Exploring probable reasons for record fatalities: Bimal Kanti Paul Mitchel Stimers

Nat Hazards (2012) 64:1511–1526
DOI 10.1007/s11069-012-0313-3
Exploring probable reasons for record fatalities:
the case of 2011 Joplin, Missouri, Tornado
Bimal Kanti Paul • Mitchel Stimers
Received: 10 June 2012 / Accepted: 24 July 2012 / Published online: 11 August 2012
Ó Springer Science+Business Media B.V. 2012
Abstract On the evening of 22 May 2011, an EF-5 tornado tore a path six miles long
across Joplin, Missouri, USA, killing 162 people as it passed through the heart of the city.
This tornado stands as the deadliest single tornado to hit the United States since modern
recordkeeping began in 1950, surpassing the tornado of 8 June 1953 that claimed 116 lives
in Flint, Michigan. The record number of deaths caused by the single tornado in Joplin was
far higher than the average annual number of US tornado deaths over the last three
decades. This study explores the reasons for the high number of fatalities caused by the
2011 Joplin tornado. Questionnaire surveys administered among tornado survivors and
informal discussions with emergency management personnel and others suggest that five
reasons are associated with the high number of tornado fatalities experienced in Joplin: (1)
the sheer magnitude of this event; (2) its path through commercial and densely populated
residential areas; (3) the relatively large size of damage area; (4) the physical characteristics of affected homes in Joplin; and (5) the fact that some residents ignored tornado
warnings. Several recommendations are offered, the implementation of which should
reduce future tornado fatalities not only in Joplin, but elsewhere in the United States.
Keywords Joplin tornado Tornado fatalities Tornado warnings
Compliance with warnings Housing characteristics
B. K. Paul (&)
Department of Geography, Kansas State University, Manhattan, KS 66506, USA
e-mail: [email protected]
M. Stimers
Department of Natural and Physical Sciences, Cloud County Community College,
Junction City, KS 66441, USA
Nat Hazards (2012) 64:1511–1526
Fig. 1 Estimated tornado path within the city of Joplin, MO. Source: Compiled from various sources
1 Introduction
On the evening of 22 May 2011, a tornado killed 162 people and injured over 1,000 in the
city of Joplin, Missouri, USA.1 As a part of a larger late-May tornado outbreak sequence, it
initially touched down in a rural area just east of the Kansas/Missouri state line and caused
minor damage. The tornado entered Joplin at its south-west corner at 5:41 pm CDT
(Fig. 1). It was approximately a half-mile wide when it hit Joplin near Schifferdecker
Avenue. At this point, the tornado was rated EF-2 to EF-3 on the Enhanced Fujita Scale
(EF Scale) (NOAA 2011). It grew to a width of three quarters of a mile between 26th and
East 20th Street (Turner and Hacker 2011). Near South Duquesne Road to near the
Interstate 44 and 249/71 junction, the tornado began turning right and moving south-east
across Interstate 44 where it weakened; nonetheless, vehicles were flipped and mangled
near the US Route 71 (Fig. 1). The weakened tornado, rated EF-1 to EF-2, continued to
track into the rural areas of south-eastern Jasper county and north-eastern Newton County
where damage was generally minor to moderate. The tornado lifted east of Diamond at
6:12 pm CDT. The tornado’s total track length was at least 22.1 miles (35.6 km) long. It
Various sources report different fatality totals for the 2011 Joplin, MO, tornado. As of May 2012, the
Storm Prediction Center (SPC) reported a death toll of 158 (SPC 2012). SPC’s total includes only deaths
directly related to the event. However, one or more indirect deaths are also caused by this tornado. For
example, a policeman was struck by lightning and killed while assisting with recovery and cleanup efforts
the day after the tornado. Several people died from injuries more than a month after the tornado. An
evaluation report of the National Weather Service (NWS) claims that there were 159 direct fatalities and 3
indirect fatalities caused by the 2011 Joplin tornado (SPC 2012).
Nat Hazards (2012) 64:1511–1526
travelled a six-mile-long path across the city and achieved wind speeds that exceeded 200
mph (320 kph), with no fewer than four of these six-mile long paths rated EF-5 on the EF
Scale. The tornado destroyed much of the city’s south side and completely flattened many
densely populated neighbourhoods (Turner and Hacker 2011).
The Joplin tornado destroyed nearly 7,000 homes and damaged hundreds more (Joplin
Globe 2011a, b), with the damage area covering 1,800 acres (2.81 square miles), equivalent to approximately one quarter of the city of 50,150 residents (US Census Bureau
2012).2 About one-third of the population of the city was in the tornado path or an
estimated 15,000 persons (Simmons and Sutter 2012). The Joplin tornado caused an
estimated $3 billion in insured losses, not including uninsured damage to structures (Paul
and Stimers 2011).
The 2011 Joplin tornado stands as the deadliest single tornado to hit the United States
since modern recordkeeping began in 1950, surpassing the tornado of 8 June 1953 that
claimed 116 lives in Flint, Michigan (NOAA 2011). Given the magnitude and size of the
tornado and the density of the population in the area it hit, the total number of fatalities
caused by this event was not unexpected (Simmons and Sutter 2012); nonetheless, the
event set the record number of deaths. For the last 30 years, tornado fatalities in the United
States have averaged around 55 per year (Simmons and Sutter 2011), meaning that the
2011 Joplin tornado caused nearly three times more fatalities than this annual average.
Why then did so many people die in Joplin? Were there problems with the tornado warning
system? Were there problems with Joplin residents heeding any given warnings? Were
there other contributing factors?
This study explores the possible reasons for the large number of fatalities caused by the
2011 Joplin tornado and provides some perspective on the death toll. Evidence will be
provided with the help of relevant information collected from both primary and secondary
sources. Though this evidence is suggestive and not necessarily entirely conclusive,
findings of this study will aid policy makers and private and public emergency management agencies in reducing fatalities and injuries from future tornados in Joplin and elsewhere in the United States.
As background for this study, an overview of risk factors associated with tornado
fatalities is presented next, followed by a section that describes the data collection procedure for this study. Probable reasons for the high number of deaths and the conclusion of
this study are presented in the final two sections.
2 Risk factors for tornado fatalities: an overview
Existing literature (e.g. Schmidlin 1993; Brenner and Noji 1995; Schmidlin and King
1997; Daley et al. 2005; Sutter and Simmons 2010; Simmons and Sutter 2011) suggests
that tornado fatalities in the United States are associated with an event’s magnitude,
frequency, timing, season of occurrence and path (or track). Magnitude refers to the
strength or severity of a tornado and since 2007 has been expressed in terms of the EF
Scale. In general, the greater a tornado’s magnitude, the greater the potential for fatalities,
injuries and damage to property (Paul 2011; Simmons and Sutter 2011). In absolute terms,
however, more people have died from EF-0 through EF-4 tornados because EF-5 tornados
occur less frequently (Simmons and Sutter 2011, 2012). Like magnitude, in general, a
The average tornado damage area in the United States is less than one square mile (Simmons and Sutter
Nat Hazards (2012) 64:1511–1526
higher frequency of tornado occurrence usually results in more fatalities; however, frequency is not relevant for the present study because the focus here is on only one tornado.
Timing is also considered a contributory factor for fatalities caused by tornadoes (e.g.
Ashley 2007; Simmons and Sutter 2008, 2011, 2012; Sutter and Simmons 2010). Tornados
are significantly more lethal at night than during the day primarily because people are
likely to be home, they may not hear tornado warnings, or they may be asleep. For
example, Simmons and Sutter (2008) found through regression analysis that expected
fatalities are 64 % lower for tornados that occur during the day than for ones that occur at
night. They reached this conclusion after controlling for other factors such as storm path
characteristics, magnitude and issuance of a tornado warning.
Using a data set for the 1900–2007 period, Simmons and Sutter (2011) claimed that
most tornado fatalities in the United States occur between March and June. However,
tornadoes in these months are not particularly dangerous. According to their analysis,
February is the most dangerous month for fatalities per tornado (0.29). Tornadoes are
particularly benign in the late summer, with lowest fatalities per tornado occurring in July.
The difference between February and July is quite remarkable. The fatality rate in July
(0.01) is about 1/25 of the February rate (Simmons and Sutter 2011). They further found
that May to October is characterized by the lowest fatality rates per tornado, while the
highest rates are in the months from November to April. This is primarily because people
of tornado-prone areas are more prepared for tornadoes during the tornado season (March–
June) than non-tornado season. They also take warnings more seriously during the former
season and as a result compliance with warnings is also high during this season (Simmons
and Sutter 2011).
Relevant studies (e.g. Balluz et al. 2000; Sutter and Simmons 2010; Stimers 2011) also
suggest that the location of a tornado path is an important determinant of deaths caused by
the event. Not surprisingly, if the path of the tornado passes over highly populated areas,
the number of fatalities is likely to be higher than if it passes over less populated areas. In
analysing tornado events between 2000 and 2009, Stimers (2011) calculated that 981 (8 %)
of the 12,657 events that occurred in the United States during that period passed through
communities. He also reported that this percentage is even lower for states located in the
Great Plains because the overwhelming majority of tornados (during the period studied)
touched down in open spaces far from any major populated areas (Stimers 2011). In the
Great Plains region as well as in other regions of the United States, population centres tend
to be widely dispersed. It is worth mentioning that the 2011 Joplin tornado travelled over
an area of the city that had an average population density of 1,500 people per square mile
(NOAA 2011).
Tornado-related deaths also depend on two non-structural factors: (1) the nature of
tornado warnings and (2) the compliance (or lack thereof) with those warnings (Daley et al.
2005). If tornado warnings are not issued in a timely manner (or not issued at all), the
population at risk is less able to seek safe shelter, which may contribute to a greater number
of deaths and injuries (Balluz et al. 2000; Simmons and Sutter 2008). Tornado warnings
issued ahead of the event coupled with widespread compliance have saved many lives.
Simmons and Sutter (2011) claim that a warning with a lead time of 5 min or less reduces
fatalities by 19 % but warnings with a lead time of 6–10 min reduces fatalities by 52 %.
Largely due to improved warning systems, annual tornado fatalities in the United States
have consistently decreased over the last 50 years (Ashley 2007). Tornado fatalities
declined from 2.9 per million in the mid-1920s to around 0.2 per million in recent years
(Simmons and Sutter 2011).
Nat Hazards (2012) 64:1511–1526
Although not considered in this study, several individual and household characteristics
are also associated with tornado fatalities. These characteristics may dictate who receives
hazard warnings as well as who complies with such warnings (Paul and Dutt 2010). For
example, the less affluent and less educated generally have reduced access to warning
systems, thus impeding their ability to receive such warnings, and as a result, tornados pose
a greater threat to their lives (Chaney et al. 2012). Further, the less affluent are more likely
to own and/or live in an old house, which may not have a basement and/or safe room. A
‘‘safe room’’ is made of thick concrete walls and ceilings reinforced with steel bars. Safe
rooms can be installed in a basement, in the centre of the house at ground level of homes
without a basement, or under a garage (the garage floor then serves as the shelter’s ceiling)
(Murphy and Sherry 2003).
As indicated, older houses, houses with walls not anchored to the foundation and houses
without a basement are risk factors associated with increased tornado mortality rates
(Balluz et al. 2000). In addition, the less affluent are more likely to live in pre-fabricated,
manufactured, or modular (‘‘pre-fab’’ or ‘‘mod homes’’) or mobile homes (Sutter and
Simmons 2010). Based on an analysis of tornado data collected between 1996 and 2007,
Sutter and Simmons (2010) reported that, in the United States, the probability of tornado
fatalities in mobile homes is 10 times (or more) greater than that in permanent homes (see
also Brooks and Doswell 2002; Brown et al. 2002; Daley et al. 2005; Simmons and Sutter
2006; Ashley 2007; Schmidlin et al. 2009; Chaney and Weaver 2010). Between 1985 and
2007, 536 of 1,240 (43.2 %) of US tornado fatalities occurred in mobile homes, although
mobile homes made up only 8 % of US housing units in 2000 and only 7 % of the
population lived in mobile homes (Sutter and Simmons 2010).
As noted, mobile homes accounted for 43 % of all tornado fatalities that occurred
during the 1985–2007 time period. Slightly over 31 % of all deaths occurred in what the
NWS describes as permanent homes. Other locations for fatalities include 9 % in vehicles,
5.3 % in schools and churches and 5 % each in businesses and outdoor locations (Simmons
and Sutter 2011). Daley et al. (2005) reported that mobile homes situated in the path of a
tornado are consistently associated with a particularly high risk of death or injury. In
analysing deaths and injuries in Oklahoma after the 3 May 1999 tornado, Brown et al.
(2002) claimed that people living in a mobile home, an apartment complex, or outdoors
had a higher fatality rate than those living in permanent houses. Additionally, Sutter and
Simmons (2010) reported that a larger proportion of fatalities occur in mobile homes
during less powerful tornados (e.g. EF-1, EF-2, or EF-3). In other words, EF-1 through EF3 tornados are potentially lethal for residents of mobile homes (which are structurally
weaker than traditional ‘‘stick-built’’ or brick homes), yet the researchers observed no
significant difference in the number of fatalities caused by EF-4 and EF-5 tornados vis-àvis people living in mobile versus permanent homes. In terms of the timing of a tornado,
Schmidlin et al. (2009) and Sutter and Simmons (2010) claim that mobile home fatalities
are especially likely to occur at night, particularly from midnight to 6:00 am.
It is evident from the preceding discussion that tornado-related deaths not only are
associated with the physical characteristics of the event, but are also a function of the
social, economic and demographic factors of people exposed to tornados (Cutter et al.
2003; Wisner et al. 2004; Cutter and Finch 2008; Simmons and Sutter 2011, 2012). Risk
factors associated with tornado fatalities can broadly be divided into two groups: (1)
structural (e.g. housing characteristics and availability of public tornado shelters) and (2)
non-structural. The latter can be subdivided as (1) warning characteristics, including
compliance with warnings; (2) physical characteristics of the tornado itself (e.g. frequency,
magnitude and timing of occurrence); and (3) individual and household characteristics
Nat Hazards (2012) 64:1511–1526
(e.g. age, education level, gender, income and immigration status). It is important to note
that there is considerable overlap among risk factors for these different groups. For
example, whether a person owns a permanent house largely depends on his/her economic
situation. Likewise, economic conditions may determine access to tornado warning systems; those without a television or access to the Internet may not be as prepared to seek
shelter as those who own a television or can access online weather reports.
3 Methods
Data and information used in this study were drawn from multiple sources, with the major
source of data coming from a questionnaire survey administered to tornado survivors of
Joplin, Missouri, selected through a snowball sampling procedure. The survey was conducted between 24 June 2011 and 16 February 2012. A prestructured interview schedule
was used to collect relevant information from residents who were in the city on the evening
of the tornado. The schedule used contained three types of information: (1) questions
related to knowledge regarding tornado warnings and compliance with these warning
messages; (2) questions regarding property damage, injuries sustained, the number of
deaths in household, past tornado experience and characteristics of the residential structure;
and (3) demographic and socio-economic questions (e.g. the respondents’ gender, age,
marital status, educational attainment, employment status and annual household income).
The primary purpose of the questionnaire was to explore the probable reasons for the
high number of fatalities caused by the 2011 Joplin tornado. This was partially accomplished by examining the nature and extent of warnings residents of Joplin received prior to
the tornado’s touchdown and how they responded to those warnings. Such an examination
was necessary because many people, including the authors of this study, after hearing
about tornado fatalities in Joplin, initially suspected that tornado warnings were not adequate and/or residents of the city did not appropriately respond to the warnings. The
disaster literature (e.g. Balluz et al. 2000; Paul et al. 2003; Sherman-Morris 2005; Simmons and Sutter 2011, 2012) also suggests a strong positive correlation between tornado
fatalities and lack of warnings, inadequate warnings, or non-compliance with warnings.
Data collected contained no identifiers; resident participation in the study interview was
voluntary and non-coercive, and participant confidentiality was respected throughout the
study. Face-to-face interviews were conducted among 133 respondents by six trained
personnel. Interviews were conducted in various locations within the Joplin area, including
malls, retail outlets, libraries, restaurants, pubs, residences, golf courses, gas stations,
temporary shelters and offices of emergency agencies responsible for processing disaster
aid applications. In addition, eight interviews were completed using a combination of
telephone and the social media Internet site Facebook, for a total of 141 responses.
Although the intent was to survey every available tornado survivor, not many were
present at the interview locations. Additionally, seven survivors declined to participate
because they had scheduled appointment in other places, and several exited before they
were approached for interviews. Each interview lasted approximately 10–15 min and the
research team went to Joplin four times to conduct interviews.
Relevant data were also collected from local emergency management personnel for
Joplin and surrounding areas, Joplin city officials, fire and law enforcement dispatchers,
volunteers responsible for clearing debris, local professionals (e.g. lawyers, teachers and
businessmen) and residents of Joplin and neighbouring areas through informal discussions
and conversations. The Joplin Globe—the local daily newspaper—and other print media
Nat Hazards (2012) 64:1511–1526
provided valuable information for this study. This research attempted to synthesize the
most common responses received from the various sources.
Spatial data used in this study were collected from the City of Joplin mapping division,
the Federal Emergency Management Agency (FEMA) and the US Census Bureau. These
data were analysed within a Geographic Information System (GIS) to create the warning
siren buffer. Among the spatial data collected was a shape file of 8,440 points data that
originated from FEMA.
4 Results
Of the 141 respondents, 86 (61 %) were male and the mean age of all respondents were
44 years. Slightly over 54 % of the respondents were married at the time of survey, 28 %
were unmarried and the remaining 18 % were divorced, legally separated, or lost their
spouse. The questionnaire survey reveals that nearly 3 % did not graduate from high
school, 62 % completed high school and the remaining 35 % had an education above the
high school level. Nearly 46 % respondents indicated they were employed full-time and
another 20 % were employed on part-time basis. Only 8 % were unemployed and the
remaining respondents were students, disabled, retired, or home makers. In terms of yearly
household income, slightly over 36 % respondents reported an income under $20,000,
while nearly one-fourth of the total respondents earned more than $59,999 per year.
Prior to the 2011 tornado, all respondents interviewed in this study were residents of
Joplin and the overwhelming majority of them were still living in the city at the time of
questionnaire survey. Nearly 49 % of all respondents reported damage to their home and/or
other possessions. This means slightly over half of the respondents were from outside the
path of the devastating tornado. Two such respondents experienced total destruction of
their Main Street real state properties which were in the direct path of the tornado. Of the
141 respondents, 84 (60 %) were at home when the tornado struck Joplin, MO and the
remaining 57 (40 %) were outside home at different places (e.g. churches, place of works,
friend’s and/or relatives’ houses within Joplin, retail stores, on the roads, restaurants, gas
stations, the movie theatre and Webb City, which is located immediately north and
adjacent to Joplin). The questionnaire survey reveals that a total of 26 persons sustained
injuries from 18 households. Among the injured, no one was outside the houses or
buildings, but their houses were along the tornado paths. Only three respondents reported a
death in their households.
In addition to present empirical evidence regarding the nature of warnings residents of
Joplin received prior to the tornado touchdown and the extent of compliance with such
warnings, the subsequent sub-sections will also analyse the housing characteristics of this
community with a view to find possible reason(s) for the high tornado fatalities in Joplin.
4.1 Tornado warnings
As noted, a questionnaire survey was administered among survivors of the Joplin tornado
to ascertain their knowledge regarding such warnings. The survey data revealed that nearly
90 % respondents received tornado warnings before it struck Joplin, MO. Exactly the same
percentage of respondents was aware that the emergency sirens had sounded prior to the
advancing tornado. Apart from sirens, a considerable proportion of respondents also
received tornado warnings from other sources, such as commercial and weather radio,
television, cellular and landline telephone, e-mail and word-of-mouth. Respondents were
Nat Hazards (2012) 64:1511–1526
Fig. 2 Warning siren locations with a one-mile buffer applied
also warned by their neighbours who came to their homes, the announcement of a tornado
warning coming over retail stores’ intercom system, looking at the sky, or through local
Of the respondents who were unaware of any warnings, the most common reason given
for their lack of awareness was that they were inside their homes and not watching
television or listening to a radio. They were also not able to hear the warning sirens
because they were inside their homes. Warning sirens are not designed to be heard from
inside a house, particularly when doors and windows are closed.
Although a 90 % siren warning awareness rate can be considered very high, the reasons
for the 10 % who remained unaware are still noteworthy, and there appear to be several
reasons. The first reason concerns siren audibility. Keith Stammer, Director of Jasper
County Emergency Operations Center (JCEOC), informed us that all 28 tornado sirens
located within the city of Joplin and surrounding area sounded on the evening of 22 May
2011 without any failure. However, the entire affected area was not within audible range of
the sirens. Fig. 2 shows the location of all 28 sirens, along with a one-mile buffer applied
to each site, indicating the audible range of the tornado warning siren.3 The figure suggests
that about 10 % of the total area of Joplin and its vicinity are located outside the audible
range of the sirens. Fig. 2 indicates that an area near the eastern-most section of the city in
the tornado’s path was not covered by a siren network. It is reasonable to expect that
people in that area should have been able to hear the siren as the wind at that time was
This audible range was suggested by Keith Stammer of JCEOC.
Nat Hazards (2012) 64:1511–1526
Table 1 Tornado lead time (in
min) by respondents’ opinions
regarding adequacy of the time to
seek safety (N = 126)
Lead time
Provided enough time
Number (%)
13 (13)
31 (32)
37 (29)
27 (28)
32 (25)
20 (20)
23 (18)
Number (%)
26 (22)
7 (7)
8 (6)
98 (100)
126 (100)
moving in a west-to-east direction. This movement created elliptically shaped buffers
rather than circular ones.
Normally, a siren can serve areas within its audible range, assuming there is little or no
topographical or artificial interference with the sound waves produced (Current and
O’Kelly 1992). However, the topography of Joplin consists of many low hills and valleys,
which might have created barriers to coverage of the entire area within the audible range.
Secondly, the Joplin tornado contained pounding rain and hail; conditions may have
contributed to some residents’ lack of awareness of the siren.
Because sufficient lead time between the warning and the arrival of the tornado is
critical for taking shelter (and thusly for reducing deaths and injuries), this study examined
the lead time Joplin residents experienced prior to tornado touchdown. Table 1 contains
this information in the form of respondent opinions regarding whether the warnings provided them sufficient time to seek safety. Although lead time is divided into five categories,
the opinions of the respondents are dichotomized as: ‘‘had enough time to seek safety’’
(yes) and ‘‘not enough time to seek safety’’ (no). Information collected from the survey
instrument revealed that the reported tornado warning lead time ranged from one to
27 min. Table 1 shows that 95 (76 %) of the 126 respondents had lead time 15 min or less
and the remaining 31 respondents (24 %) received warnings more than 15 min before
tornado hit Joplin. It is interesting to note that a warning with a lead time of 5 min or less
reduces fatalities by 19 % and a warning in the 11–15-min interval reduces fatalities by
33 %. A lead time over 15 min may not reduce fatality; rather, it may increase expected
fatalities (Simmons and Sutter 2011).
Table 1 further shows that the overwhelming majority of respondents surveyed agreed
that the tornado warnings provided enough time to seek shelter. Only 28 respondents
indicated they felt they did not have sufficient time to take shelter despite the fact that 12 of
them had a lead time of at least 5 min. Respondents were also asked to indicate whether
they felt the National Weather Service (NWS) did an adequate job of warning and/or
preparing Joplin residents for the approaching tornado. Slightly over 84 % (106 of 126)
respondents answered this question affirmatively.
In analysing response to the 27 April 2011 tornado outbreak across northern Alabama,
Klockow (2011) reports that many tornado survivors had difficulty determining whether
they were at risk or not. They had a hard time determining a storms trajectory because the
looping radar images being shown on the television were too short and did not clearly show
where the storm was moving. To a considerable extent, these problems occurred because of
Storm Based Warnings (SBWs) use as opposed to traditional county-based warnings. The
SBWs were adopted by the NWS in October 2007, and this system warns only areas directly
in the path of a tornado instead of entire counties (Simmons and Sutter 2011, 2012).
Nat Hazards (2012) 64:1511–1526
The 2011 Joplin tornado warning also used the SBW system and the entire path of the
storm through Joplin was contained within the SBW polygons (NOAA 2011).
However, no respondents or any other persons in Joplin raised any concerns with the
SBW system. Moreover, conversations with non-respondents and emergency personnel
working in Joplin (at the time the questionnaire surveys were administered) revealed that
advance warnings of the tornado were issued in a timely manner and reached almost all the
people in the affected community and its immediate environs. The NOAA assessment
report on the 22 May 2011 Joplin tornado, and articles published on tornado warnings in
the daily Joplin Globe also support this view (NOAA 2011). Simmons and Sutter’s recent
book on the 2011 deadly tornado season also concur with the above contention (Simmons
and Sutter 2012).
4.2 Compliance with tornado warnings
Of the 126 respondents who received the tornado warnings, 97 (77 %) stated that they did
comply with warnings and took shelter after hearing sounds of siren. This response rate can
be considered high.4 At least three respondents who complied with the warning indicated
they went outside to see the tornado, to visually verify the tornado threat before taking
shelter; such action is not recommended as it involves considerable risk. Several others did
not react to the first warning sirens, which sounded at 5:11 pm CDT, but did seek shelter
after hearing the second round of sirens beginning at 5:38 pm CDT. The first sirens
provoked some respondents to seek additional information from televised reports, social
media via the Internet, or going outside to view the tornado and/or weather conditions for
themselves. A similar finding was also included in the NOAA report on this tornado
(NOAA 2011).
Having no faith in the warnings and consequently not seeking shelter was reported by
10 of the 29 respondents. Four respondents who did not seek shelter thought that there
might be some hail and/or wind damage that would result from the storm, and one
respondent stated: ‘‘It would not pass through our town.’’ Only two respondents indicated
that they did not understand the warning instructions. One of these two thought the second
round of sirens meant the tornado had passed. Three respondents explained that they did
not receive any warning as they were in their car returning home. The remaining eight
respondents who did not take shelter after receiving the warnings stated that emergency
siren activation is such a common occurrence in this area that they hesitated in their
response. One of them stated: ‘‘We hear tornado sirens all the time, and nobody pays
attention to them.’’ This illustrates an example of the indifferent attitude of some Joplin
residents towards tornado warnings (also see Turner and Hacker 2011).
Historically, tornado activity around the Joplin area is well above the average for
Missouri and it is 161 % greater than the national average (Joplin, Missouri (MO) Profile
2011). The city has experienced two other tornadoes since May of 1971. However, the vast
majority of tornados occurring in the Joplin area are low magnitude, ranging from EF-0 to
EF-2. In addition, tornados in this area usually display short-tracks, and are often fastmovers on the ground. Even the 2011 Joplin tornado was rated as only an EF-0 just a few
miles west of the city (NOAA 2011). Given the high frequency of tornados and their low
magnitude, it is not unusual that some people in Joplin did not take the severe weather
warnings seriously.
In reviewing studies dealing with response to tornado warnings in the United States, Simmons and Sutter
(2011) found that the reported response rates range from just under 30 % to almost 90 %.
Nat Hazards (2012) 64:1511–1526
Table 2 Areas used for tornado shelter by residential and non-residential structures (N = 97)
Residential structure
Number (%)
Interior room
18 (25)
7 (28)
25 (26)
9 (13)
6 (24)
15 (16)
Non-residential structure
Number (%)
15 (20)
13 (18)
Tornado shelter
9 (12)
4 (6)
5 (20)
5 (5)
4 (16)
Crawling space
2 (3)
13 (13)
12 (12)
2 (3)
72 (100)
15 (16)
3 (12)
Safe room
Number (%)
25 (100)
4 (4)
4 (4)
2 (2)
2 (2)
97 (100)
Includes under non-resident structure
Places used for shelter by the 97 respondents who complied with the tornado warnings
are listed in Table 2. These are divided into two groups: (1) residential structures (e.g.
mobile homes and permanent houses, including duplexes and apartment complexes) and
(2) non-residential structures (e.g. churches, restaurants, retail stores, and hospitals).
Table 2 suggests that the area most widely used as shelter (25 of 97 respondents) was an
interior room, specifically, a room inside the house without windows. Interior rooms of
non-residential buildings, primarily churches where they were attending Sunday services,
were used by seven of these respondents. The remaining 18 respondents used an interior
room of their residential building. Fifteen respondents reported seeking shelter in basements, six of them used church basements for shelter and one sought shelter in a neighbour’s basement. Another 15 respondents sought shelter in a closet, and bathtubs and
bathrooms were used for safety by 13 and 12 respondents, respectively.
Five respondents, who were in a gas station, restaurant, or retail store, used a commercial cooler as shelter. Four respondents went to nearby tornado shelters; two of these
four were living in mobile homes. Another four respondents were already in a car, or
moved to a car and drove perpendicular (either to the right or left) of the approaching
tornado for safety. The NOAA assessment report also claims that some Joplin residents
‘‘mistakenly drove their vehicles into the tornado path, but somehow lived to tell of it’’
(NOAA 2011, 2). Two respondents went to a safe room and another two to the home’s
crawl space for protection (see Table 2). Use of other forms of protection (e.g. standing
behind a table, desk, or bench) was not reported by any respondents. Some respondents
mentioned that they covered themselves with a blanket or mattress. As noted, 26 people
were injured and they took shelter in the interior rooms, closets, bathtubs, or bathrooms of
both residential and non-residential buildings. People who took shelter in other places did
not sustain any injuries. All three deaths occurred in the interior room, closet, or in
Evidence provided above suggests that the tornado warning system in Joplin performed
at least adequately on 22 May 2011; the overwhelming majority of residents of this city
obtained warnings via many sources, including commercial and weather radio, telephone
Nat Hazards (2012) 64:1511–1526
Table 3 Reported construction year of residential structures by period and type (N = 141)
Number (%)
Single detached
Number (%)
Number (%)
Before 1960
Number (%)
Number (%)
18 (62)
22 (58)
42 (89)
24 (89)
106 (75)
5 (18)
13 (34)
2 (4)
2 (7)
22 (16)
Mobile home
3 (10)
3 (10)
3 (8)
29 (100)
38 (100)
3 (7)
47 (100)
6 (4)
1 (4)
7 (5)
27 (100)
141 (100)
calls, and television. In addition to tornado sirens, weather radio began providing tornado
warnings at 5:17 pm CDT—24 min before the tornado hit the city. Additionally, response
to the tornado warning was also satisfactory.
4.3 Housing characteristics
Adequate warning is not effective if those at risk have no access to shelter (Balluz et al.
2000). The lack of a basement in a house and/or the inability to gain access to a nearby
shelter is an important factor for responding to tornado warnings. Only eight respondents
had a basement in their home. The absence of basements in houses in Joplin is not
surprising; according to the Jasper County Assessor’s Office, nearly 78 % of houses in the
county lack basements, due in part to rocky ground and a high water table (Joplin Globe
2011b). Joplin has an even lower percentage of basements compared to Jasper County
communities as a whole. The city is located near the borders of four states (Missouri,
Kansas, Oklahoma, and Arkansas); this area contained numerous lead and zinc mines that
were operational from the late 1800s and early 1900s through the early 1950s. There are
many locations situated over old mine shafts, which has resulted in subsidence problems;
obviously, this makes the construction of basements difficult and impractical.
Several studies (e.g. Grazulis 1993; Balluz et al. 2000) have recommended the use of an
interior room in the house as an alternative for people without basements. However,
depending on the specific conditions of such a house, an interior room may fail to provide
adequate protection, particularly for high magnitude tornadoes. Most of the houses in
Joplin are old; nearly 53 % of all respondents reported living in homes constructed more
than 30 years earlier (Table 3). Moreover, older houses were constructed according to the
standards of the time, which are far less rigorous than today’s much stricter building codes.
According to Keith Stammer, Director of JCEOC, many of these older houses are not
secured to their foundation; some do not even have a foundation. Field survey data
revealed that 82 % of the respondents lived in houses made of wood. The majority of
respondents had no idea if their domicile was anchored to its foundation or whether
hurricane roof straps were installed. After the tornado the Joplin City Council encouraged
residents to use hurricane straps to strengthen any new housing construction (Joplin Globe
An analysis of 2011 other USA tornadoes and Joplin tornado fatalities by location
showed that the largest percentage of deaths occurred in business facilities (e.g. hospitals,
nursing homes, restaurants, churches, gas stations, and retail stores), respectively
(Table 4). Compared to the other USA tornado fatalities, 35 % more fatalities occurred in
Joplin in business facilities largely because the tornado directly hit many such facilities. No
other known tornado has destroyed or damaged so many business facilities in the past.
Nat Hazards (2012) 64:1511–1526
Table 4 Tornado fatalities by
location, 2011
Mobile home
112 (28)
Permanent home
164 (41)
Number (%)
65 (41)
19 (5)
15 (10)
26 (7)
66 (42)
Source: Based on data provided
from SPC (2012)
Other USA tornado fatalities
Number (%)
6 (2)
2 (1)
69 (17)
10 (6)
396 (100)
158 (100)
Some of the many severely affected business facilities in Joplin included St. John’s
Regional Medical Center, the 15th Street Wal-Mart Supercenter, the Home Depot store, the
Pizza Hut and AT&T store on South Range Line Road, and several other stores, churches
and nursing homes. Six people died in St. John’s Medical Center, two died in the Pizza
Hut, seven in Home Depot, three in the Wal-Mart Supercenter, three in Heights Baptists
Church, another three in Elks Lodge, and 10 in the Greenbriar nursing home (Zagier 2011).
Because the tornado occurred on Sunday evening and on Joplin High School graduation
day, many people from within and outside the city were present in such facilities, particularly in churches, big box and other stores, and restaurants. The graduation ceremony
was concluded by the time when the tornado struck Joplin.5 Stores, though, were still open.
Although many of these facilities had designated safer areas, these areas did not offer a
high level of protection against an FE-5 tornado.
Table 4 further shows that exactly equal proportion of tornado fatalities occurred in
permanent homes in the United States and Joplin in 2011. However, using the US tornado
fatalities data from 1985 through 2007 Simmons and Sutter (2011) reported that permanent
homes accounted for 31 % of all deaths during the study period.6 Ten per cent more
tornado deaths occurred in Joplin relative to the US in permanent homes probably because
of the absence of a basement and inadequate structural conditions in housing units of the
city. According to a NOAA assessment team report, many Joplin residents took shelter
after receiving the tornado warnings in the most appropriate location (e.g. interior rooms
and hallways) within their permanent homes. Unfortunately, a considerable number of
them did not survive (NOAA 2011).
Information presented in Table 4 suggests that 112 (28 %) of the 396 US tornado
victims died in mobile homes; the corresponding national percentage was 43 for
1985–2007 period (Simmons and Sutter 2011). This means that a much lower proportion of
deaths occurred in mobile homes in the United States in 2011 than the 1985–2007 period.
Surprisingly, no tornado deaths occurred in mobile homes in Joplin (Table 4). According
to the Joplin Community Profile, of a total 21,362 housing units, 350 (2 %) were mobile
homes in 2009 (Joplin, Missouri (MO) Profile 2011). This percentage is much lower than
Many respondents suspected that the number of deaths would have been higher if the Joplin High School
graduation had been held at the school auditorium rather than on the Southwest Missouri State campus,
which was outside of the tornado’s path. After such a ceremony, people generally remain in the area for
some time to socialize, and maintenance personnel also stay for cleaning purposes.
For 2000–2010 period, the permanent homes accounted for 32 % of all tornado fatalities in the United
States (SPC 2011).
Nat Hazards (2012) 64:1511–1526
the corresponding national percentage. Field surveys combined with a GIS analysis
revealed that almost all the mobile home units in Joplin were outside the tornado path.
5 Conclusion
This study has explored the probable reasons for the high number of deaths caused by the
22 May 2011, Joplin, Missouri, tornado. Five reasons seem to be associated with the high
number of tornado-related fatalities in Joplin: (1) the sheer magnitude of the event; (2) its
path through commercial and densely populated areas; (3) the size of the damage area; (4)
the physical characteristics of homes in Joplin; and, to some extent, (5) a desensitized
attitude towards tornado warnings on the part of some residents. Because of holiday
weekend and the Joplin High School graduation there were many out-of-town people in
Joplin on the day of the tornado. However, neither of these is pursued as possible reasons
for so many fatalities in the city. Even on weekdays, Joplin attracts visitors from neighbouring communities who seek medical treatment and other services.7 Moreover, many
people work in Joplin, but live in the neighbouring communities.
Contrary to initial suspicion, tornado warning failure and/or timeliness should not be
blamed for the large number of deaths caused by this event in Joplin. Still, 10 % of
surveyed respondents did not receive tornado warnings. Furthermore, 23 % of respondents
did not comply with warnings after receiving them. This indicates a need on the part of
emergency management officials to further improve the existing systems of disseminating
tornado warning information so that all at-risk populations will receive the intended
warning. Increased public education is also needed to improve residents’ response to
tornado warnings.
This study found that both residential and non-residential buildings in Joplin failed to
protect many people from a high magnitude tornado. Although an EF-5 tornado is unlikely
to hit Joplin in the near future, Joplin residents need shelter options. Given the physiography of the region, most notably the hard near-surface rock stratum underlying the Joplin
area, several alternatives to basements are proposed; one is the construction of safe rooms.
Building an in-house safe room costs approximately $8,000–$12,000, and public authorities should provide financial incentives for their construction.8 Crawl spaces can also be
used as tornado shelters if there is direct access to such spaces from within the house.
Given the geologic conditions that restrict the construction of basements, Joplin city
authorities should also consider construction of community tornado shelters.9 Owners of
non-residential buildings, particularly stores, should consider enhancing/creating safer
shelter facilities within such buildings. Implementations of these recommendations may
help the community to avoid such a high death toll in future tornado events.
Joplin has two large medical facilities, a growing medical spin-off industry, numerous restaurants and an
active trucking industry, as well as shopping facilities.
In the 1990s, FEMA provided residents of several cities in Kansas and Oklahoma grants in the range of
$3,000–$5,000 to incorporate a concrete safe room into their existing homes (Pattan 2003).
After a destructive tornado affected Seneca, Missouri (20 miles southeast of Joplin), in 2008, the city built
such a community shelter.
Nat Hazards (2012) 64:1511–1526
Acknowledgments The authors would like to acknowledge the grant from the Natural Hazards Center,
University of Colorado at Boulder, which supported this research. We also thank the members of the survey
team, James Wells, Brandon Haddock, Patrick Rissler (Kansas State University, Department of Geography)
and Tom Davis (Cloud County Community College, Department of Social Science), for their hard work in
collecting the data used in this research. We are grateful to Keith Stammer, Joplin Emergency Operations
Center Director, for granting us an interview, as well as arranging for the mapping division to provide us
with the very useful warning siren location data and the FEMA-derived damage data on the tornado. We also
thank the two anonymous reviewers for commenting on this work and suggesting quality improvements.
Finally, we thank the generous residents of Joplin, and those who took the time to complete our survey,
aiding us in completing this important research.
Ashley WS (2007) Spatial and temporal analysis of tornado fatalities in the United States: 1880–2005.
Weather Forecast 22:1214–1228
Balluz L, Schieve L, Holmes T, Kiezak S, Malilay J (2000) Predictors for people’s response to a tornado
warning: Arkansas, 1 March 1997. Disasters 24:71–77
Brenner SA, Noji EK (1995) Tornado injuries as related to housing in the Plainfield Tornado. Int J Epidemiol 24:144
Brooks HE, Doswell CA III (2002) Deaths in the 3 May 1999 Oklahoma City Tornado from a historical
perspective. Weather Forecast 17:354–361
Brown S, Archer P, Kruger E, Mallonee S (2002) Tornado-related deaths and injuries in Oklahoma due to
the 3 May 1999 tornados. Weather Forecast 17:343–353
Chaney PL, Weaver GS (2010) The vulnerability of mobile home residents in tornado disasters: the 2008
super Tuesday tornado in Macon County, Tennessee. Weather Clim Soc 2:190–199
Chaney PL, Weaver GS, Youngblood S, Pitts K (2012) Mobile home resident preparedness and response to
tornado warnings: the 27 April 2011 disasters in Dekalb County, Alabama. Quick Response Report
#223. Boulder: The Natural Hazards Center, University of Colorado at Boulder, CO
Current J, O’Kelly M (1992) Locating emergency warning sirens. Decis Sci 23:221–234
Cutter SL, Finch C (2008) Temporal and spatial changes to social vulnerability to natural hazards. Proc Natl
Acad Sci 105:2301–2306
Cutter SL, Boruff BJ, Shirley WL (2003) Social vulnerability to environmental hazards. Soc Sci Q
Daley WR, Brown S, Archer P, Kruger E, Jordan F, Batts D, Mallonee S (2005) Risk of tornado-related
death and injury in Oklahoma, May 3, 1999. Am J Epidemiol 161:1144–1150
Grazulis TP (1993) Significant tornados, 1680–1991. Environmental Films, St. Johnsbury
Joplin Globe (2011a) Authorities fear death toll could eclipse 100. May 23
Joplin Globe (2011b) Following May 22, how can homeowners protect themselves for the next one? June 26
Joplin, Missouri (MO) Profile (2011) Joplin, Missouri—main profile. www.city-data.com/city/Joplin-Missouri.html.
Accessed 17 Sep 2011
Klockow K (2011) Investigation of individual spatial awareness relating to response during the April 27,
2011 tornado outbreak. Quick Response Report #225. Boulder: The Natural Hazards Center, University of Colorado at Boulder, CO
Murphy K, Sherry M (2003) After devastation come the lessons: safe homebuilding again in FEMA
spotlight. The Kansas City Star, Kansas City, 15 May, A1 and A4
NOAA (National Oceanic and Atmospheric Administration) (2011) NWS Central Region service assessment: Joplin, Missouri, Tornado—May 22, 2011. U.S. Department of Commerce, Kansas City
Pattan A (2003) Grassroots homeland security. Nat Hazards Observer 27:1–3
Paul BK (2011) Environmental hazards and disasters: contexts, perspectives and management. WileyBlackwell, Hoboken
Paul BK, Dutt S (2010) Hazard warnings and responses to evacuation orders: the case of Bangladesh’s
Cyclone Sidr. Geogr Rev 100:336–355
Paul BK, Stimers MJ (2011) Tornado warnings and tornado fatalities: the case of May 22, 2011 tornado in
Joplin, Missouri. Quick Response Report #226. Boulder: The Natural Hazards Center, University of
Colorado at Boulder, CO
Paul BK, Brock VT, Csiki S, Emerson L (2003) Public response to tornado warnings: a comparative study of
the May 4 2003 tornadoes in Kansas, Missouri and Tennessee. Quick Response Report #165. Boulder:
The Natural Hazards Center, University of Colorado at Boulder, CO
Nat Hazards (2012) 64:1511–1526
Schmidlin TW (1993) Tornado fatalities in Ohio 1950–1989. In: Church C (ed) The Tornado: its structure,
dynamics, prediction, and hazards. American Geophysical Union, Washington, pp 529–533
Schmidlin TW, King PS (1997) Risk factors for death in the March 1, 1997 Arkansas tornados. Quick
Response Report # 98. Natural Hazards Research Applications and Information Center, Boulder, CO
Schmidlin TW, Hammer BO, Ono Y, King PS (2009) Tornado shelter-seeking behavior and tornado shelter
options among mobile home residents in the United States. Nat Hazards 48:191–201
Sherman-Morris K (2005) Tornadoes, television and trust—a closer look at the influence of the local
weathercaster during severe weather. Environ Hazards 6:201–210
Simmons KM, Sutter D (2006) Direct estimation of cost effectiveness of tornado shelters. Risk Anal
Simmons KM, Sutter D (2008) Tornado warnings, lead times and tornado casualties: an empirical investigation. Weather Forecast 23:246–258
Simmons KM, Sutter D (2011) Economic and societal impacts of tornadoes. American Meteorological
Society, Boston
Simmons KM, Sutter D (2012) Deadly season: analysis of the 2011 tornado outbreaks. American Meteorological Society, Boston
SPC (Storm Prediction Center) (2011) 2010 Annual U.S. killer tornado statistics. www.spc.noaa.gov/climo/
torn/fataltorn.html. Accessed 15 May 2012
SPC (Storm Prediction Center) (2012) 2011 Annual U.S. killer tornado statistics. www.spc.noaa.gov/climo/
torn/fataltorn.html. Accessed 15 May 2012
Stimers M (2011) A categorization scheme for understanding tornado events from the human perspective.
A Ph.D. Dissertation, Department of Geography, Kansas State University, Manhattan, KS 66506, USA
Sutter D, Simmons KM (2010) Tornado fatalities and mobile homes in the United States. Nat Hazards
Turner R, Hacker J (2011) 5:41: Stories from the Joplin Tornado. Kansas City Star Books, Kansas City
U.S. (United States Census Bureau) (2012) 2010 Interactive population search. http://2010.census.gov/2010
census/popmap/ipmtext.php?fl=29. Accessed 21 Mar 2012
Wisner B, Blaikie P, Cannon T, Davis I (2004) At risk: natural hazards, people’s vulnerability and disasters.
Routledge, New York
Zagier AS (2011) Joplin tornado: some died while saving others. Huff Post Green. http://www.huffingtonpost.
com/2011/06/04/joplin-tornado-death_n_871424.html. Accessed 7 June 2012