Abstract

A great deal of research has been conducted regarding tornado warnings and protective actions taken, including some studies in which respondents were presented with hypothetical tornado warning scenarios. Much less research has been conducted in which respondents were presented with tornado watch scenarios, even though they cover a larger area and longer time period, thus potentially disrupting a far greater number of people. To address this lack of research, surveys were used to determine the influence of severe weather watches on planned Saturday afternoon and evening activities away from the immediate vicinity of the respondent’s home. Respondents were presented a hypothetical watch scenario, in which they had some activity planned for later that afternoon or evening. Each respondent rated his or her likelihood to continue an activity depending on the severity of the watch and the length of the activity. Respondents were provided information about each hypothetical watch including duration and primary threats. Responses from the survey indicated that as the severity of the watch or the length of the activity increased, the likelihood of the respondent continuing the activity decreased. For a severe thunderstorm watch, a tornado watch, and a particularly dangerous situation (PDS) tornado watch, 36.1%, 51.2%, and 80.2% of the respondents, respectively, would not continue an activity lasting 30 min or longer.

1. Introduction

In the United States, particularly east of the Rocky Mountains, severe weather can occur at any time of the day and during any month of the year. Severe thunderstorms have the greatest occurrence from Texas to southern Minnesota (NSSL 2018), while tornadoes have the greatest occurrence in Florida, the Great Plains, and the Southeast (NOAA 2018). Severe weather produces a variety of threats, including flooding, lightning, damaging winds, hail, and tornadoes, and as a result, the National Weather Service (NWS) and the Storm Prediction Center (SPC) issue a variety of products to inform the public of the potential for severe weather. Local NWS offices are responsible for issuing severe thunderstorm warnings and tornado warnings for their county warning areas (NWS 2017), while the SPC issues convective outlooks, mesoscale discussions, public severe weather outlooks, and severe weather watches (SPC 2017). On average, 715 severe weather watches (severe thunderstorm and tornado) were issued annually from 2007 to 2016 (an average of 1.96 watches per day). During this period, severe thunderstorm watches were issued more frequently (466 on average) than tornado watches (249 on average) (SPC 2018). Rarer are particularly dangerous situation (PDS) tornado watches (9 on average) (SPC 2018), issued when the confidence of multiple intense tornadoes is high (SPC 2017). Averages were calculated by accessing the SPC severe weather events archive page (SPC 2018) and counting the number of watches that occurred each year. A few watches from each year were not available and were not considered when calculating the averages.

Tornado watches typically cover a large geographic region, around 51 800–103 600 km2 (20 000–40 000 mi2), and last around 6–8 h (SPC 2017), unlike tornado warnings, which have an average size of 999 km2 (385.7 mi2) and 38 min (Harrison and Karstens 2017). Thus, many more people may be under tornado watches, and for a longer period of time. This would suggest that they could potentially impact more people for a longer time if individuals vary their routine when watches are issued. In Ripberger et al. (2014), U.S. Census Bureau data were used to determine that over 6 times as many people were affected by watches than warnings from 25 April 2012 through 11 November 2012. Unlike tornado warnings, however, little is known about how individuals respond to severe thunderstorm or tornado watches. This study sought to fill this gap by determining the influence of severe weather watches (severe thunderstorm watch, tornado watch, and PDS tornado watch) in decision making using a hypothetical experiment involving planned Saturday afternoon and evening activities away from the immediate vicinity of the respondent’s home.

2. Background

Considerable research has been conducted on tornado warning communication, risk perception, and response. Because individuals must first hear or otherwise become aware of the warning message in order to act upon it (Mileti and Sorensen 1990), much of the prior research on actual events has focused on how warnings have been received. Results of 11 studies showed, on average, over three quarters of the people surveyed were aware of a tornado before it hit (Sherman-Morris 2013). Individuals have historically reported becoming aware of the possibility of a tornado through television, sirens, or a combination of the two (Balluz et al. 2000; Legates and Biddle 1999; Hammer and Schmidlin 2002; Mitchem 2003; Paul et al. 2003; Sherman-Morris and Brown 2012). Evidence from more recent events indicates that television is still an important source. Stokes and Senkbeil (2017) found that local television was the most common source of warning information during the Tuscaloosa tornado of 2011 even among the college-age samples, with those who lived in the area over 20 years strongly preferring it. It was also the dominant source in three Oklahoma tornadoes in 2013 (Miran et al. 2018). Sirens also contribute to awareness. Surveyed residents of an Alabama town with a tornado siren were more than 3 times more likely to have received a warning than residents of a town without sirens (Liu et al. 1996). Receipt of tornado watch information has not been analyzed to nearly the same degree as warning information.

Studies have provided a measure of the number of people who were aware that severe weather was possible prior to an event, but frequently without specifying whether or not individuals had received information about a watch. Schumacher et al. (2010) reported only a few individuals interviewed were aware of the possibility of severe weather prior to tornado warnings. Other research indicated 26.7% of students and 40% of faculty knew about the possibility of severe weather on a college campus before a watch was even issued, but only 9.8% of students and 15.5% of faculty first became aware from the tornado watch (Sherman-Morris 2010). The difference in awareness between these two events may be due to meteorological conditions (Schumacher et al. 2010). It is not well known to what level individuals are aware of tornado watches, or to what extent they use watch information in decision making. Individuals report frequently obtaining weather information, with around 80% checking it at least once per day (Lazo et al. 2009; Silver 2015). However, this does not necessarily translate to awareness of severe weather threats. One individual interviewed following an EF5 Mississippi tornado was not aware tornadoes were possible despite getting a weather forecast “multiple times per day” (Sherman-Morris and Brown 2012, p. 97). Attention is necessary for a message to lead to action, and has been a neglected variable in tornado studies (Ripberger et al. 2014). Research using Twitter data supports the idea that tornado watches heighten public attention, but not to the same extent as tornado warnings (Ripberger et al. 2014). Sutter and Simmons (2014, p. 22) found that the presence of a watch, once other factors are controlled, does not influence fatalities. The authors explain this lack of effect by saying that watches may be “too diffuse to prompt immediate action.” However, it is not known whether the information the watch conveys or a difference in awareness between watch and warning would cause this lack of effect.

A number of studies have examined how individuals respond to tornado warnings. While not always the case (e.g., Balluz et al. 2000; Mitchem 2003), research indicates that a majority of individuals who hear a tornado warning do engage in some form of protective behavior (e.g., Hammer and Schmidlin 2002; Tiefenbacher et al. 2001; Paul et al. 2003; Comstock and Mallonee 2005; Chaney and Weaver 2010; Sherman-Morris and Brown 2012). However, the response may not come immediately. Most individuals seek to confirm warning information prior to taking shelter (Drabek and Boggs 1968) often by trying to visually locate the tornado (Tiefenbacher et al. 2001; Chaney and Weaver 2010). In some events, individuals have attempted to drive to a safer location (Hammer and Schmidlin 2002; Sherman-Morris 2010). In responding to an Oklahoma City tornado, only 1% of the respondents who fled their homes chose to do so within 5 min of receiving the warning and 48% waited more than 30 min to leave (Hammer and Schmidlin 2002). Similarly, following a Mississippi tornado warning resulting in the cancellation of college classes, almost a quarter of students left campus rather than sheltering in place, with some indicating they believed it was safe to drive to their safe place (Sherman-Morris 2010).

Very little research has been conducted on tornado watches and their impacts on planned activities (Mason and Senkbeil 2015). Research has indicated a relationship between checking weather when severe weather was possible and avoiding travel (Silver 2015), but this did not specifically focus on tornado watches. Studies that show the importance of confirmatory actions during warnings, combined with research that indicates longer lead times may cause people to delay protective action (Hoekstra et al. 2011), raise questions about the influence of watch information when individuals are carrying out other activities. When warnings do not prompt immediate actions, it is reasonable to expect watches, which are already less urgent than warnings, to have an even more limited effect. As part of the risk identification and risk assessment process, the protective action decision model suggests individuals ask whether there is a real threat they need to pay attention to, and then use information about the likelihood and the immediacy of the threat to determine whether action needs to be taken (Lindell and Perry 2012). Individuals generally understand the difference between tornado watches and warnings (Liu et al. 1996; Legates and Biddle 1999; Balluz et al. 2000; Mitchem 2003), and view tornado warnings as both a more likely as well as a more serious threat than tornado watches (Schultz et al. 2010). Thus, the lack of immediacy and reduced certainty of the watch could lead to no protective action taken.

Tornado risk perception has been studied following actual events as well as scenarios. Studies focusing on actual tornado events have examined how perception of the threat fits into preconceived ideas about where tornadoes can occur (Klockow et al. 2014), and examined the role of environmental cues or reports (e.g., Schumacher et al. 2010). Frequently, the perception of threat is somehow based on the tornado path or the warning polygon. In a study of three Oklahoma tornadoes, odds of taking protective action were higher when one was located within 5 miles of the tornadoes’ paths, which was likely due to the relationship between perceived risk and proximity to the threat (Miran et al. 2018). A more direct relationship has been the focus of several studies of hypothetical tornado warnings. For example, multiple studies have demonstrated that individuals perceive the highest level of risk near the center of a warning polygon, and not closest to the location of the storm when the warning is issued (Ash et al. 2014; Lindell et al. 2016; Sherman-Morris and Brown 2012).

Decision-making experiments using scenarios have been used in a variety of weather contexts to measure the influence of varying levels of severity on protective actions taken. While not a tornado scenario, individuals faced with the decision of whether to bring plants inside when provided a frost forecast were more likely to take the protective action when told the frost would be more severe (Kox and Thieken 2017). In another study, participants playing the role of a plant manager were more likely to make a shelter-in-place decision when provided a warning with any sort of strongly worded impact statement (Casteel 2018). One tornado watch study was found that focused on communicating varying levels of severity through a tornado watch scale (TWS) (Mason and Senkbeil 2015). The TWS altered the standard tornado watch and the PDS tornado watch into six levels (levels 0–5) based on the forecasted severity of the event and also provided suggested protective actions (Mason and Senkbeil 2015). Overall, the TWS prompted safer protective actions when compared to an NWS tornado watch and PDS tornado watch (Mason and Senkbeil 2015). Our study examines whether watches can prompt action in a specific hypothetical decision-making scenario. Based on the influence of severity levels on protective actions as revealed in the previous studies, the authors expected different types of watches would also have different levels of influence on actions in the current study.

3. Methods

An overall hypothesis that severe weather watches have statistically significant effects on planned Saturday afternoon and evening activities away from the immediate vicinity of the respondent’s home was tested using scenarios in which the type of watch was varied. The following three specific hypotheses were tested:

  • H1: Severe weather (severe thunderstorm, tornado, and PDS tornado) watches have a statistically significant effect on planned Saturday afternoon and evening activities away from the immediate vicinity of the respondent’s home.

  • H2: Each increase in watch severity level will show a significantly greater effect on planned Saturday afternoon and evening activities away from the immediate vicinity of the respondent’s home.

  • H3: The likelihood of continuing an activity during a watch will decrease with increasing length of the activity.

Respondents viewed the scenarios in an online survey generated using Qualtrics, a private online survey tool, and distributed via social media. Students, faculty, and meteorologists across the country shared a link to the online survey through Twitter and Facebook. A total of 2176 people participated in this study. Of the 2176 respondents, 2008 provided their zip code. Of the respondents, 99.30% (1994) lived in U.S. climate regions (Karl and Koss 1984) that experience severe weather. The largest percentage of respondents were from the South, the Southeast, and the Ohio Valley with 39% (783), 29.2% (586), and 22.2% (445), respectively. The remaining respondents were located in the upper Midwest, the Northeast, the Southwest, the northern Rockies and Plains, the West, and the Northwest with 3.7% (74), 3.3% (67), 1.1% (21), 0.9% (17), 0.6% (11), and 0.2% (3), respectively. The survey contained 32 questions, including text entry, multiple choice, and Likert-type questions. Thirty-two survey questions were used to better understand each respondent’s 1) weather knowledge, 2) sources of weather information, 3) actions taken during a severe weather watch, and 4) household characteristics. This paper focuses on the effects of severe weather watches on planned Saturday afternoon and evening activities away from the immediate vicinity of the respondent’s home. Other information provided by the respondents will be addressed in a separate paper. The scenarios were designed to determine if an individual would change his or her plans because of the issuance of one of the three severe weather watches. First, respondents were told that each hypothetical watch was issued on a Saturday at 1500 LT, and that they (the respondent) had an activity planned for later that afternoon or evening. The activity was located at least 20 min from the respondent’s home, and shelter opportunities at the activity were not known. Saturday was used as it typically avoids conflicts with school, work, and religious services. Respondents were then provided information about the watch including duration and primary threats (Table 1). The type of severe weather watch (severe thunderstorm, tornado, and PDS tornado) was not explicitly stated. This was to ensure that the respondents would assess the threat based on the text, not by the type of watch. All respondents saw every watch scenario. Using a Likert-type scale, each respondent rated his or her likelihood to continue an activity during the watch. The length of the activity was progressively increased, similar to Kox and Thieken (2017), to determine whether the length of time required for the activity would lead to differences in response or if the length of the activity would interact with the type of watch. For each type of watch, respondents rated activities that lasted 30 min, 1 h, 2 h, 4 h, and the entirety of the watch. The respondent selected ratings ranging from “definitely would continue with activity” to “definitely would not continue with activity.” Our response choices were inspired by Strawderman et al. (2018), whose response choices ranged from “definitely not drive” to “definitely drive” in a hypothetical winter weather scenario (Strawderman et al. 2018). Respondents also rated how likely they would be to monitor the weather in each scenario. Specific questions are provided in Table 2.

Table 1.

Severe weather watch information.

Severe weather watch information.
Severe weather watch information.
Table 2.

Survey questions.

Survey questions.
Survey questions.

SPSS Statistics was used to complete the statistical analysis. The Friedman test calculated the significance of the difference in effect of severe weather watches on planned Saturday afternoon and evening activities away from the immediate vicinity of the respondent’s home from one scenario to another. The same process was used to test for differences within watch categories among different lengths of activities. The Friedman test is the nonparametric alternative to the one-way repeated measures analysis of variance (ANOVA). Since the data were not normally distributed and the dependent variable was ordinal, the Friedman test was used instead of the one-way repeated measures ANOVA.

4. Results

Of the 2176 respondents, 99.3% (2161), 96.8% (2105), and 99.1% (2157) provided their sex, age, and highest level of education completed, respectively. A majority (65.4%, or 1413) of the respondents were female. The mean age of the respondents was 42 with a standard deviation of 13.5. Nearly half (46.8%, or 984) of the respondents were between the ages of 31 and 50; 24.1% (507) were 30 and under, and 29.2% (614) were 51 and over. Nearly a third (32.4%, or 698) of the respondents received a bachelor’s degree and about a quarter (25.6%, or 552) received a graduate degree (master’s or doctorate). Additionally, 21.5% (463) of the respondents received a high school diploma or equivalent, and 15.7% (338) received an associate’s degree. Of the remaining 4.9%, 2.6% (55) of the respondents attended some college, 0.9% (20) attended some high school, 0.9% (19) completed vocational school or received a certificate, and 0.6% (12) could not be classified.

a. Responses per watch category and activity scenario

When provided the severe thunderstorm watch, almost half (47.9%) of the respondents indicated that they “definitely would” or “probably would” continue an activity lasting any duration, while 38.9% “probably would not” or “definitely would not.” A smaller percentage (13.3%) were “not sure” if they would continue the activity. Regarding activities with progressively longer lengths of time, 49.0%, 44.5%, and 32.3% of the respondents stated that they “probably would” continue an activity lasting 30 min, 1 h, or 2 h, respectively. Nearly 26% (25.9%) of the respondents stated that they “probably would not” continue an activity lasting 4 h followed by 25% indicating that they “definitely would not” continue the activity. For an activity lasting the entire duration of the severe thunderstorm watch, 29.2% of the respondents “definitely would not” continue the activity (Fig. 1a). During a severe thunderstorm watch, 84.9% of the respondents “definitely would monitor the weather.”

Fig. 1.

Likelihood of continuing an activity during severe weather watches. (a) Likelihood of continuing an activity during a severe thunderstorm watch; (b) likelihood of continuing an activity during a tornado watch; (c) likelihood of continuing an activity during a PDS tornado watch.

Fig. 1.

Likelihood of continuing an activity during severe weather watches. (a) Likelihood of continuing an activity during a severe thunderstorm watch; (b) likelihood of continuing an activity during a tornado watch; (c) likelihood of continuing an activity during a PDS tornado watch.

For a tornado watch, one third (33%) of the respondents indicated that they “definitely would” or “probably would” continue an activity lasting any duration, while 54.8% “probably would not” or “definitely would not.” Similar to the severe thunderstorm watch, 12.2% were “not sure” if they would continue the activity. The greatest number (37.4% and 33.4%) of the respondents indicated that they “probably would” continue an activity lasting 30 min or 1 h, respectively. However, 30.9% of the respondents stated that they “probably would not” continue an activity lasting 2 h. For an activity lasting 4 h or the entire duration of the tornado watch, 35.7% and 40.9% of the respondents “definitely would not” continue the activity, respectively (Fig. 1b). During a tornado watch, 92.3% of the respondents “definitely would monitor the weather.”

Regarding a PDS tornado watch, 12.6% of the respondents indicated that they “definitely would” or “probably would” continue an activity lasting any duration, while 79.1% “probably would not” or “definitely would not.” A smaller 8.2% were “not sure” if they would continue the activity. Much larger percentages (44.3%, 51.6%, 58.9%, 63.6%, and 66%) of the respondents indicated that they “definitely would not” continue an activity lasting 30 min, 1 h, 2 h, 4 h, or the entire duration of the PDS tornado watch, respectively (Fig. 1c). During a PDS tornado watch, 96.5% of the respondents “definitely would monitor the weather.”

b. Influence of length of activity within watch categories

Means were compared to determine how the length of an activity and the severity of the watch affected the respondents’ decisions. Lower averages represent a higher likelihood of continuing an activity, while higher averages indicate a higher likelihood of not continuing an activity. A Friedman test was conducted to identify the statistical difference between the respondents’ likelihood to continue an activity based on the length of each activity. For each severe weather watch, the likelihood of continuing an activity decreased as the duration of the activity increased (Table 3). A post hoc test was used to determine the statistical significance between each variable. A Bonferroni adjustment was applied to minimize the effects of a Type 1 error by lowering the alpha value required for significance.

Table 3.

Mean likelihood of continuing an activity for a specified severe weather watch among each activity duration. Lower averages represent a higher likelihood of continuing an activity. Standard deviations are provided in parentheses.

Mean likelihood of continuing an activity for a specified severe weather watch among each activity duration. Lower averages represent a higher likelihood of continuing an activity. Standard deviations are provided in parentheses.
Mean likelihood of continuing an activity for a specified severe weather watch among each activity duration. Lower averages represent a higher likelihood of continuing an activity. Standard deviations are provided in parentheses.

For a severe thunderstorm watch, there was a significant effect of the length of the activity on the respondents’ likelihood of continuing an activity (N = 2110, X2 = 3188.043, df = 4, p < 0.001). The greatest difference between ranks, −29.323, occurred between 30 min and the entire duration of the watch, while the smallest difference, −6.511, occurred between 4 h and the entire duration of the watch (Table 4).

Table 4.

Severe thunderstorm watch, tornado watch, and PDS tornado watch pairwise comparison. The asterisk in the “Z” column indicates that time A is statistically significantly different from the time B variables.

Severe thunderstorm watch, tornado watch, and PDS tornado watch pairwise comparison. The asterisk in the “Z” column indicates that time A is statistically significantly different from the time B variables.
Severe thunderstorm watch, tornado watch, and PDS tornado watch pairwise comparison. The asterisk in the “Z” column indicates that time A is statistically significantly different from the time B variables.

For a tornado watch, there was a significant effect of the length of the activity on the respondents’ likelihood of continuing an activity (N = 2102, X2 = 3155.437, df = 4, p < 0.001). The greatest difference between ranks, −28.891, occurred between 30 min and the entire duration of the watch, while the smallest difference, −9.751, occurred between 4 h and the entire duration of the watch (Table 4).

For a PDS tornado watch, there was a significant effect of the length of the activity on the respondents’ likelihood of continuing an activity (N = 2080, X2 = 1814.924, df = 4, p < 0.001). The greatest difference between ranks, −22.282, occurred between 30 min and the entire duration of the watch, while the smallest difference, −4.448, occurred between 4 h and the entire duration of the watch (Table 4).

With respect to increasing watch severity (severe thunderstorm watch, a tornado watch, and a PDS tornado watch) as the length of the activity increased, the likelihood of a respondent continuing an activity decreased. Each independent variable was statistically significantly different from every other independent variable, with the smallest difference observed between 4 h and the entire duration of the watch conditions (Table 4).

c. Influence of watch categories on activities of a given length

The Friedman test was conducted to identify the statistical difference between the respondents’ likelihood to continue an activity based on the severity of the watch. The likelihood of continuing an activity decreased for each activity as the severity of the severe weather watch increased. A post hoc test was used to determine the statistical significance between each variable. A Bonferroni adjustment was applied to minimize the effects of a Type 1 error by lowering the alpha value required for significance.

For an activity lasting 30 min, there was a significant effect of the type of the watch on the respondents’ likelihood of continuing an activity (N = 2129, X2 = 2305.347, df = 2, p < 0.001). The greatest difference between ranks, −33.861, occurred between a severe thunderstorm watch and a PDS tornado watch, while the smallest difference, −18.771, occurred between a severe thunderstorm watch and a tornado watch (Table 5).

Table 5.

Pairwise comparison for an activity lasting 30 min, 1 h, 2 h, 4 h, and the entire duration of the severe weather watch. The asterisk in the Z column indicates that type A is statistically significantly different from the type B variables.

Pairwise comparison for an activity lasting 30 min, 1 h, 2 h, 4 h, and the entire duration of the severe weather watch. The asterisk in the Z column indicates that type A is statistically significantly different from the type B variables.
Pairwise comparison for an activity lasting 30 min, 1 h, 2 h, 4 h, and the entire duration of the severe weather watch. The asterisk in the Z column indicates that type A is statistically significantly different from the type B variables.

This pattern repeated itself for each of the lengths of time provided. This was true for an activity lasting 1 h (N = 2090, X2 = 2288.620, df = 2, p < 0.001); an activity lasting 2 h (N = 2083, X2 = 2054.262, df = 2, p < 0.001); an activity lasting 4 h (N = 2085, X2 = 1684.151, df = 2, p < 0.001); and an activity lasting the entire duration of the watch (N = 2072, X2 = 1598.014, df = 2, p < 0.001). The greatest difference between ranks also occurred between a severe thunderstorm watch and a PDS tornado watch for each of the remaining activity durations (ranging from −33.863 for 1 h to −28.464 for the entire duration of the watch). Similarly, the smallest difference between ranks occurred between a severe thunderstorm watch and a tornado watch for each of the activity durations (ranging from −20.744 for 1 h to −19.442 for a 4 h activity; Table 5).

As the severity of the watch increased, the likelihood of a respondent continuing an activity lasting the 30 min, 1 h, 2 h, 4 h, and the entire duration of the watch decreased. Each independent variable was statistically significantly different from every other independent variable.

The threshold for not continuing an activity was calculated for each severe weather watch. The threshold was calculated by determining the first time a respondent would not continue an activity. In each watch category, the threshold for the greatest number of respondents was 30 min. For a severe thunderstorm watch, a tornado watch, and a PDS tornado watch, 53% (1112), 67.8% (1425), and 84.7% (1762) of the respondents, respectively, indicated that at least one threshold existed. For a severe thunderstorm watch, 36.1% (401) of the 1112 respondents would not continue an activity lasting 30 min or longer (Table 6), and 51.2% (729) of the 1425 respondents would not continue an activity lasting 30 min or longer for a tornado watch. Regarding a PDS tornado watch, 80.2% (1413) of the 1762 respondents would not continue an activity lasting 30 min or longer. All values appear in Table 6. The percentage for whom each length of time is the threshold for continuing the activity generally decreases with increasing length of time. However, in each type of watch, 2 h is a threshold for more people than any length of time other than 30 min.

Table 6.

Threshold for not continuing an activity.

Threshold for not continuing an activity.
Threshold for not continuing an activity.

5. Discussion and conclusions

Watches are longer in duration than tornado warnings, and cover a larger spatial area, thus having the potential to disrupt the normal daily routine for many people. In Ripberger et al. (2014), over 6 times as many people were affected by watches than warnings. Watches are perceived as less serious and less likely to result in a threat (Schultz et al. 2010); therefore, it is not clear to what extent they influence behavior. Results in this study suggested that both the type of watch and the length of activity a person was (hypothetically) going to engage in influence whether or not one chooses to carry on with the activity. Respondents were more likely to continue with an activity during a severe thunderstorm watch than during a tornado watch. Similarly, they were more likely to continue with an activity during a tornado watch than a PDS tornado watch. This was true for each individual activity length. As the severity of the watch increased, the likelihood of the respondent continuing the activity decreased. This was illustrated in the percentage who definitely would not continue an activity during an entire PDS tornado watch (66%), compared to the percentages who definitely would not continue an activity during the entire severe thunderstorm watch (29.2%) or the entire tornado watch (40.9%). Similarly, the percentages who would probably or definitely continue on with an activity lasting only 30 min decreased from almost three quarters (71.5%) during a severe thunderstorm watch to only 20% in a PDS tornado watch, with about half (54.7%) continuing a 30-min activity during a tornado watch. Casteel (2018) found that the extreme wording associated with impact-based tornado warnings resulted in a more likely decision to shelter in place. Differences in decisions varied between some, but not all, threat levels. The author explains “that the heightened risk transmitted by the stronger impact statements does communicate increased risk and that extreme language is not necessary to craft an effective tornado warning” (Casteel 2018).

Respondents also exhibited increasing levels of caution when the length of the activity increased. For each type of watch, the likelihood of continuing the activity decreased as the activity length increased. There was some overlap in likelihood among watch categories and length of activity between the severe thunderstorm watch and tornado watch. For example, participants were more likely to continue with an hour-long activity during a severe thunderstorm watch than any length of time in a tornado watch. However, the average likelihood of continuing an activity of any length during a tornado watch was greater than the likelihood of continuing an activity of only 30 min during a PDS tornado watch. The mean differences were greater between both severe thunderstorm watch and tornado watch compared to PDS tornado watches for each length of activity, with the difference between tornado watch and PDS tornado watch becoming smaller as the length of activity increased.

In this study, the distinction of a PDS watch seemed to make a difference in the actions taken by the participants. This was not the case in past research that found no difference in participants who would choose a safer shelter option in a tornado watch versus PDS watch scenario (Mason and Senkbeil 2015). However, the fact that each increase in watch severity corresponded to a lower likelihood to continue an activity of similar duration was consistent with Mason and Senkbeil’s (2015) general conclusions that increasing severity levels of a tornado watch resulted in significant increases in cautious behavior. This difference could be the result of participants in Mason and Senkbeil (2015) not understanding what the PDS tornado watch meant. In their study, respondents were not told the specific watch type but instead the threats appropriate to each watch.

The research demonstrates that different levels of watches may have varying levels of influence on actions taken, and that individuals may take into account the length of the activity when deciding whether or not to continue on with it. However, in order for a watch to have any influence on behavior, a person must be aware of the watch. Similar to Silver (2015), the current study demonstrated individuals’ willingness to alter travel arrangements during severe weather. In Silver (2015) this was especially true for those individuals who checked the forecast more frequently during severe weather. We did not compare forecast use with response to our scenarios.

Sutter and Simmons (2014, p. 22) found that a watch has no effect on fatalities. However, our study demonstrates that individuals will alter their hypothetical plans if they are presented with adequate information about the watch. Therefore, a watch may not be effective in preventing fatalities since people may not be fully aware of the watch and the associated threats. People may not be aware enough of the watch in order for it to have a measurable influence on decisions that lead to fatalities, which only affect a very small number of individuals each year. The research reveals the importance of a broader range of research on societal impacts associated with severe weather forecasting and severe weather watches, not just casualties and damage.

While research has shown an increase in public attention, via tweets, when watches are issued, the attention did not rise to the same level as when a tornado warning was issued (Ripberger et al. 2014). Future research is still required to examine to what extent individuals receive current tornado watch information, and whether they are receiving specifically SPC-issued watches, or a similar message of the likelihood of severe weather only from another source. The influence on behavior may not be important, but the knowledge could help frame any discussion about how to modify existing watch products.

The current study also adds to the discussion about how much lead time is desirable during a tornado warning by demonstrating that people may take timing into account when deciding on protective actions. The average lead time for a tornado warning is 13 min, while the preferred lead time measured in past research is 34.3 min (Hoekstra et al. 2011). Respondents in that study indicated that taking shelter was a lower priority with a 1-h lead time, and also indicated that this longer lead time was not as life threatening (Hoekstra et al. 2011). Severe weather watches typically last 6–8 h and thus it makes sense that respondents in the current study would be more likely to continue activities of shorter lengths or activities during watches they perceived as less threatening.

This paper focuses on the effects of severe weather watches on planned Saturday afternoon and evening activities away from the immediate vicinity of the respondent’s home. Other information provided by the respondents—including their weather knowledge, sources of weather information, household characteristics, and geographic locations—will be addressed in a separate paper to better understand the effects on risk-taking decisions.

Limitations of this research should be noted. The sample size was large, but somewhat more educated than the general U.S. population. The sample did include a sizeable percentage with a high school diploma or less education (22.4%), but this was balanced by 25.6% of respondents who had an advanced degree, which is much greater than the general population. Because it was not a randomly recruited sample, respondents may also be more interested in weather than the average person. However, each comparison was conducted within subject, so the conclusions about length of activity and watch level would still be valid. Each respondent viewed each scenario in the same order. This could have led to some desirability bias in the responses if respondents believed the researchers desired for them to alter their responses to the different scenarios.

The term “daily activity,” which was presented in the hypothetical scenario, may have been misleading or misunderstood by the respondents. It was not stated whether the activity was a responsibility or reward, or whether the activity was located indoors or outdoors. The respondents’ risk-taking decisions may have been influenced by their outlook on the activity. Respondents with rewarding activities planned may take more risks than respondents with less desirable activities. Also, respondents might have felt safer continuing the activity if the activity was located indoors opposed to outdoors. Even though the duration of the watch was specified in the hypothetical scenario, respondents may have associated daily activities with activities completed specifically during the daytime hours. Respondents may not have considered daily activities as those located at least 20 min from their house. Future research should examine a variety of hypothetical scenarios that are not constricted to times, locations, or daily activities. Future research should also attempt to further examine the influence of activity length and watch type in an actual, as opposed to hypothetical, setting. It should attempt to examine the social, moral, and hedonistic effects on risk-taking decision as well as the type of activity planned.

REFERENCES

REFERENCES
Ash
,
K. D.
,
R. L.
Schumann
III
, and
G. C.
Bowser
,
2014
:
Tornado warning trade-offs: Evaluating choices for visually communicating risk
.
Wea. Climate Soc.
,
6
,
104
118
, https://doi.org/10.1175/WCAS-D-13-00021.1.
Balluz
,
L.
,
L.
Schieve
,
T.
Holmes
,
S.
Kiezak
, and
J.
Malilay
,
2000
:
Predictors for people’s response to a tornado warning: Arkansas, 1 March 1997
.
Disasters
,
24
,
71
77
, https://doi.org/10.1111/1467-7717.00132.
Casteel
,
M. A.
,
2018
:
An empirical assessment of impact based tornado warnings on shelter in place decisions
.
Int. J. Disaster Risk Reduct.
, https://doi.org/10.1016/j.ijdrr.2018.01.036, in press.
Chaney
,
P. L.
, and
G. S.
Weaver
,
2010
:
The vulnerability of mobile home residents in tornado disasters: The 2008 Super Tuesday tornado in Macon County, Tennessee
.
Wea. Climate Soc.
,
2
,
190
199
, https://doi.org/10.1175/2010WCAS1042.1.
Comstock
,
R. D.
, and
S.
Mallonee
,
2005
:
Comparing reactions to two severe tornadoes in one Oklahoma community
.
Disasters
,
29
,
277
287
, https://doi.org/10.1111/j.0361-3666.2005.00291.x.
Drabek
,
T. E.
, and
K. S.
Boggs
,
1968
:
Families in disaster: Reactions and relatives
.
J. Marriage Fam.
,
30
,
443
451
, https://doi.org/10.2307/349914.
Hammer
,
B.
, and
T. W.
Schmidlin
,
2002
:
Response to warnings during the 3 May 1999 Oklahoma City tornado: Reasons and relative injury rates
.
Wea. Forecasting
,
17
,
577
581
, https://doi.org/10.1175/1520-0434(2002)0170577:RTWDTM2.0.CO;2.
Harrison
,
D. R.
, and
C. D.
Karstens
,
2017
:
A climatology of operational storm-based warnings: A geospatial analysis
.
Wea. Forecasting
,
32
,
47
60
, https://doi.org/10.1175/WAF-D-15-0146.1.
Hoekstra
,
S.
,
K.
Klockow
,
R.
Riley
,
J.
Brotzge
,
H.
Brooks
, and
S.
Erickson
,
2011
:
A preliminary look at the social perspective of warn-on-forecast: Preferred tornado warning lead time and the general public’s perceptions of weather risks
.
Wea. Climate Soc.
,
3
,
128
139
, https://doi.org/10.1175/2011WCAS1076.1.
Karl
,
T. R.
, and
W. J.
Koss
,
1984
: Regional and national monthly, seasonal, and annual temperature weighted by area, 1895–1993. National Climatic Data Center Historical Climatology Series 4-3, 38 pp., https://repository.library.noaa.gov/view/noaa/10238.
Klockow
,
K. E.
,
R. A.
Peppler
, and
R. A.
McPherson
,
2014
:
Tornado folk science in Alabama and Mississippi in the 27 April 2011 tornado outbreak
.
GeoJournal
,
79
,
791
804
, https://doi.org/10.1007/s10708-013-9518-6.
Kox
,
T.
, and
A. H.
Thieken
,
2017
:
To act or not to act? Factors influencing the general public’s decision about whether to take protective action against severe weather
.
Wea. Climate Soc.
,
9
,
299
315
, https://doi.org/10.1175/WCAS-D-15-0078.1.
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.
Legates
,
D. R.
, and
M. D.
Biddle
,
1999
: Warning response and risk behavior in the Oak Grove–Birmingham, Alabama tornado of 8 April 1998. Natural Hazards Research and Applications Information Center Quick Response Research Rep. 116, 43 pp., https://hazdoc.colorado.edu/bitstream/handle/10590/692/NHC-A-QR-1999-116.pdf?sequence=1.
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.
Lindell
,
M. K.
,
S.-K.
Huang
,
H.-L.
Wei
, and
C. D.
Samuelson
,
2016
:
Perceptions and expected immediate reactions to tornado warning polygons
.
Nat. Hazards
,
80
,
683
707
, https://doi.org/10.1007/s11069-015-1990-5.
Liu
,
S.
,
L. E.
Quenemoen
,
J.
Malilay
,
E.
Noji
,
T.
Sinks
, and
J.
Mendlein
,
1996
:
Assessment of a severe-weather warning system and disaster preparedness, Calhoun County, Alabama, 1994
.
Amer. J. Public Health
,
86
,
87
89
, https://doi.org/10.2105/AJPH.86.1.87.
Mason
,
J. B.
, and
J. C.
Senkbeil
,
2015
:
A tornado watch scale to improve public response
.
Wea. Climate Soc.
,
7
,
146
158
, https://doi.org/10.1175/WCAS-D-14-00035.1.
Mileti
,
D. S.
, and
J. H.
Sorensen
,
1990
: Communication of emergency public warnings: A social science perspective and state-of-the-art assessment. Oak Ridge National Laboratory Tech. Rep. ORNL-6609, 166 pp, http://www.cires.org.mx/docs_info/CIRES_003.pdf.
Miran
,
S. M.
,
C.
Ling
, and
L.
Rothfusz
,
2018
:
Factors influencing people’s decision-making during three consecutive tornado events
.
Int. J. Disaster Risk Reduct.
,
28
,
150
157
, https://doi.org/10.1016/j.ijdrr.2018.02.034.
Mitchem
,
J. D.
,
2003
: An analysis of the September 20, 2002 Indianapolis tornado: Public response to a tornado warning and damage assessment difficulties. Natural Hazards Research and Applications Information Center Quick Response Research Rep. 161, 76 pp., https://hazdoc.colorado.edu/bitstream/handle/10590/682/NHC-A-QR-2003-161.pdf?sequence=1.
NSSL
,
2018
: Severe weather 101—Thunderstorms. Accessed 14 March 2018, https://www.nssl.noaa.gov/education/svrwx101/thunderstorms/.
NWS
,
2017
: NWS Norman watch, warning and advisory criteria. Norman, OK Weather Forecast Office, accessed 1 March 2015, http://www.srh.noaa.gov/oun/?n=spotter-wwa-definitions.
Paul
,
B. K.
,
V. T.
Brock
,
S.
Csiki
, and
L.
Emerson
,
2003
: Public response to tornado warnings: A comparative study of the May 4, 2003, tornados in Kansas, Missouri, and Tennessee. Natural Hazards Research and Applications Information Center Quick Response Research Rep. 165, 27 pp., https://hazdoc.colorado.edu/bitstream/handle/10590/700/NHC-A-QR-2003-165.pdf?sequence=1.
Ripberger
,
J. T.
,
H. C.
Jenkins-Smith
,
C. L.
Silva
,
D. E.
Carlson
, and
M.
Henderson
,
2014
:
Social media and severe weather: Do tweets provide a valid indicator of public attention to severe weather risk communication?
Wea. Climate Soc.
,
6
,
520
530
, https://doi.org/10.1175/WCAS-D-13-00028.1.
Schultz
,
D. M.
,
E. C.
Gruntfest
,
M. H.
Hayden
,
C. C.
Benight
,
S.
Drobot
, and
L. R.
Barnes
,
2010
:
Decision making by Austin, Texas, residents in hypothetical tornado scenarios
.
Wea. Climate Soc.
,
2
,
249
254
, https://doi.org/10.1175/2010WCAS1067.1.
Schumacher
,
R. S.
,
D. T.
Lindsey
,
A. B.
Schumacher
,
J.
Braun
,
S. D.
Miller
, and
J. L.
Demuth
,
2010
:
Multidisciplinary analysis of an unusual tornado: Meteorology, climatology, and the communication and interpretation of warnings
.
Wea. Forecasting
,
25
,
1412
1429
, https://doi.org/10.1175/2010WAF2222396.1.
Sherman-Morris
,
K.
,
2010
:
Tornado warning dissemination and response at a university campus
.
Nat. Hazards
,
52
,
623
638
, https://doi.org/10.1007/s11069-009-9405-0.
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.
Sherman-Morris
,
K.
, and
M. E.
Brown
,
2012
:
Experiences of Smithville, Mississippi residents with the 27 April 2011 tornado
.
Natl. Wea. Dig.
,
36
,
93
101
, http://nwafiles.nwas.org/digest/papers/2012/Vol36No2/Pg093-Sherman-Brown.pdf.
Silver
,
A.
,
2015
:
Watch or warning? Perceptions, preferences, and usage of forecast information by members of the Canadian public
.
Meteor. Appl.
,
22
,
248
255
, https://doi.org/10.1002/met.1452.
SPC
,
2017
: SPC products. Accessed 1 March 2015, http://www.spc.noaa.gov/misc/about.html.
SPC
,
2018
: SPC severe weather events archive. Accessed 14 March 2018, http://www.spc.noaa.gov/exper/archive/events/.
Stokes
,
C. T.
, and
J. C.
Senkbeil
,
2017
:
Facebook and Twitter, communication and shelter, and the 2011 Tuscaloosa tornado
.
Disasters
,
41
,
194
208
, https://doi.org/10.1111/disa.12192.
Strawderman
,
L.
,
D. W.
Carruth
,
K.
Sherman-Morris
,
P.
Menard
,
M.
Warkentin
, and
K. S.
McNeal
,
2018
:
Individual transportation decisions under conditions of risk and uncertainty
.
Nat. Hazards
,
92
,
927
942
, https://doi.org/10.1007/s11069-018-3232-0.
Sutter
,
D.
, and
K. M.
Simmons
,
2014
:
Preparing for danger: On the impact of tornado watches on tornado casualties
.
Int. J. Mass Emerg. Disasters
,
32
,
1
25
, http://www.ijmed.org/articles/647/download/.
Tiefenbacher
,
J. P.
,
W.
Monfredo
,
M.
Shuey
, and
R. J.
Cecora
,
2001
: Examining a “near-miss” experience: Awareness, behavior, and post-disaster response among residents on the periphery of a tornado-damage path. Natural Hazards Research and Applications Information Center Quick Response Research Rep. 137, 21 pp., https://hazdoc.colorado.edu/bitstream/handle/10590/643/NHC-A-QR-2001-137.pdf?sequence=1.

Footnotes

© 2018 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).