means across varying conditions. Participants were asked to evaluate the impact of the storm names through exposure to a single stimulus, and it may be the case that this manipulation was fairly weak in the context of long standing attitudes and dispositions toward weather events of this kind. It may be fruitful for future research to explore how the presentation of storm-related information embedded within a timeline of other Twitter content may influence outcomes. Similarly, studying these variables through some type of field experiment taking place in real time, with a real-life event and controlling for standing attitudes and dispositions, would also be beneficial to better understanding new technology usage for weather information. Valuable future research may seek to expand methodologies to require a participant to engage throughout multiple stages of a forecast or mock weather scenario with certain fixed particulars (housing type, simplified task, etc.) to better understand more realistic choices, behaviors, and perceptions related to their scenario.
Overall, the results and discussion forwarded here are not to justify or debunk the practice of naming winter storms. Instead, they are presented to further the discussion of testing the usefulness and impact of naming winter storms. Additional explorations of this process are necessary to better understand any implications that storm naming may have on generating risk perceptions, behavioral intentions, or perceptions of hazard or affective expressions.
The authors thank the four anonymous reviewers for their comments and feedback on this manuscript. The authors have no funding declarations.