1

International Journal of Mass Emergencies and Disasters March 2007, Vol. 25, No. 1, pp. 1–36

Call-to-Action Statements in Warnings: Do They Reflect Recent Developments in Tornado-Safety Research?

John E. Farley Professor Emeritus Department of Sociology and Criminal Justice Studies Southern Illinois University Edwardsville [email protected]

Call-to-action statements in tornado warnings are content analyzed to determine to what extent their wording has been influenced by recent research calling into question official safety guidelines and traditional advice regarding vehicles and mobile homes. While the statements do not directly contradict official guidelines and advice, there is significant variation among NWS offices regarding what advice is given and what guidelines are emphasized in call-to-action statements in tornado warnings. Some of this variation is regional, and interviews with NWS meteorologists reveal a frequent opinion that what is best to do if in a vehicle during a tornado warning may vary by region, time of day, and terrain. The interviews also reveal widespread awareness among NWS meteorologists of debates over tornado safety in vehicles and mobile homes, and strong support for local office autonomy in decisions about the wording of call-to-action statements.

Key Words: tornado, warning, advice, safety, recommendations.

Introduction The research reported in this paper examines “call to action” (CTA) statements in tornado warnings issued by the National Weather Service (NWS) in light of scientific controversies associated with

1 2 International Journal of Mass Emergencies and Disasters recent research by Thomas W. Schmidlin and others regarding the safety of vehicles relative to mobile homes and ditches. This research presents a dilemma to the NWS and other emergency management agencies because it calls into question the validity of safety guidelines that have been used for a number of years. It also presents a challenge because the recent research suggests that the safest action to take is highly conditional upon the specifics of the situation (Schmidlin, 2002), while research on risk communication indicates that repeated provision of clear and understandable messages about the nature of the risk and what to do about it is the means of communication most likely to lead to adoption of appropriate preparedness and response measures (Mileti et al., 1993; Mileti, 1997). For example, research has confirmed that in tornadoes and other emergencies, knowledge of safety rules does lead people to take actions that can prevent death and injury (Aguirre, et al., 1987; Burby and Wagner, 1996; Perry, Lindell, and Green, 1980; Rogers, 1985). Relatedly, people are more likely to respond to a warning message if it contains information about what to do to protect themselves (DiGiovanni et al., 2002; Mileti and Darlington, 1997). Warning messages are most effective when they are not only clear but also repeated (Drabek, 1969; Mikami and Ikeda, 1985; Mileti and Fitzpatrick, 1993) and consistent (Christiansen and Ruch, 1980; DiGiovanni et al., 2002; Nigg, 1987). For example, Mileti and Fitzpatrick (1993) have shown that when the same information is received multiple times from multiple different sources, its impact on risk perception and on people’s efforts to obtain further information is maximized (see also Pfister, 2002). Unfortunately, it can be difficult to give repeated and consistent recommendations when the best course of action is highly situation- dependent. Moreover, research has shown that a clear and simple message is more likely to evoke action than one that is complicated or contains multiple response options (DiGiovanni et al., 2002; Rogers and Sorensen, 1989). Hence, the research shows that the advice given in warnings for tornadoes and other hazards has real impacts on what people do, but also that giving good safety advice that will be heeded represents a significant challenge when the best response is situation-dependent. The purpose of this paper Farley: Tornado Warnings 3 is to determine the extent to which actions recommended in CTA statements are consistent or inconsistent with the recommendations of Schmidlin and his colleagues based on their research findings, and how these CTA statements, if at all, have been influenced by debates surrounding those findings. This represents an important issue, because as Golden and Adams (2000) have pointed out, considerably more research has focused on the physical sciences and technology portion of the warning process than upon the warning communication process, the behavioral response to warnings, and the epidemiology of tornadoes and their consequences. Schmidlin’s Research on Vehicles, Mobile Homes, and Tornado Safety Research findings by Schmidlin and his colleagues (Schmidlin 1997; Schmidlin et al. 2002) suggest that, while vehicles are far from the safest place to be in a tornado, they may in many instances be safer than mobile homes (so that a person in a mobile home with no sturdy shelter nearby would be safer in a vehicle than in the mobile home). Nonetheless, there are no tornado safety guidelines currently in use by the National Weather Service or American Red Cross that convey the idea that a vehicle may be safer than a mobile home. Schmidlin’s research and that of others also suggests that vehicles may be safer than outdoor locations such as ditches, and Schmidlin notes there is no research supporting the idea that a ditch is safer than a vehicle . Yet, it is often recommended in safety guidelines that people abandon mobile homes and vehicles for a ditch, if no sturdy shelter is available. Schmidlin and his colleagues conducted site surveys of mobile home parks and vehicles that had been hit by tornadoes of various strength. They found that in a large number of instances, mobile homes were severely damaged or destroyed, while vehicles parked next to them or nearby suffered no major damage. In these cases, a person remaining in the mobile home would have been killed or severely injured, while a person in the nearby vehicle would have been unharmed. In general, Schmidlin found that winds of about 80 miles per hour were required to tip a mobile home, while much stronger winds of 120 miles per hour were required to tip a car. 4 International Journal of Mass Emergencies and Disasters

Specifically, he found that in tornadoes rated F1 or F2 on the Fujita Scale 1, just 4 percent of the vehicles he observed were tipped over and just 28 percent were moved. The upper end of the F2 range is estimated wind of 157 mph at a height of 10 meters, which translates to a speed of about 120 mph at door-handle height. At this wind speed, most mobile homes are tipped or blown apart, even if tied down, in contrast to the overwhelming majority of cars that survive and are not tipped, or in most cases even moved or significantly damaged. At higher Fujitia Scale ratings, cars suffer greater, sometimes devastating, damage, but the great majority of tornadoes are F2 or weaker. For example, between 1985 and 1996, more than 85 percent of tornadoes were F0 or F1, and fewer than 4 percent were F3 or stronger (Thompson and Vescio, 1998). In addition to the site survey research, Schmidlin and his colleagues also performed wind tunnel tests on one-sixth scale models of a midsize sedan and a minivan (Schmidlin et al., 2002). The tests indicated that it would take winds around 115-130 mph to tip the sedan (depending on wind direction) and a little stronger to tip the minivan. These findings are very consistent with those of the site-survey research, indicating that it takes much stronger winds to tip an automobile than to tip or blow apart a mobile home. It has also been pointed out by Schmidlin (2002) and by Bernner and Noji (1993) that there is no research evidence establishing that outdoor locations such as ditches are safer than vehicles, and multiple studies of tornadoes in which it was found that people who stayed in their vehicles fared better than those who tried to take shelter outside. In the Marion, IL tornado of 1982, DuClos and Ing (1989) found that people who tried to drive to safety were less likely to be injured than people who took shelter outside, and that nobody was injured in a vehicle even in a few instances in which the vehicle was overturned by the tornado. Carter, Millson, and Allen (1989) also found in the Barrie, tornado of 1985 that people who remained in vehicles were less likely to be injured than people who took shelter outdoors. Two studies of the devastating tornadoes in Oklahoma on May 3, 1999 also produced similar findings. Daley et al. (2005) found that people in vehicles were less likely to be injured or killed than persons who were either outdoors or in mobile homes. Additionally, Hammer Farley: Tornado Warnings 5 and Schmidlin (2002) found that people who left their homes to drive out of the path of the tornado fared better than even people who remained in their homes. This finding may be tied to 1) the lack of basements in most homes in that area and 2) the long lead time (30 minutes) between the warning and the time the tornado struck. Additionally, there are multiple documented cases in recent years in which people have followed the recommendation to abandon their vehicles and get in the ditch and have been killed as a result, either by their own vehicle or other objects. On February 16, 2001, a woman saw a tornado near Goodman, Mississippi and abandoned her car. The tornado, which was rated F2, rolled her car over onto her, causing her death (National Weather Service Storm Prediction Center, 2004). In 1999, a woman and a fifteen year old girl were killed near Logan, Iowa when they took shelter in a ditch, and their abandoned car and a 3-ton soybean head from a combine were blown on top of them by an F3 tornado (National Climatic Data Center, 2000, p. 109). On the other hand, there are also multiple documented cases in which people died in vehicles, not necessarily because the vehicle was tipped, but because it was hit by flying projectiles or falling tree branches. For example, on June 7, 2001, a man died when a tree fell on the pickup truck he was driving near Zachary, LA (Storm Prediction Center, 2002). The tornado was only rated F1. Hence, because of falling trees or flying projectiles, vehicles may be a dangerous place to be, even if they are not moved or tipped by the wind. It is also true that a handful of deaths and injuries in the May 3, 1999 Oklahoma City tornado occurred outdoors when people were attempting to reach their vehicles to drive out of the path or to shelter, or when they abandoned vehicles to get under an overpass (it is unknown whether these individuals had initially gotten into a vehicle to flee the tornado). While getting under an overpass is not recommended, either by the NWS and other emergency management agencies or by Schmidlin and his supporters, it is possible that some of the individuals in both groups might have survived had they remained in shelter rather than trying to flee. Nonetheless, the overall death and injury rates were far lower for those who drove away than for those who remained in mobile homes or were outdoors, and even somewhat lower than for those who remained in houses or apartments. 6 International Journal of Mass Emergencies and Disasters

Advice Given to the Public Concerning Vehicles, Mobile Homes, and Tornadoes There are a number of ways in which the NWS communicates with the public regarding what actions should be taken when a tornado threatens. First, the NWS and other public-safety and emergency- response agencies publish safety guidelines advising people what actions should be taken when a tornado threatens. Second, information on tornado safety is given to the public by NWS personnel at severe weather safety talks and storm spotter trainings . Finally, information on actions to take to protect oneself is included with tornado warnings in the form of CTA statements. The research reported here focuses primarily on CTA statements, although the NWS personnel who were interviewed also gave useful information on their practices in safety talks and spotter trainings. However, a brief review of the content of safety guidelines is in order, because these are these publicly- promulgated guidelines give a sense of the “official” NWS position. Of course, the degree to which this “official” position is reflected in the actual practices followed by individual NWS professionals in local offices throughout the country is uncertain, and represents the main research question addressed by this paper. An examination of safety guidelines promulgated by the NWS and related agencies reveals that the advice given remains in general inconsistent with the findings of the Schmidlin’s research. In fact, this has been acknowledged by Joseph Golden of the National Oceanic and Atmospheric Administration (NOAA). Golden (2002) states that the safety guidelines should not be changed because 1) although relatively few tornado fatalities occur in vehicles, many windstorm fatalities do, often due to trees falling on the vehicles, 2) there are documented cases of vehicles being lifted and smashed by tornadoes, and 3) attempts to flee tornadoes in vehicles are dangerous if traffic congestion occurs and account for many of the tornado deaths that do occur in vehicles. This points to a common concern among supporters of the current guidelines: It is difficult and confusing to issue guidelines that say to do one thing in one situation (drive away if you are in open country with no congestion) but something else in another (avoid vehicles if congestion is likely). Additionally, changing the guidelines causes confusion. Farley: Tornado Warnings 7

For example, it has been twenty years since the NWS dropped the recommendation that people open windows before a tornado strikes, but NWS meteorologists still receive questions about this recommendation (Golden, 2002). Golden does indicate support for changing guidelines to support abandonment of mobile homes for a stationary automobile, but no such recommendation yet appears in any of the safety guidelines issued by the NWS or American Red Cross. For example, the NWS Severe Weather Preparedness Guide brochure (National Weather Service, 1999)—probably the most “official” source of advice from the NWS, contains the following: MYTH If you are driving and a tornado is sighted, you should turn and drive at right angles to the storm. TRUTH The best thing to do is to seek the best available shelter. Many people are injured or killed when remaining in their vehicles. MYTH People caught in the open should seek shelter under highway overpasses. TRUTH Take shelter in a sturdy reinforced building if at all possible. Overpasses, ditches, and culverts may provide limited protection from a tornado, but your risk will be greatly reduced by moving inside a strong building. Note that the advice here is specifically against driving at right angles to the tornado. Advice on ditches and overpasses is ambiguous, as it (correctly, in the view of most researchers) states that they are less safe than a sturdy building but indicates they may “provide limited protection.” This could be, and often is, interpreted as saying that if no building is available, they may offer the “best available shelter” as referenced in the preceding statement advising abandonment of vehicles. Similar advice, but with a more positive take on getting in ditches, appears in NOAA Outlook (National Oceanic and Atmospheric Administration, no date), a Web page of the National Oceanic and Atmospheric Administration (NOAA), the parent organization of the NWS: 8 International Journal of Mass Emergencies and Disasters

Q. What actions should people take to protect themselves during a tornado? A. Move to a pre-designated shelter such as a basement in a home or building. If an underground shelter is not available, move to an interior room or hallway on the lowest floor and get under a sturdy piece of furniture. Stay away from windows; get out of cars—do not try to outrace a tornado. If you’re caught outside, lie flat in a nearby ditch or depression. Again, this statement advises against driving away from a tornado, and says to “get out of cars.” If one is caught outside (one way this could happen is by getting out of a car, if no better shelter is available nearby), the advice is to lie in a ditch. The “tornado safety tips” listed on the Web page of the NWS Storm Prediction Center (Edwards, 2004), is the most nuanced set of advice. It is written by an individual author (Roger Edwards of SPC) and does not appear to constitute an official guideline. In some regards, it offers slightly different advice than the above items: In a mobile home: Get out! Even if your home is tied down, you are probably safer outside, even if the only alternative is to seek shelter out in the open. Most tornadoes can destroy even tied-down mobile homes; and it is best not to play the low odds that yours will make it. If your community has a tornado shelter, go there fast. If there is a sturdy permanent building within easy running distance, seek shelter there. Otherwise, lie flat on low ground away from your home, protecting your head. If possible, use open ground away from trees and cars, which can be blown onto you.

In a car or truck: Vehicles are extremely dangerous in a tornado. If the tornado is visible, far away, and the traffic is light, you may be able to drive out of its path by moving at right angles to the tornado. Otherwise, park the car as quickly and safely as possible -- out of the traffic lanes. [It is safer to get the car out of mud later if necessary than to cause a crash.] Get out and seek shelter in a sturdy building. If in the open country, run to low ground away from any cars (which Farley: Tornado Warnings 9

may roll over on you). Lie flat and face-down, protecting the back of your head with your arms. Avoid seeking shelter under bridges, which can create deadly traffic hazards while offering little protection against flying debris. This advice is different from all of the other sources cited here in that it does mention the possibility of driving out of the path of the tornado under some circumstances. However, if the tornado cannot be avoided, the advice is still to get out, though the possibility of being hit by a vehicle moved by the tornado is mentioned. The advice to mobile home residents is the same as in the other statements: If there is no sturdy building or storm shelter available, get in a low place on the ground. Specifically, there is no mention of getting out of a mobile home and into a vehicle. Finally, the American Red Cross advises the following on its Web site: When a Tornado WARNING Is Issued...

฀ If you are inside, go to the safe place you picked to protect yourself from glass and other flying objects. The tornado may be approaching your area. ฀ If you are outside, hurry to the basement of a nearby sturdy building or lie flat in a ditch or low-lying area.

฀ If you are in a car or mobile home, get out immediately and head for safety (as above).

This advice, which the American Red Cross described as based on NWS guidelines, is the most specific in telling people that if they are in a car or mobile home and no sturdy building is available, it should be abandoned in favor of a ditch or low-lying area. Clearly, the advice from these agencies is not entirely in line with the principle that messages should be repeated and consistent. There are some inconsistencies among these messages, and someone reading all three would not be sure what to do. For example, one message suggests the possibility of driving out of the path of a tornado, while another says to get out of the car and do not try to outrun a tornado. One message does not even mention mobile 10 International Journal of Mass Emergencies and Disasters homes, the place where tornado deaths are most likely to occur. One equates ditches and overpasses, while another advises against going under overpasses under any conditions. Additionally, each set of safety recommendations is inconsistent with the recent findings of Schmidlin and others in at least one way. None of them indicate that a vehicle is safer than a mobile home, though Schmidlin argues that his research and that of others shows that it clearly is. Only one mentions the risk of objects falling on people taking cover in a ditch, although, as noted above, people have been killed in that manner in recent years (Goldhammer, 2002). And some advise abandonment of vehicles, even though at least some studies suggest people are probably safer in vehicles than in the ditch. Schmidlin’s research, and debate among Schmidlin, Golden, and others on what the safety rules should say, is well-known among the disaster-preparedness and meteorological communities. The debate, for example, received high-profile coverage in the University of Colorado’s widely-disseminated Natural Hazards Observer (Schmidlin, 1997; Lopes, 1997). Additionally, in a policy statement issued in 2000, the American Meteorological Society suggested in relation to mobile homes “If there is no shelter and there is enough warning lead time, it might be best to evacuate in a vehicle to avoid an approaching tornado or to reach a sturdy shelter” and “motorists should use all information to avoid an approaching tornado” and use the ditch only “as a last resort, to be used only when a tornado cannot be avoided” (American Meteorological Society, 2000). The purpose of my research was to examine to what extent, if any, the new research and the surrounding debate have impacted the day- to-day operations of NWS offices when issuing tornado warnings. Clearly, the official policy of the NWS has not changed, but the purpose of this research is to see what impact of the debates and of Schmidlin’s research has occurred on the local, operational level. CTA Statements in Tornado Warnings When the NWS issues a tornado warning, it is normally accompanied by a CTA. For example, the following is excerpted from a tornado warning issued by the Morristown, Tennessee NWS office on July 14, 2004: Farley: Tornado Warnings 11

344 PM EDT WED JUL 14 2004 THE NATIONAL WEATHER SERVICE IN MORRISTOWN HAS ISSUED A *TORNADO WARNING FOR...POLK COUNTY IN EAST TENNESSEE IF IN A MOBILE HOME OR VEHICLE...LEAVE IT IMMEDIATELY AND GET INSIDE A SUBSTANTIAL SHELTER. IF NO SHELTER IS AVAILABLE...LIE FLAT IN THE NEAREST DITCH AND COVER YOUR HEAD WITH YOUR HANDS. The wording immediately above is an example of a CTA statement in a tornado warning—a short statement telling people what to do to protect themselves. Most tornado warnings contain one or two CTA statements. These do not attempt to list all of the safety rules; any given CTA statement will usually focus on just one situation. Thus, vehicles and/or mobile homes may or may not be mentioned in any given CTA statement. In the research reported here, I analyze the content of CTA statements in tornado warnings issued by the NWS, to determine to what extent advice given is consistent with, neutral, mixed, or inconsistent with ideas arising from the new research findings by Schmidlin and others regarding vehicles, mobile homes, and ditches. I also analyze the extent and nature of variation among NWS offices in the consistency of the advice given with Schmidlin’s research. As discussed in greater detail later, I also interviewed several warning coordination meteorologists in local NWS offices to determine their views about the appropriateness of the current guidelines. This research is useful in seeing how a large-scale public governmental organization, the NWS, along with its individual forecast offices, deals with new research findings and the attendant debates and controversies in their everyday operations. It is also a way to see whether and to what extent individual forecasters in local offices may have been influenced by the findings and controversies, given the presence of official agency guidance that is contrary to the new arguments raised by Schmidlin and others. 12 International Journal of Mass Emergencies and Disasters

Determining the Degree of Discretion in CTA Statement Wording An important consideration in the feasibility of doing this research was the degree of discretion that NWS offices have in the wording of CTA statements. Unless there is some flexibility and discretion extended to forecasters in local offices, it is not reasonable to expect either 1) significant variation among local offices in the advice given in CTA statements or 2) efforts on the part of local forecasters to adjust the wording of CTA statements based on the new ideas about safety actions. To ascertain this before beginning the research, I spoke with a NWS meteorologist whose duties include issuing warnings. The meteorologist informed me that NWS local forecast offices indeed do have a good deal of discretion in the wording and content of CTA statements, so long as they do not directly contradict official guidelines. For example, he indicated that it would be permissible for a local NWS office to issue a CTA that mentioned abandoning mobile homes but did not mention abandoning automobiles. Given this discretion, it appears that—within broad limits—there is considerable ability for NWS offices to word these products in ways that are more or less consistent with Schmidlin’s argument; that are more or less influenced by recent debates about tornado safety in vehicles, that do or do not make clear statements about safety actions of debated effectiveness, and that do or do not include identical statements about what to do if you are in a vehicle or mobile home in the same CTA.

Procedures Content Analysis of CTA Statements I obtained through the Internet the wording of all tornado warnings issued by all NWS local offices between August 1, 2004 and July 30, 2005. This was done by accessing a database containing the exact wording of all tornado warnings issued in the . 2 This represented just over 3,000 tornado warnings. I drew a representative sample of one-sixth of the total number of tornado warnings, resulting in a representative sample of 510 tornado warnings. Farley: Tornado Warnings 13

For each warning, I content-analyzed the text of the warning to determine whether the warning contains one or more CTA statements. The main unit of analysis for this study was the tornado warning. For each warning sampled, I coded whether or not it contained a CTA statement and if so, how many. For each warning, the CTA statement or statements were coded as consistent with , neutral or mixed , irrelevant to , or inconsistent with ideas arising from the new research findings by Schmidlin and others regarding vehicles, mobile homes, and ditches. Detailed definitions were developed for each of these categories; these definitions are shown in Table 1.

Table 1. Coding of Tornado Warnings with Regard to Consistency of their CTA Statements with Schmidlin Recommendations Concerning Vehicles.

Consistent: Mentions abandoning or evacuating vehicles or mobile homes for buildings, but contains no information recommending abandonment of vehicle for ditch or low spot. Also, contains any recommendation pertaining to vehicle other than avoidance of overpasses, and does not recommend abandonment of vehicle for ditch or low spot. (May or may not include recommendation to abandon mobile home for ditch)

Mixed or ambiguous: Either contains recommendation to abandon vehicle without clear instructions as to what location to abandon the vehicle for, or contains what appears to be advice that is partly consistent and partly inconsistent with Schmidlin (e.g. mentions driving away but also getting in ditch, or recommends ditch, but only “if the tornado is nearby,” thereby implying that otherwise some other action would be better).

Inconsistent: Contains any recommendation to abandon or evacuate vehicle for ditch or low spot.

No relevant advice: Contains no advice regarding abandonment of vehicles or mobile homes, or contains no such advice except for avoidance of overpasses.

Additionally, I looked for patterns in the CTA statements: what are common combinations of recommended actions? This resulted in a second content coding, based on descriptions of the action(s) recommended in the CTA statement. 14 International Journal of Mass Emergencies and Disasters

Qualitative Interviews with NWS Meteorologists Additionally, I conducted qualitative interviews with meteorologists involved in issuing storm warnings in seven different NWS offices in different regions of the country. Attempts were made to interview meteorologists from a total of ten NWS offices. Seven interviews were completed, yielding a response rate of 70 percent. The purpose of this part of the research was to learn more about the opinions of those responsible for issuing warnings regarding what content should appear in CTA statements and in safety guidelines. To preserve confidentiality, the persons interviewed are identified in my research only as meteorologists involved in issuing warnings at local NWS offices around the country (usually the Warning Coordination Meteorologist).

Findings Content Analysis of CTA Statements in Tornado Warnings Of the 510 tornado warnings in the sample, 458, about 90 percent, contained one or more CTA statements (Table 2). About 44 percent (223 warnings) contained one CTA statement, while 46 percent (235 warnings) contained two or more (usually two; only about 7 percent contained 3 and just four warnings had 4 CTA statements). This shows that, as I was told by the meteorologist I interviewed prior to beginning this study, it is routine for tornado warnings to contain CTA statements, and the overwhelming majority of them do. A few of the tornado warnings were test or practice warnings, usually issued in conjunction with statewide tornado drills or severe weather awareness weeks. While 487 of the warnings were actual tornado warnings, 23, or 4.5 percent, were test or practice warnings. Both were included in the analysis, because some of the test or practice warnings (about half) do contain CTA statements, and these, like the ones issued with actual warnings, provide information to people about what they should do when a tornado threatens. Farley: Tornado Warnings 15

Table 2. Does Warning Contain Recommendation to Abandon Vehicle for Ditch? All Warnings Warnings with Warnings CTA Mentioning Vehicles CTA Included in Warning 458 (89.8%) Vehicles Mentioned in CTA 307 (67.0%) Abandoning vehicles 216 (70.4%) recommended Abandoning vehicles not mentioned 91 (29.6%) Vehicles not mentioned in CTA 151 (33.0%) No CTA in warning 52 (10.8%) N 510 458 307 2 (1) = 323.2 2 (1) =53.1 (p 2 (1) =50.9 (p < .01) < .01) (p < .01) * CTA=Call to action.

Same or different advice for mobile homes and vehicles? Of the 458 warnings that contain CTA statements, 307, or 67.0 percent, mention vehicles, and 297, or 64.8 percent, mention mobile homes or recreational vehicles (RVs). Of warnings containing statements that mention either vehicles or mobile homes/RVs, 86.8 percent contain at least one statement offering the same advice for vehicles as for mobile homes or RVs. From the standpoint of Schmidlin’s research, this is problematic, because a key point he argues is that the best actions to take are often not the same when you are in a vehicle as when you are in a mobile home. Furthermore, just 58 warnings – 18.1 percent of those that mentioned vehicles and/or mobile homes – contained any separate advice for people in vehicles or for people in mobile homes that was not offered for people in both locations. Moreover, about half of these consisted of advice for persons in vehicles not to seek shelter under overpasses, an issue not directly relevant to the issues being studied in this paper. Hence, only around 30 warnings contained any separate advice for people in vehicles and people in mobile homes/RVs, other than for drivers to avoid seeking shelter under overpasses. Clearly, the norm is that when advice is given for either vehicles or mobile homes – as it is 16 International Journal of Mass Emergencies and Disasters in nearly two-thirds of the tornado warnings with CTA statements— the same advice is given for both. Get in the ditch? What advice is given? The most common advice is to evacuate or abandon the vehicle or mobile home, and seek shelter in a sturdy building if available, but if a building is not available, to seek shelter in a ditch . In fact, as shown in Table 2, nearly half of all warnings that have CTA statements, and more than two-thirds of warnings that mention vehicles, contain a recommendation to abandon the vehicle and get into a ditch under at least some circumstances. Usually, the advice is to get in a ditch or low spot as a last resort or if no other shelter is available, but clearly the message is that if it comes to that choice, it is better to be in a ditch than in one’s vehicle. In a handful of cases, there is no qualification on the advice. In six tornado warnings, people were simply told to get out of their vehicle and into the ditch. Clearly, this is not the best thing to do – getting into a sturdy building is better, even as recognized in official tornado safety guidelines. Yet, this handful of tornado warnings contained statements such as: LEAVE VEHICLES AND GET INTO A DITCH OR LOW SPOT. A few warnings also mentioned both buildings and ditches, but did not indicate a preference, such as: ABANDON CARS AND MOBILE HOMES FOR A STURDIER BUILDING OR INTO A DITCH OR CULVERT. Additionally, some warnings simply advised people to get out of cars, with no advice about where to go, such as: GET OUT OF MOBILE HOMES AND CARS. Use of standardized wording. Most commonly, though, the advice was to get into the ditch as a last resort or if no sturdier shelter was available. The following standardized wording, mentioned earlier, appeared in 191 of the warnings, 41.7 percent of all warnings that contained CTA statements: IF IN MOBILE HOMES OR VEHICLES...EVACUATE THEM AND GET INSIDE A SUBSTANTIAL SHELTER. IF NO Farley: Tornado Warnings 17

SHELTER IS AVAILABLE...LIE FLAT IN THE NEAREST DITCH OR OTHER LOW SPOT AND COVER YOUR HEAD WITH YOUR HANDS. These 191 warnings constituted about 59 percent of all tornado warnings with CTA statements that mentioned vehicles. As discussed in greater detail below, this standardized wording appeared with great regularity in tornado warnings issued by numerous NWS offices. In fact, warnings that recommended leaving vehicles for ditches contained this wording 87.5 percent of the time. Consistency of CTA statements with Schmidlin’s recommendations. Given the prevalence of this wording, it is not surprising that a high proportion of warnings that mentioned vehicles were found to be inconsistent with Schmidlin’s recommendations. Using the categories described in Table 1, each warning was rated as consistent with, mixed or ambiguous, inconsistent with, or irrelevant to Schmidlin’s recommendations regarding vehicles and ditches. The distribution of these ratings is shown in Table 3. As shown in Table 3, nearly half of the warnings that contained CTA statements included advice inconsistent with Schmidlin’s recommendations, three times as many as contained recommendations consistent with Schmidlin’s recommendations. Usually, the reason the ratings were coded as inconsistent with Schmidlin’s recommendations is that they recommended getting out of a vehicle and into a ditch under at least some circumstances. Note from Table 1 that if a warning advised getting out of vehicles and into sturdy shelter without recommending ditches or other outdoor locations, it was coded as consistent with Schmidlin’s recommendations. This category accounts for the

Table 3. Are Recommendations Consistent with Schmidlin Research?

Consistent 70 (15.3%) Mixed or ambiguous 29 ( 6.3%) Inconsistent 210 (45.9%) No relevant advice 149 (32.5%) N 458 2 (3) =159.0 (p < .01) Note: the universe for this table is warnings with CTA statements. 18 International Journal of Mass Emergencies and Disasters overwhelming majority of recommendations coded as consistent. In 63 tornado warnings—13.8 percent of all tornado warnings that contained a CTA statement, the statement advised abandoning vehicles for a sturdy building or for more substantial shelter only, with no mention of leaving them for an outdoor location. There was also one warning that gave this advice with regard to mobile homes, with no mention of vehicles. Recommendations for mobile homes and RVs. Additionally, CTA statements in some tornado warnings made distinctions between mobile homes and vehicles in regard to what one should do. Some warnings contained relatively detailed instructions for evacuating mobile homes or RVs, such as: IF YOU ARE IN AN RV PARK...LEAVE YOUR RECREATIONAL VEHICLE. MANY PARKS HAVE CENTRALIZED BUILDINGS SUCH AS REC CENTERS...POOL HOUSES...COMMUNITY SHOWERS AND BATHS...OR A MAIN OFFICE WHICH CAN OFFER BETTER PROTECTION. IF YOU ARE IN A MOBILE HOME...ABANDON IT AND GO TO A NEARBY STURDY BUILDING. AS A LAST RESORT... LIE FLAT IN A DITCH AND COVER YOUR HEAD! IN ADDITION TO THE TORNADO...DAMAGING WINDS AND LARGE HAIL ARE LIKELY IN THE WARNED AREA. IF YOU ARE IN OR NEAR THE PATH OF THIS TORNADO... GO TO A BASEMENT OR A SMALL INTERIOR ROOM ON THE LOWEST FLOOR! IF YOU ARE IN A MOBILE HOME... CAMPER OR ARE CAUGHT OUTDOORS SEEK STURDY SHELTER. AS A LAST RESORT...LIE FLAT IN A DITCH AND COVER YOUR HEAD! Others gave similar advice in a more concise manner: EVACUATE MOBILE HOMES AND GO TO A STURDY BUILDING. MOVE TO A PLACE OF SAFETY...AWAY FROM WINDOWS...IN AN INTERIOR ROOM ON THE LOWEST FLOOR. EVACUATE MOBILE HOMES. IF DRIVING...DO NOT SEEK SHELTER UNDER A HIGHWAY OVERPASS. Farley: Tornado Warnings 19

Significantly, all of these wordings appeared in tornado warnings that did not contain a recommendation to leave vehicles and get in a ditch or other outdoor location, although some did recommend doing this as a last resort if in a mobile home. One NWS office also altered the wording of the standardized wording discussed above, to delete the reference to cars so that the advice applied only to mobile homes: IF IN MOBILE HOMES...EVACUATE THEM AND GET INSIDE A SUBSTANTIAL SHELTER. IF NO SHELTER IS AVAILABLE...LIE FLAT IN THE NEAREST DITCH OR OTHER LOW SPOT AND COVER YOUR HEAD WITH YOUR HANDS. There were eight warnings that contained CTA statements with this wording, all issued by the same NWS office. As this example and the others discussed just above show, there clearly are cases where advice is given to abandon or evacuate mobile homes, without similar advice being given for cars. Drive out of the path? Some warnings also contained CTA statements that were in some ways consistent with Schmidlin’s advice and in other ways inconsistent. For example, while there were no warnings in the sample that directly recommended driving away from the path of a tornado (though at least one not included in the sample did), there were a handful that did mention or imply that possibility, even while also mentioning the ditch option:

DO NOT USE HIGHWAY OVERPASSES FOR SHELTER. OVERPASSES DO NOT PROVIDE PROTECTION FROM TORNADIC WINDS. VEHICLES STOPPED UNDER BRIDGES BLOCK TRAFFIC AND PREVENT PEOPLE FROM GETTING OUT OF THE STORM’S PATH AND TO SHELTER. IF YOU CANNOT DRIVE AWAY FROM THE TORNADO...GET OUT OF YOUR VEHICLE AND LIE FLAT IN A DITCH AS A LAST RESORT. DO NOT USE YOUR CAR TO TRY TO OUTRUN A TORNADO. CARS ARE EASILY TOSSED AROUND BY TORNADO WINDS. IF YOU ARE CAUGHT IN THE PATH OF A TORNADO...LEAVE THE CAR AND GO TO A STRONG BUILDING. IF NO SAFE STRUCTURE IS NEARBY... 20 International Journal of Mass Emergencies and Disasters

SEEK SHELTER IN A DITCH OR LOW SPOT AND COVER YOUR HEAD. BRIDGES AND OVERPASSES DO NOT PROVIDE PROTECTION FROM TORNADIC WINDS. VEHICLES STOPPED UNDER BRIDGES AND OVERPASSES BLOCK TRAFFIC AND PREVENT OTHERS FROM REACHING SAFETY. IF YOU CANNOT SAFELY DRIVE AWAY FROM THE TORNADO... SEEK SHELTER IN A CULVERT...DITCH... OR A LOW LYING AREA AWAY FROM YOUR VEHICLE. LIE FLAT AND COVER YOUR HEAD WITH YOUR HANDS.

While both of the above warnings contain advice to get in the ditch if other alternatives do not exist (and the first specifically says not to try to outrun a tornado), both of them also contain wording advising the ditch if you cannot drive away from the tornado . Implicitly, this would certainly suggest driving out of the path as a preferable alternative to getting in the ditch – which is what Schmidlin argues based on his research. Variation among NWS offices. As noted above, there is considerable variation among NWS offices with regard to the issues discussed above. I examined the total distribution of warnings with regard to leaving vehicles for a ditch for all NWS offices that had five or more tornado warnings in the sample. Among these, there were fifteen local NWS offices that recommended leaving vehicles to get in the ditch under at least some circumstances in 70 percent or more of the warnings issued by that office that contained CTA statements. These fifteen offices accounted for a total of 175 of the 510 tornado warnings in the sample. In five of these offices, every warning in the sample that had a CTA statement included this advice. On the other hand, there were twenty-three local NWS offices that gave this recommendation in less than 30 percent of the warnings issued that contained CTA statements. These twenty-three offices accounted for a total of 167 of the 510 tornado warnings in the sample. Eleven of these offices never recommended abandoning vehicles to get in the ditch in any warning in the sample. Among offices that usually did not recommend leaving a vehicle to get in the ditch, there were two subcategories. The first, which accounted for nine offices, were ones that usually mentioned vehicles but did Farley: Tornado Warnings 21 not recommend abandoning them to get in a ditch. The other, which accounted for twelve offices, were ones that usually did not mention vehicles at all in their CTA statements. One question that might be raised about these offices would be whether or not they may have been avoiding the issue of vehicles due to the controversy about what is the best action to take. The other two offices reflected a mix of the two patterns, being about equally likely to mention vehicles but not recommend the ditch option and to not mention vehicles at all. There were three offices, which had a total of 27 warnings in the sample, that issued at least five warnings but showed no clear pattern – sometimes they recommended abandoning a vehicle for the ditch and sometimes they either did not mention vehicles or, in the case of one office, mentioned them but did not recommend abandoning them for the ditch. I examined the geographic distribution of the offices to see whether that might afford any clue as to why some offices almost always recommend abandoning vehicles to get in the ditch under at least some circumstances, and others almost never do. The distribution did not appear to be related to which NWS regional office the local office reported to, as there was a mix among all of the regional offices. However, it did appear that offices in the southeastern United States and the lower Mississippi valley were more likely to recommend abandoning vehicles to get in the ditch, while offices in the Great Plains region were less likely to do so. Eleven of the offices that usually recommended abandoning vehicles for the ditch under at least some circumstances were located in the general Southeast or lower Mississippi Valley regions, while ten of the offices that usually either did not mention vehicles or recommended abandoning them only for sturdy buildings were located in the Great Plains. As was suggested to me in two or three of my interviews with Warning Coordination Meteorologists (discussed in greater detail below), this may reflect the different terrain and storm structure of the regions. In the wide-open Great Plains, where storms are often higher-based and visibility is good, tornadoes are often visible at great distances. This may make driving away from the tornado a more viable option relative to getting out of the car and into the ditch. In the southeast and much of the lower Mississippi valley, 22 International Journal of Mass Emergencies and Disasters in contrast, the terrain is often hilly and wooded, roads curve and follow river valleys, clouds are low, and tornadoes more often occur at night. All of these conditions mean that tornadoes are much more difficult to see in these areas, so that driving away from one is not an option. In these situations, a last-minute decision may be more necessary, making a last resort-option such as a ditch more relevant. Hence, it is reasonable to argue that the best advice in one time and place is not the best advice in another. Interviews with Warning Coordination Meteorologists As described above, I conducted qualitative telephone interviews with seven Warning Coordination Meteorologists in seven different NWS offices in different parts of the country. These included offices that usually recommend abandoning vehicles for ditches, as well as offices that rarely do so. I asked a series of open-ended questions and used probes and follow-up questions as appropriate. The interviews lasted anywhere from around 20 minutes to over an hour, and averaged a little more than 30 minutes. In general, the interviewees were very interested in the topic and eager to talk about their efforts to communicate severe weather safety to the public. Familiarity with and reaction to debates about what to do if in vehicles. I began by asking whether they had heard any debates or disagreements about what was the best course of action when a person is in a vehicle and a tornado warning is issued. Six of the seven responded that they had heard such debates. The seventh said that he had not heard much debate, but responses to later questions indicated knowledge of the debates over Schmidlin’s research. Thus, the people I interviewed were familiar with debates over whether it is better to drive away from a tornado or to abandon a vehicle for the ditch. I next asked a very open-ended question asking them what they thought about these debates, and received a wide range of responses. One respondent said that neither option (getting in the ditch or driving away) was good – the only good option was to get out of the vehicle and into a sturdy building. This meteorologist stressed that situational awareness was the key – they way to avoid having to choose between getting in the ditch and trying to outrun a tornado Farley: Tornado Warnings 23 was to be aware of the situation and get to sturdy shelter before having to make that choice. Another said more briefly that the best option was a sturdy building, but that the decision came down to common sense and individual decisions in specific situations. One meteorologist supported the official guidelines, emphasizing that a ditch is a last resort but better than staying in a car. Two interviewees said that the best choice varies according to region or location, i.e. that driving out of the path of a tornado is best in the Plains, where tornadoes are easy to see, but should be avoided in the Southeast and other regions that are heavily wooded. Two other meteorologists did not say this in response to this question, but did volunteer it on multiple occasions in responses to later questions. It is notable that, in response to this question or another, the majority of meteorologists interviewed volunteered the idea that whether it is safer to drive away from a tornado or to abandon one’s vehicle for the ditch depends on region, location, or situation. Reactions to Schmidlin’s arguments. Subsequent questions asked the meteorologists whether they had heard about Schmidlin’s arguments that 1) it is safer to remain in one’s car than to get in the ditch and often possible to drive out of the tornado’s path, and 2) it is better to be in a car than in a mobile home. If they said yes, they were asked what they thought about the argument. All of the meteorologists interviewed indicated that they had heard at least something about both of these arguments. On the question of whether it is better to remain in the car and possibly drive out of the path versus getting out and into the ditch, three of the meteorologists indicated support for the official NWS position, i.e. abandon cars, the ditch is better as a last resort than staying in cars. Three of them said that which action is best depends on the situation, and one said that neither alternative is good. Two of those who saw the best choice as situation-dependent specifically said that it depends on the region or location, mentioning considerations such as whether it is open terrain where the tornado can be easily seen. In this kind of situation, they argued, it is best to drive out of the path, but that would not be the case for a tornado in an urban or wooded area, or at night. The third said that if you can see the tornado and it is a long distance away and moving away, or if you are in a different part of 24 International Journal of Mass Emergencies and Disasters a county from where the storm is moving, it would be foolish to abandon the car for the ditch. Those who supported the NWS position mentioned the risk of flying debris, even when the tornado is relatively weak, and two specifically mentioned having seen 2x4s driven through the doors of cars. These meteorologists also questioned Schmidlin’s conclusion that most tornadoes that are F2 or weaker will not move a car, citing experiences in damage surveys where cars were moved by weaker tornadoes. One mentioned driving in a microburst (severe, localized straight-line wind) which he estimated at 60 mph, and said that his car was lifted and turned around. Get out of mobile home and into car? On the question of whether it is better to get out of a mobile home into a car than to abandon the mobile home for an outdoor location, most of the responses indicated ambivalence. Four agreed that a car is safer than a mobile home in at least some situations, a fifth indicated agreement with Schmidlin’s position “to some extent,” and another mentioned that straight-line winds of as little as 80-90 mph can overturn a mobile home. Yet nearly all were unwilling to make a blanket recommendation to abandon a mobile home in favor of a vehicle. Concerns included being injured by falling objects while trying to get to the car, the risk of driving into the tornado while trying to escape, and the notion that even if vehicles are sometimes better than mobile homes, neither is really safe. Several stressed that a better option than the mobile home, car, or ditch is to get to a sturdy structure, and that the key to that is planning ahead and acting early. A good summary of the comments would be that the meteorologists’ own experiences in damage surveys supported Schmidlin’s findings that a vehicle is sometimes safer than a mobile home, but that they saw it as still risky enough that getting into a vehicle from a mobile home should not be a recommended action. Should safety recommendations be changed? The meteorologists were also asked a series of questions about possible changes in safety recommendations in light of Schmidlin’s research and the accompanying debates. I asked them 1) whether it is best to give the same advice for cars and mobile homes, or different advice, 2) whether it is better to mention getting in the ditch as a last resort, or just recommending to get in a sturdy building, 3) whether the safety Farley: Tornado Warnings 25 guidelines should be changed in any way, and if so, how, 4) whether the CTA should mention the possibility of driving out of the path of the tornado under some circumstances, and 5) whether it should mention that under some conditions or as a last resort, a vehicle might be safer than a mobile home. These questions elicited varied responses. Five of the seven respondents said the advice should be the same for cars and mobile homes, while one said different and one gave a mixed answer. However, later, more specific questions suggested more mixed opinions, as we shall see shortly. On the point of whether it is best to include mention of getting in the ditch or to only recommend getting in a sturdy building, four favored mentioning the ditch (often to cover situations where no building is available) while two favored mentioning only a sturdy building, because they did not view the ditch as safe. One who mentioned the ditch qualified his recommendation by stating that his area had deep ditches, and that he would rather be six feet underground than in a ditch. When asked the open-ended question about whether or not the safety guidelines should be changed in any way, about two-thirds said no and one-third said yes, but responses to other questions showed support for some specific changes. The change most supported, at least for some situations, was driving out of the path of the tornado. Four out of the seven meteorologists felt that in some situations this should be mentioned in CTA statements. The situations in which this was favored were open country with good visibility, good road networks, and daytime tornadoes. The idea of having different guidelines in different geographic areas was mentioned by several of these meteorologists. A complicating factor mentioned by at least two of the meteorologists was that their regions contained both areas of flat, open country with good visibility where driving away would be a viable option, and hilly and/or wooded areas where it would not. Hence, this made offering a blanket recommendation even for their regions of responsibility difficult. The other three meteorologists did not think that driving out of the tornado’s path should be mentioned in CTA statements, for varying reasons. One said it is simply too dangerous. However, one who did not favor including this recommendation in the CTA statement did acknowledge mentioning it at trainings of spotters, public officials, and emergency managers. His logic was that the general public 26 International Journal of Mass Emergencies and Disasters would often get confused and mistakenly drive toward rather than away from the tornado, while the groups at the trainings would be the people most likely to be driving in an area near tornadoes. The third meteorologist who did not want this recommendation in CTA statements was concerned about changing the recommendation based on one study by one researcher, but was open to the idea if it is supported by subsequent research elsewhere. Finally, in regard to the five questions outlined above, the idea of recommending that people leave mobile homes for vehicles received only limited support. Just two of the meteorologists favored including this recommendation in the CTA statements, while five opposed making this recommendation for various reasons. Variation in CTA statements and local discretion. One other thing I asked the meteorologists about was the variation in CTA statements that I had found in my research. I noted my finding that some offices mentioned vehicles and mobile homes in nearly every tornado warning, while others almost never did so. I also noted that some offices routinely gave the same advice for mobile homes and vehicles, while others either offered separate advice or mentioned either vehicles or mobile homes but not both in the CTA statements in their tornado warnings. And I mentioned that some offices routinely recommend abandoning vehicles for the ditch, while other offices rarely mention this advice. I asked the meteorologists what they thought accounted for these variations. In response to these questions, virtually every meteorologist mentioned the discretion that local NWS offices have in regard to what is included in their tornado warnings. One, for example, noted that there is a standard wording, but offices are free to edit it, add information, or delete parts of it. Moreover, the majority of the meteorologists interviewed clearly want such discretion at the local level, and view it as a good thing. One meteorologist stated, “The CTA will vary depending on where the tornado is – you need to use local expertise. You could automate so all tornado warnings have the same wording, but you should not do that- you need the human element to match the situations.” Another stated, “There’s been talk, but it’s difficult to standardize because of the differences – geography, number of people. They tried to do something like that Farley: Tornado Warnings 27 with the snow, but it didn’t work . . . It happens, but we’ll do what it takes to protect people here in ____. We have had 52 tornadoes and no fatalities and no serious injuries. . . I would hate to have directions to always do the same thing; that we can’t do what fits our local area, and then have people get killed.” Yet another said, “You can’t blanketly indicate before the fact what is the best decision . . . Part of our job is to feed information back uphill – so decisions are made by those in the field offices that encounter the situation.” It is notable that these very similarly-themed statements came from three different meteorologists in three different NWS offices. At the same time, this viewpoint was not unanimous; a meteorologist in another office (who did not favor changes in the guidelines) stated, “It has to do with we are one agency – we are designated by Congress as the sole warning voice – and we need to speak with a consistent voice.” Discussion, Policy Recommendations, and Future Research If Schmidlin’s arguments about the relative safety of vehicles and ditches are correct, then about two-thirds of warnings that mention vehicles contain advice that may not be the safest thing to do. At the same time, however, there is wide variation among NWS offices in the advice given in tornado warnings with regard to vehicles and mobile homes. For example, there are some offices that nearly always recommend abandoning vehicles for the ditch under at least some circumstances (a recommendation inconsistent with Schmidlin’s argument), while there are other offices that never mention getting out of a vehicle and into a ditch. At the same time, it is very rare for offices to advise people to drive out of the path of a tornado – very few even suggest it indirectly. In a sense, this is not surprising, because to do so would be to contradict the official NWS guidelines. Yet it would appear that a number of NWS offices do avoid advising people to get out of vehicles and into ditches, and it is evident from my interviews with Warning Coordination Meteorologists that 1) many feel that, in some situations and locations, it may be a good idea to drive out of the path of a tornado, and 2) some of them are mentioning this possibility in spotter trainings and safety talks. This study, like all research, has its limitations. For example, the number of NWS meteorologists interviewed is small, and in many 28 International Journal of Mass Emergencies and Disasters cases the number of warnings from any given NWS office is also small, even though the overall sample is sizable. Nonetheless, it seems fair to say that, on an informal level, Schmidlin’s research has had some impact, especially in portions of the country where tornadoes are usually visible at some distance and road networks are good. I reach this conclusion for several reasons. First, all of the meteorologists I spoke with were aware of Schmidlin’s research, and some stated to me that they agreed in part with the recommendations arising from it. In fact, the majority agreed that, under at least some circumstances, it was a good idea to drive out of the path of a tornado. Second, there are clearly regional variations in the extent to which the safety guideline recommending abandoning cars for the ditch as a last resort appears in CTA statements. It appears commonly in the southeast and lower Mississippi Valley, but much less often in the Great Plains. Third, these variations are consistent with comments I heard from the meteorologists I interviewed, many of whom said that a recommendation to drive away from the storm made sense in the Great Plains but not in the Southeast or the eastern or upper Midwest, hilly wooded areas where visibility is limited. Moreover, I heard this opinion from meteorologists in both kinds of region. A Great Plains meteorologist told me that driving out of the path of the storm made sense in his region, while one from the Southeast said that it did not in his region. But both said that the best thing to do would be different in another region. Together, these findings clearly show a recognition among Warning Coordination Meteorologists that in some situations it may be better to drive out of the path of a tornado, even if there are others in which it may be better to get out of a car and into the ditch if no other shelter is available. Since they are aware of Schmidlin’s research, it seems reasonable to assume that it has contributed to their thinking. At the same time, there is less willingness to change recommendations based on Schmidlin’s ideas about getting out of mobile homes and into cars to escape a tornado, even though several agreed that this can sometimes be useful. Additionally, there is also a sense among some that there is not enough research evidence to change long-standing safety recommendations. It is not surprising that the impact of Schmidlin’s research on day- to-day actions in NWS offices has been more informal than formal. Farley: Tornado Warnings 29

Some offices – mainly in the Great Plains, where Schmidlin’s advice seems most applicable—seem quite clearly to avoid advising people to get out of cars and into ditches in tornado warning CTA statements. Yet they do not directly contradict the official NWS position by including advice to drive out of the path of a tornado that can be seen. At the same time, a number of NWS meteorologists to appear to give this advice orally during safety talks and spotter trainings, if they are in areas with open views and good road networks. This would seem to be a less direct challenge to the official policy, since 1) it is given to a specialized audience that is arguably more knowledgeable and motivated than the general public, and 2) it is less public than the text of tornado warnings, which are widely disseminated and available on the Internet for anyone who wishes to see their wording. At the same time, it involves an exercise of discretion that is recognized and valued by the meteorologists responsible for severe-storm safety on the local level. It was clear from their comments that most of the meteorologists interviewed feel that they need the discretion to tell people what fits the needs of their local area. According to several of those that I interviewed, current NWS practice allows local offices to choose, from a number of pre-written CTA recommendations, which ones to include or not include in any given tornado warning. Moreover, there is freedom to edit the recommendations and delete or add material, so long as NWS guidelines are not directly contradicted. In practice, there is considerable variation in the extent to which such editing occurs. Recent research by Fine (2006) is helpful in understanding this variation. In the course of his observations of day-to-day operations in three NWS offices, he found considerable variation in office culture. Specifically, he found variation in the extent to which offices take pride in exercising their autonomy versus emphasizing the need to follow NWS guidelines and policies. I found this variation to some extent as well. Nonetheless, most of the meteorologists I interviewed indicated that they value their discretion (even if they vary in the extent to which they exercise it), and that the exercise of some discretion is essential to tell people what they need to know in their particular situations. This suggests two policy recommendations. First, in light of Schmidlin’s research findings, of variations in local NWS practices, 30 International Journal of Mass Emergencies and Disasters and of comments from the meteorologists I interviewed, NWS offices should probably be given the additional discretion to mention or not mention the possibility of driving out of the path of a tornado dependent on local geography. Hence, offices in areas with wide- open, flat terrain and mostly daytime tornadoes would have the discretion to recommend driving out of the path, while offices in areas with opposite characteristics would continue to recommend getting out of vehicles as is the case at present. At the same time, it might be noted in the official national safety guidelines that the best course of action may depend on whether or not the tornado is clearly visible and movement can be determined. Second, there should be no reduction in the amount of discretion that local NWS offices now enjoy. One meteorologist expressed concern to me that more standardization could be on the way, because of an incident in which one NWS office made a recommendation regarding highway overpasses that was contrary to the NWS position that they should be avoided because they do not offer safe shelter. This meteorologist told me that, “I heard that headquarters will come down with CTA statements they want us to use in the software. It has to do with we are one agency – we are designated by Congress as the sole warning voice – and we need to speak with a consistent voice.” I responded to this information with a probe asking, “Was this an official announcement, or did you hear it through the grapevine?” The meteorologist responded, “It was not official, but I was told by someone in our office that follows the process closely and manages our software, so I see it as reliable.” I asked several of the other meteorologists I interviewed if they had heard this, and they all said that they had not and that while there had been some efforts at standardization, they did not foresee any strict limits on their ability to edit the content of CTA statements. One meteorologist said this: “What I’ve heard isn’t about the CTA but about other formatting in warnings. Once a warning leaves us, others are putting out information – so warnings are becoming inflexible with regard to spaces, lines, number of words . . . So I think it’s about trying to preserve the integrity of the format, not a particular CTA.” Whether or not any effort is under way to reduce local NWS office discretion in the wording of CTA statements, this research suggests Farley: Tornado Warnings 31 strongly that to do so would be a bad idea. The meteorologists I interviewed recognize that different actions may be appropriate in different regions and in different situations, and this is reflected in regional variations in the content of the CTA statements. These variations parallel what the meteorologists told me about regional differences in what is the best action to take if you are threatened by a tornado while in a vehicle, and Schmidlin’s research supports the notion that in some situations – open country, good visibility – the official advice of getting in the ditch may not be the safest thing to do. For all these reasons, discretion on the part of local offices should be maintained, and indeed it should probably be increased, at least to allow a recommendation to drive out of the path of the tornado in situations where that seems to be the best course of action. Previous research has shown that how risk information is communicated to the public has important effects on how the public responds to natural hazards (Mileti, et al., 1993; Turner, et al., 1986; Farley, 1998). For this reason, the advice that is given should be the best that is possible, and this study, along with the research of Schmidlin and his colleagues, supports the notion that what is the best thing to do varies by location, among other things. Consistency, as discussed earlier in this paper, is also important, but it may be possible to be consistent at the local level, even if it is not at the national level. In Kansas, Oklahoma, or the Dakotas, it may be quite possible to consistently tell people that if it is daytime and you can see the tornado and can determine which way it is moving, to drive away from the path. This advice could be extended to mobile home residents as well – if the above conditions apply, leave the mobile home and drive out of the path of the tornado or to a safe shelter. In Tennessee or Alabama, on the other hand, it may be advisable to continue telling people that their vehicle is not safe and to get out of it, into a sturdy building if possible, but perhaps into a low spot away from the vehicle if no building is reachable. In either case, the advice is consistent at the local level. Moreover, such consistency at the local level can be reinforced by giving the same message in different local forums such as CTA statements in tornado warnings, guidelines on locally-oriented Websites and media outlets, advice during safety talks, school presentations, and spotter trainings, and 32 International Journal of Mass Emergencies and Disasters so forth. However, to give the best advice at the local level, some flexibility at the national level is a necessity. Finally, in future research, it would be helpful to generate more information on 1) the situations people are in when they receive tornado warnings (i.e. how many are driving, in buildings, outdoors, in mobile homes, etc.), 2) the ways in which they view the actions recommended in CTA statements (i.e. as blanket recommendations or as situation-specific ones), and 3) the degree to which people follow the recommendations given in CTA statements versus doing what they individually think best (e.g. how many actually get out of the car and into the ditch versus attempting to drive out of a tornado’s path, if no nearby sturdy building is available). For all of these kinds of information, quick-response survey or interview studies of people in areas covered by tornado warnings could generate useful information on peoples understanding of and response to CTA statements in tornado warnings.

Notes 1In 2007, the Fujita Scale was replaced by the , which includes more detailed rating criteria and different wind speed estimates. However, here we refer to the Fujita Scale, since it is what was in use when Schmidlin conducted this research and it was what Schmidlin used to estimate tornado strength. 2 The tornado warnings were obtained from the WX-TOR database maintained at the University of Illinois. This database consists of a set of files, produced weekly, that contain all of the tornado warnings issued in the United States for that week. The archive is maintained for a period of one year from the time a tornado warning is issued. The user can email the listserv at [email protected] to obtain the listing of all of the weeks for which files of tornado warnings are available. Once this list is obtained, the user can then email the listserv again to request the desired weeks. Through this procedure, I obtained all tornado warnings issued for anywhere in the United States from August 1, 2004 to July 30, 2005. Farley: Tornado Warnings 33

References Aguirre, Benjamin E. 2000. “Social Science and Severe Weather Warnings.” Pp. 98-108 in Storms Volume I . London, UK: Routledge. American Meteorological Society. 2000. Policy Statement: Tornado Preparedness and Safety, adopted by the AMS Council 28 February, 2000. Bulletin of the American Meteorological Society 81, 5 (May): 1061-1065. Burby, R.J., and F. Wagner. 1996. “Protecting Tourists from Death and Injury in Coastal Storms.” Disasters 20 (1):49-60. Carter, A. O., M. E. Millson, and D. E. Allen, 1989. “Epidemiologic Study of Deaths and Injuries due to Tornadoes.” American Journal of Epidemiology 130: 1209–1218. Christensen, Larry and Carlton E. Ruch. 1980. “The Effect of Social Influence on Response to Hurricane Warnings.” Disasters 4 (2):205-210. Daley, W. Randolph, Sheryll Brown, Pam Archer, Elizabeth Kruger, Fred Jordan, Dahna Batts, and Sue Ballonee. 2005. “Risk of Tornado-related Death and Injury in Oklahoma, May 3, 1999. American Journal of Epidemiology 161: 1144-1150. DiGiovanni, Clete, Barbara Reynolds, Robert Harwell, and Elliot B. Stonecipher. 2002. “A Prospective Study of the Reactions of Residents of an American Community to a Bioterrorist Attack.” Bethesda, MD: Defense Threat Reduction Agency. Drabek Thomas E. 1994a. “Disaster Evacuation and the Tourist Industry.” Boulder, CO: Natural Hazards Research and Applications Information Center, Institute of Behavioral Science, University of Colorado. DuClos, P. J., and R. T. Ing. 1989. “Injuries and Risk Factors for Injuries from the 29 May 1982 Tornado, Marion, Illinois.” International Journal of Epidemiology 18: 213-219. Edwards, Roger. 2004. “Tornado Safety.” World Wide Web, http:// www.spc.noaa.gov/faq/tornado/safety.html (downloaded January 20, 2006). Farley, John E. 1998. Earthquake Fears, Predictions, and Preparations in Mid-America. Carbondale and Edwardsville: Southern Illinois University Press. 34 International Journal of Mass Emergencies and Disasters

Fine, Gary Alan. 2006. “Shopfloor Cultures: The Idioculture of Production in Operational Meteorology.” Sociological Quarterly 47: 1-19. Golden, Joseph. 2002. “Comments on ‘Unsafe at Any (Wind) Speed? Testing the Stability of Motor Vehicles in Severe Winds.’” Bulletin of the American Meteorological Society 83: 1831-1833. Golden, Joseph H. and Christopher R. Adams. 2000. “The Tornado Problem: Forecast, Warning, and Response.” National Hazards Review 1 (2):107-118. Goldhammer, Robert G. 2002. “Comments on ‘Unsafe at Any (Wind) Speed? Testing the Stability of Motor Vehicles in Severe Winds.’” Bulletin of the American Meteorological Society 83: 1835-1837. Hammer, Barbara, and Thomas W. Schmidlin. 2002. “Response to Warnings During the 3 May 1999 Oklahoma City Tornado: Reasons and Relative Injury Rates.” Weather and Forecasting 17: 577-581. Lopes, Rocky. 1997. “The Alternate Dilemma: How to Explain and Encourage Counterintuitive Behavior.” Natural Hazards Observer 22: 3-4. Mikami, Shunji and Ken’ichi Ikeda. 1985. “Human Response to Disasters.” International Journal of Mass Emergencies and Disasters 3(1): 107-132. Mileti, Dennis S. 1997. The Role of Searching in Shaping Reactions to Earthquake Risk Information. Social Problems 44: 89-103. Mileti, Dennis S. and Joanne DeRouen Darlington. 1997. “The Role of Searching in Shaping Reactions to Earthquake Risk Information.” Social Problems 44:89-103. Mileti, Dennis S., JoAnne D. Darlington, Colleen Fitzpatrick, and Paul W. O’Brien. 1993. Communicating Earthquake Risk: Societal Response to the Revised Probabilities in the Bay Area. Fort Collins: Colorado State University, Hazards Assessment Laboratory and Department of Sociology. Mileti, Dennis S., and Colleen Fitzpatrick. 1993. The Great Earthquake Experiment: Risk Communication and Public Action . Boulder, CO: Westview Press. National Climatic Data Center. 2000. Storm Data and Unusual Weather Phenomena . May, 1999. Vol. 41, No. 5. Asheville, NC: National Farley: Tornado Warnings 35

Climatic Data Center. Available for download at http://www7. hcdc.noaa.gov/SerialPublications/SDPubs?action=createPDF (downloaded October 27, 2005). National Oceanic and Atmospheric Administration. (No date). “Frequently Asked Questions About Tornadoes.” NOAA Outlook. World Wide Web, http://www.outlook.noaa.gov/tornadoes/ q%26a.htm (downloaded January 20, 2006). National Weather Service. 1999. “Thunderstorms, Tornadoes, Lightning: Natures Most Violent Storms, A Preparedness Guide.” World Wide Web, http://www.nws.noaa.gov/om/brochures/ttl. pdf (downloaded January 20, 2006). National Weather Service Storm Prediction Center. 2004. 2004 Tornado Fatality Information. World Wide Web, http://www.spc. noaa.gov/climo/torn/2004deadlytorn.html. (downloaded October 27, 2005). Nigg, Joanne M. 1987. “Communication and Behavior: Organizational and Individual Response to Warnings.” Pp. 103-117 in Sociology of Disasters , edited by R. R. Dynes, B. DeMarchi, and C. Pelanda. Milan, Italy: Franco Angeli Libri. Perry, Ronald W., Michael K. Lindell, and Marjorie R. Greene. 1980. “The Implications of Natural Hazard Evacuation Warning Studies for Crisis Relocation Planning.” Seattle, WA: Battelle Human Affairs Research Center. Pfister, Neil. 2002. “Community Response to Flood Warnings: The Case of an Evacuation from Grafton, March 2001.” Australian Journal of Emergency Management 17:19-29. Rogers, George O. 1985. “Some Policy Implications of Human Components of Emergency Warning.” Pittsburgh, PA: Center for Social and Urban Research, University of Pittsburgh. Rogers, George O. and John H. Sorensen. 1989. “Warning and Response in Two Hazardous Materials Transportation Accidents in the U.S.” Journal of Hazardous Materials 22:57-74. Schmidlin, Thomas. 1997. “Closet, Car, or Ditch? The Mobile Home Dilemma During a Tornado.” Natural Hazards Observer 22: 1-3. Schmidlin, Thomas, Barbara Hammer, Paul King, Yuichi Ono, L. Scott Miller, and Gregory Thumann. 2002. “Unsafe at Any (Wind) Speed? Testing the Stability of Motor Vehicles in Severe 36 International Journal of Mass Emergencies and Disasters

Winds.” Bulletin of the American Meteorological Society 83: 1821-1830. Storm Prediction Center. 2002. 2001 Tornado Fatality Information. World Wide Web, http://www.spc.noaa.gov/climo/torn/ 2001deadlytorn.html (downloaded March 1, 2006). Thompson, Richard L., and Michael D. Vescio. 1998. “The Destruction Potential Index: A Method for Comparing Tornado Days.” Paper presented at 19 th Conference on Severe Local Storms, Minneapolis. Also available on Storm Prediction Center Web page, http://www.spc.noaa.gov/publications/thompson/dpi/ dpi.htm (downloaded January 20, 2006). Turner, Ralph H., Joanne M. Nigg, and Denise Heller Paz. 1986. Waiting for Disaster: Earthquake Watch in Southern California . Berkeley: University of California Press.