Designing a Risk Communication Campaign:
Applying the Insights from the Flood of '93 to other Flood
Events
The Diversity of the Field: Drawback or Drawbridge
Since the field's inception, risk communication scholars have profited
from diverse academic outlooks. The academy knows something about
how
people will behave in risky, sometimes disastrous, situations (Drabek
1986), how various types of people, from scientific elites to average
citizens, conceive of risk (Fischhoff 1982; Douglas and Wildavsky
1982,
Doble and Richardson 1992), and how certain institutions, primarily
political institutions, relief agencies and the mass media, can be
expected to perform in response to hazards, disasters and risk
(Quarantelli 1989). This quantity of information has been amassed using
a variety of methodologies, from case studies to public opinion
polling
to content analysis to experimental work (Boyce et al 1992).
As national investigations of the field have noted (National Research
Council 1989), the field owes some of its insights to its historic
roots
in the study of hazards and disasters developed primarily by
geographers
and sociologists. A different set of insights inform the field of
risk
assessment, traditionally the purview of engineers, systems
analysts,
and statisticians. Yet another set of findings focusing on risk
perception and policy making emerge from psychology, sociology and
political science (Krimsky and Plough 1988). Mass communication
scholars
have relied on the findings from all these disciplines to inform
their
research on risk communication.
While the enormous diversity of effort has failed to yield an
integrative and generalized theory of risk perception and behavior,
there have been consistent insights open to emperical validation and
testing. Using as a data base the insights gained from the mid-west
flood of 1993, this study develops a model for individual risk taking
during flood events, correlates that model with individual projective
responses to one highly likely situation during flood events,
integrates
these findings with survey research on information seeking behavior
and
trust in governmental institutions during flood events and then
develops an outline for a risk communication campaign about flash
flooding.This study is an attempt to integrate research findings across
disciplines with needs of policy makers. The goal of the proposed
communication campaign is to use insights from a rare event to inform
policy makers, civic and government leaders, and disasters agencies
how
to communicate effectively about a much more frequently occurring
flood
risk.
The 1993 flood study used many of the better documented findings from
previous research on risk perception and behavior. They include:
y That certain sorts of risks, for example natural hazards, are linked
to both geography and socio-economic status (White 1994; Drabek
1986).
General knowledge about what specific actions to take in hazardous
situations will vary with education, will decay relatively quickly over
time, and will not readily transfer from one hazardous condition or
event to another (Drabek 1986).
Research question: In the flood of 1993, were those living near rivers
and streams at greater risk than those living in other areas? How
well
will people understand the risks of their geographic locations? Will
people retain specific knowledge of what to do in the event of
flooding--particularly in the area of personal hygiene and the care of
contaminated food, etc.,--after the event? Will people be able to
transfer their knowledge about past experience with floods or other
hazards to the flood of 1993?
y Different perceive risk in an idiosyncratic fashion, what has been
characterized as the "expert vs. lay" view of risk (Fischhoff 1982).
Experts evaluating systems, events and conditions with which they are
familiar evaluate associated risks in probabilistic terms, in much
the
same fashion that systems analysts evaluate the mathematical
likelihood
of potential risky events. Average people, or experts evaluating
systems with which they are not familiar, tend to apply a risk
heuristic. That heuristic ignores such mathematical conditions as base
rates in favor of evaluations of perceived as opposed to
mathematical
likelihood, control over events, and potential harm (Kahneman and
Tversky 1982; Slovic 1987; Granberg and Brown 1995). Such individual
evaluations include elements of ethics (fairness) and emotion
(Gregory
1991; Boyce et al 1992) and are subject to internal individual
dynamics
(Ferguson and Valenti 1991 a and b). The lay risk heuristic also
may
include the persistent acceptance of discredited information
(Anderson,
Lepper and Ross 1980).
Research question: Will people placed in a hypothetical situation
involving personal risk will tend to evaluate the risk based on issues
of past personal knolwedge, individual control over events, and
perceived likelihood of possible injury or death? Under these
condition, will people assert they would engage in appropriate or
inappropriate risk behaviors? Will these behaviors have an emotional as
well as a rational component? Will these projected behaviors
correlate
with actions actually taken during the flood?
y That institutions will tend to behave in ways that while, not
uniform, are predictable. For example, emergency management officials
will tend to operate from a command post point of view (Quarantelli
1981; Perry 1989). All institutions, including the mass media, will
tend to cooperate, during the onset phase of the disaster ( Drabek
1986;
Wilkins 1987; Smith 1992). Political institutions, and some
economic
ones as well, will respond to a dynamic that includes public
opinion,
local and regional history, and, depending on the issue, a national
cultural and political context (Douglas and Wildavsky 1982; Krimsky and
Plough 1988; Priest 1995). The media will cover risks in
predictable
ways (Singer and Endreny 1992) that will vary according to stage of
the
disaster, craft considerations such as community role (Griffin and
Dunwoody 1993), and the specific medium involved (Nimmo and Combs 1985;
Smith 1992). Generally, media coverage of risk has been found to
focus
on events rather than context, to emphasize victims and loss rather
than
potential benefit, and fails to connect specific events to larger
social
and political issues (Wilkins 1987).
Research questions: What sort of information will people seek during
the flood and to which institutions will they turn to obtain
necessary
information? How will people evaluate the response of governmental
institutions to the flood?
While the field of risk communication has mushroomed with specific
findings about specific problems, it has failed to produce many
insights
that could be characterized as integrative or generalizable. This
relatively low level of integration is particularly noteworthy during
the decade of the 1990s, when the United Nations has established a
world-wide goal of mitigating natural hazard losses and when, in the
wake of two large California earthquakes, Hurricane Andrew, and the
1993
midwest flood, Congress has established the goal of developing an
integrated a national policy of hazard mitigation. This study begins
that work of integration through emperical testing of some well
accepted
findings that have seldom been placed in a research context where
they c
ould inform and enrich each other, synthesizes them, and then
suggests a
practical application for the work.
Background: The great flood of '93 (1)
The 1993 midwest floodDdepending on geographical locationDmay be viewed
as anywhere from a 50 to a 500-year event. In Missouri, where both
the
Mississippi and Missouri rivers reached record levels, statistics
collected by the U.S. Geological Services indicate the rising water
should be evaluated as a 500-year event. In other states, for example
many places in Illinois and Wisconsin, the flood should be
considered a
50-year event.
The flooding in the upper Mississippi and Lower Missouri basins from
mid June through early September 1993 was caused by intense pulses
of
rain that fell in late June, July and early September and that
followed
six months of heavy and persistent rainfall. Precipitation between
January and July in the affected area was 1.5 to 2 times normal. In
June, a stalled weather pattern caused by a strong low pressure
system
in the western U.S. and a large high pressure system in the
southeast
resulted in large amounts of rain in the upper midwest. By late
June,
flood storage reservoirs were at or near capacity and soils
throughout
the area were saturated. Even though peak discharges were not
exceeded
everywhere, human-inducted changes in the landscape made it possible
for
a less-than-peak volume of water to produce a higher flood stage, a
phenomenon that was specifically noted in St. Louis (Water Resources
Update 1994).
Flooding caused significant damage in nine states: Illinois, Iowa,
Kansas, Minnesota, Missouri, North and South Dakota, Nebraska, and
Wisconsin. All the counties in Iowa were declared disaster areas. More
than 1,000 levees stretching nearly 6,000 miles in length were
breached
or overtopped. Many others sustained significant damage. A total
of
487 counties were included in the Presidential Disaster Declaration
(Commission 1994).
National statistics attribute about 50 deaths region wide to the flood,
including one cluster of six deaths in metropolitan St. Louis where
a
group of youngsters on an outing with a clergyman were trapped in a
cave
by rising water and ultimately drowned. However, this death total
itself is in dispute. The state of Missouri, for example, suggests that
50 Missourians died as a result of the flood, 30 of them in flash
floods
that were ancillary to the flooding on the Mississippi and Missouri
and
hence are not included in the national death toll. Similar debates
over
death and injury tolls exist in other states. It was these
ancillary
deaths, the result of events much more common than a 500-year flood,
that prompted this more detailed investigation of the risk factors and
behaviors associated with the larger event.
Estimates about dollar damages from the flood also vary widely. The
most commonly cited at the time of the flood were from $12 to $16
billion, more than half of those losses sustained by agriculture. The
National Weather Service estimated damaged at $15.7 billion and in
August 1993 the New York Times published a damage estimate of $12
billion based on information provided by state and federal sources.
However, these losses were not evenly distributed. Only four states,
among them Missouri, ultimately recorded a decrease in the corn crop
for
the 1993 year. Later estimates pegged agricultural losses at
between $4
and $5 billion, most of it in upper Mississippi basin (Committee
1994).
It is also important to note that the nation's investment in flood
protection projects paid off. It is estimated that reservoirs and
levees built by the U.S. Army Corps of Engineers prevented more than $19
billion in potential damages. Large areas of Kansas City and St.
Louis
were protected by both private and public levee systems, although
several suburbs suffered extensive damage. Again, the fact that some
government efforts at flood prevention and mitigation were successful
during the 1993 event provides some indication that additional steps
by
government to conserve both life and property may be warranted.
The total number of people affected will never be certain. However,
the American Red Cross estimated that 56,295 family dwellings were
affected in some way. The Red Cross spent more than $30 million in
flood relief efforts, sheltered 14,502 people in 145 shelters in the
region and served more than 2.5 million meals. Surveys made by Red
Cross workers immediately after the flood identified more than 55,000
flooded residences, an estimate FEMA later updated to 70,545. As
of
April 1994 the federal government had received 167,224 registrations
for
individual assistance and 112,042 applications for the Disaster
Housing
Program, 89,734 of which had been approved. The Disaster Housing
program data indicate that more than 100,000 residences were flooded.
Property damage has been estimated as high as $12 billion; Congress
allocated in excess of $6 billion for disaster relief. FEMA will
cover
about $650 million for public assistance. No overall damage
estimates
for businesses are available, but Small Business Association loans
after
the flood exceeded $334 million for physical damage and economic
injury
(Committee 1994, p. 18). Much of this damage occurred in Kansas
City
and St. Louis where both private and publicly-built levees were
overtopped. In aggregate, at least 30,000 jobs were disrupted.
Flooding that wiped out access roads, and sometimes major arteries,
added hours to the time it took to get to work and resulted in much
lost
business. The Federal Aviation Administration also identified 33
airports with varying degrees of flood damage. While many of them were
small airfields, repair costs were estimated at $5.4 million. Many
of
the region's water treatment plants, which are often located at the
lowest possible location, also sustained damage. The Environmental
Protection Agency has identified 200 water treatment plants impacted by
the flood, the most well known of which is the Des Moines Water
Works
that was flooded, remained out of operation for 12 days, and could
not
produce potable water for 19 days. Damages to public buildings
exceeded
$27 million. Water control facilities sustained more than $20
million
in damages, and facilities such as parks and other recreation
facilities
recorded more than $22 million in damages (Committee 1994, p. 19).
Superfund sites were flooded and huge volumes of farm chemicals and raw
sewage were flushed into the rivers (Water Resources Update, p. 21).
More than 100 towns have taken steps to partially or completely
relocate, with a potential cost of $500 million (Water Resources
Update, p. 21.)
The Federal Crop Insurance Corporation estimated that about 57 percent
of the flooded acreage was insured; those claims are expected to top
$600 million. (Environment 35 (10): December 1993, p. 7-35).
Officials
with the Department of Agriculture say perhaps 400,000 acres will
remain
covered with sand much of which was the materiel eroded from
breached
levees. In Missouri alone, 455,000 acres or 60 percent of the
cropland
in the Missouri River floodplain, were damaged by sand deposits and
scouring. Of that, 77,500 acres are covered with six to 24 inches of
sand, and 59,000 acres were covered by more than two feet of sand.
Although these figures are staggering it is essential to note that
these impacts were not uniformly spread throughout the population. A
generalized statement about risk previously noted in the literature
(Vaughn 1994) applied to the 1993 flood: the riskiest thing to be was
poor. Flooded neighborhoods tended to be lower income, have a high
percentage of rental properties, and generally house more elderly
residents, more young families and more people on assistance. Homes in
the floodplain often had market values of less than $25,000
(Committee
1994, p. 7). Thus, the Great Flood of 1993, which is widely
regarded as
the third most costly domestic disaster in U.S. history, hit some
portions of the population much harder than others. Those who were
among the least able to cope economically often bore the largest social
costs.
Individual responses to the event
Social scientists have often studied human response to disasters
through what is called quick response researchDarriving in the field as
soon as possible after disaster onset and collecting data from those
at
specific locales. This type of research has yielded a somewhat
consistent set of findings. For example, hazards scholars know that
disasters result in some psychological problems for a subset of the
population, and the economic dislocation caused by disasters will also
indirectly contribute to such human problems as increased drug
usage,
domestic abuse, etc. Individual daily routines are often disrupted,
and
in the early days of disaster onset and recovery, people spend a
great
deal of time trying to reconstruct their lives. These activities
includ
e fulfilling immediate needs for medical care, food, and shelter as
well
as the more long term needs for rebuilding and reconstruction
(Drabek
1986).
However, social scientists seldom have the luxury of asking people how
they have been affected by a disaster in the long-term. Few studies
have asked people to evaluate their experiences after the recovery
period is well underway. Both time and distance from the event may allow
individuals to be more reflective about their experiences.
Missouri residents have had such a chance to be reflective. In
statewide survey funded by the Missouri Department of Health and
conducted by the University of Missouri's Center for Advanced Social
Research during March and April of 1994--about seven months after the
"official" end of the eventDabout 1,950 Missouri residents were
surveyed
regarding a variety of issues raised by the flood. Some of the
questions dealt with specific actions residents took during the height
of the disaster such as boiling water or seeking certain sorts of
information in making daily decisions. Other questions asked residents
to evaluate the response of state, local and federal officials to
the
flood. And, still other questions attempted to determine some of the
longer-term impacts the flood may have had on Missouri residents, for
example by asking them what they "worried" about six months after the
event.
The survey was conducted by telephone and the methodology used allowed
the interviewers to speak with those with unlisted and new phone
numbers. Residents of every county in the state were surveyed, and
rural residents responding to the survey equaled the percentage of
rural
residents in the state. About 53 percent of the respondents were
male,
slightly higher that the state's population (2), and about nine
percent
of the survey respondents were people of color, less than the 13
percent
of the state's population that is listed as African-American. About
65
percent of those responding said they had been born in the state;
more
than 89 percent had lived in Missouri for at least 10 years. The
results are accurate plus or minus three percent.
Because the 1993 flood was an event of such magnitude, previous
research would have predicted Missourian's long-term response to the
flood. Missourians indicated the flood had had significant and long
term
impacts. For example, when asked "what is it that you worry about"
one-third of those responding said "nothing" and an additional 18
percent listed crime. These responses are predictable, particularly
because of the high crime rates in the metropolitan Kansas City and
St.
Louis areas. However, 17 percent of those responding said they
worried
about environmental issues, among them flooding, rain, etc. More
than
double the number of people said they worried about the environment
than
said they worried about health or the economy. When probed for
additio
nal worries, about 15 percent of those responding listed the
environment, ranking it fourth behind "miscellaneous" concerns, the
economy and health. In contrast, most national polling conducted by
organizations such as Gallup and Roper report that spontaneous mention
of concern over environmental issues normally ranges from 1 to 3
percent.
When respondents were asked to change their focus from general concerns
to concerns specifically about "the weather or climate in Missouri,"
more than 18 percent listed flooding as their top concern. Worry
about
tornadoes/storms and cold and ice was ranked as the top concern by
19
percent of those responding. When probed for second-level worries
connected to weather and climate, flooding became the top
categoryDlisted by more than 19 percent of those responding. This
worry
outranked concern about tornadoes by more than four percent.
(Missouri
is one of the midwestern states considered part of the northern
hemisphere's tornado alley.) Thus, more than six months after the
flood, the disaster remained salient. These results parallel the
findings by Turner (1986) about Californian's concern over earthquakes.
Indeed, if the Turner study is taken as an base, then the flood of
1993
left Missourians more concerned about flooding than Californians
appear
to be concerned about earthquakes. This concern is evident despite
the
statistical fact that earthquakes are a more likely event in
California
than a 500-year flood is in the midwest.
The fact that the flood remained salient to Missourians so many months
after the event was probably linked to the life experiences many
Missourians had during the summer months. Social science research in a
variety of field indicates that personal experience with an event or
issue tends to heighten salience. Missourians reported many such
personal experiences. For example, more than 88 percent of those
responding said they had talked to people about the flood, and 64
percent said they had visited a flooded area. Almost half--48
percentDsaid they had volunteered in some way to help with flood relief,
13.5 percent of them as sandbaggers and 37.4 percent who helped in
other
capacities. And, in a finding that may be linked to the number of
deaths
in the state associated with flash floods, more than 45 percent of
those
responding said they had had to drive on flooded streets at least
once
during the disaster. Other scholars have noted a similar trend
throughout the midwest.
Thousands filled and stacked sandbags to hold weakening levees;
others worked day after day to help clean the homes and business
of
people they had never met. Dry communities adopted those in
need....Those who were recipients of this assistance will never
forget this demonstration of true caring (Committee 1994, p.
xxi).
The impact of personal experience with the flood also was correlated
with knowledge and retention of facts about the flood itself. Flood
relief work, particularly work as a sandbagger, was the strongest
predictor of knowledge about how to handle risks associated with flood
waters, including the appropriate ways to deal with contaminated
food
stuffs and with personal hygiene.
Probably because the survey was taken six months after the flood, the
vast majority of those responding characterized their mental health
as
good. Seventy-three percent of those responding said the flood had
no
impact on their psychological or physical health, and 92 percent
reported the flood had not curtailed their daily activities. Fewer than
100 of those surveyed indicated that the flood had made it
impossible
for them to reach their homes; 29 of the survey respondents said they
had to move because of the flood and 27 reported that their homes had
flooded.
While people said they were relatively psychologically unscathed by the
event, responses to another set of questions seemed to indicate that
the
flood reminded many of the potential dangers in an environment they
normally considered benign. For example, almost 60 percent of those
surveyed said that they lived near creeks, streams and rivers. Of
those
who did, about two-thirds said they did not regard those creeks,
streams
and rivers as dangerous before the 1993 flood. After the event,
opinion
shifted dramatically. Now, 63 percent asserted that they regarded
those same geographic features as dangerous. Of those surveyed, 2.7
percent reported that their children had played near flood waters.
Thus, geography and climate appear to have played their predicted roles
in the risks associated with the flood. Furthermore, at a level of
generalized concern as well as an individual evaluation of specific
geographic features, Missourians found the flood of continuing salience
in their daily lives. Furthermore, work in flood relief efforts, as
well as just daily activity during the flood, appears to be correlated
with some risk taking behaviors--for example driving on flooded
streets
or visiting areas that were flooded--and some retention of knowledge
about specific safety and health procedures. The flood event appears
to
have heightened generalized concern about the weather, climate and
geography in the state. This heightened awareness coupled with personal
experience with the flood means Missourians have been primed to
think
about the weather, climate and their own local geography is somewhat
different ways than past state and individual history would indicate.
But, when the questions moved from generalized concern to the
acquisition and retention of specific facts, Missourians again reflected
previously documented patterns: many of their beliefs were
inaccurate.
For example, about 61 percent of those surveyed agreed that Missouri
received about the same amount of flooding as other states and
sustained
about the same number of deaths. In fact, Missouri was among the
hardest hit by the 1993 flood, both in terms of property damage and
deaths. Of those who responded, almost 88 percent said they did not
live in a flood plain; about 16 per cent said they carried flood
insurance. However, 40 percent of Missouri's total land area is within a
100-year flood plain.
Similarly, Missouri residents were vague about specific facts linked to
reducing their risk of illness and injury during a flood. Whether
the
issue was how to decontaminate well water, whether to keep food that
had
been exposed to flood waters, how long it was necessary to boil
water if
informed a water source had been contaminated, or what to do to
decontaminate themselves if they had been exposed to flood water, the
majority of respondents to the survey could be viewed as a potential
public health problem. Missouri residents either didn't know the
correct answers or were positive that doing the wrong thing was
appropriate.
Not surprisingly, the better educated a Missourian was, the more likely
he or she was to provide the correct answers to such questions.
Women
also were somewhat more likely to know the correct answers to
questions
about how to appropriately handle contaminated or potentially
contaminated food than were men.
Although it would be inaccurate to suggest that Missourian's responses
to such questions would duplicate those of all other midwestern
residents who were flooded in the summer of '93, there are some trends
in the responses that are worth noting. Missourians say they are
concerned about the hazards connected with flooding, but they have
retained relatively little specific information about how to cope with
flood events. This finding again replicates other studies. In
addition, Missourians may have a difficult time linking specific
behaviors during the "great flood" to behaviors during times that seem
much more routine, for example, the strong thunderstorms associated
with
tornados that often cause flash flooding in the state. In this
sense,
Missourians fail to see the link between "tornados" on the one hand
and
"flooding" on the other--a linkage that the disaster mitigation
community understands only too well. Combating the apparent inability
to generalize from one type of hazard to another, as well as
reinforcing
information gain that tends to decay rapidly, should thus become
central
parts of any mitigation strategy aimed at something less rare than a
500-year event.
In addition, Missourians also apparently found the national, state and
local governments a trustworthy source of information and aid for
dealing with the flood. This "vote of confidence", however, was far
from unanimous. About 73 percent of those responding said they
believed
the federal government had been helpful during the flood; 90 percent
said their state and local government had been helpful. More than 53
percent disagreed with the statement "there would have been less
death
and injury due to the flood if the federal government has done a
better
job". About 45 percent of those responding disagreed with the same
statement when the referent was changed to "state and local"
governments.
This finding about the trustworthiness of government is noteworthy
because these opinions about the role of government in times of crisis
are far different than the results of national opinion polls about
the
role and impact of government in Americans' daily lives (Times
Mirror
1994). The results of the 1994 election have been widely analyzed
and
accepted as a plebiscite on the question of whether government
should do
more for citizens and whether it has been effective. Missourians'
opinions about the role of government in the flood of 1993 run counter
to national trends. At a time when generalized trust in government
is
not high, working through the problems associated with the flood,
and
perhaps disasters in general, may be one place to help rebuild the
governmentDcitizen relationship. Community grew out of the flood. As
mitigation strategies are developed, government-sponsored efforts to
link these newly-formed communities with preparedness and mitigation
efforts may prove effective in the same way that interpersonal
communication was found effective in the prevention of heart disease
(Maccoby and Farquhar 1975). It is clear that citizens value
government
efforts on their behalf in the area, a positive starting point for
any
cohesive mitigation effort.
Risk Taking During the Flood: A Model
Missouri residents who responded to the survey also were asked a
detailed set of questions about their behavior during the flood
itselfDan event which, it must be emphasized, lasted for weeks in most
parts of the state. For example, in addition to being asked whether
they had worked as sandbaggers, people also were asked whether they
had
been in water up to their ankles, waist or over their heads during
the
flood. People also were asked whether they had ever driven on
flooded
roads. The point of this model, therefore, is to discern the risk
factors which may place individuals in danger during serious floods.
Using these individual responses as a base, a model was developed
that
associated individual activities with a variety of other responses
(Amemiya 1985; Fox 1985).
The model employed 16 variables, is based on 1950 responses, and
explains about 18.2 percent of the variance in risk behavior.
Other responsesDthe independent variablesDwere divided into four
categories. They were:
1. Demographic factors: Sex (SEX), race (RACE) income (INCOME) and
education (EDU);
2. Measures of proximal association with creeks, streams and rivers:
live in flood plain (LIFP), live near a river (RIVER), the area
where
you live flooded previously (PRIORFL);
3. Individual activities during the flood: where information about
flood was obtained (INFO), who is to blame for flood woes (BLAME),
how
often did you talk about the flood (TALK), did you visit flooded
areas
(VISIT), did you fill sandbags (BAG), did you volunteer to help in
some
other way (VOLUN);
4. Level of knowledge about specific actions to take during the flood:
how to decontaminate food or water supplies (DECON); sources for
information about decontamination procedures (SOURCES), and what
specific acts you took (for example, boil water) to decontaminate
(ACTS).
The model distinguished between activities that people engaged in that
were associated with risks and other activity, thus preventing
contamination of the dependent variable with some of the independent
variables. ----------------------------------INSERT TABLE 1
HERE-------------------
The model suggests that a very strong relationship exists for
respondents who live in the flood plain and risk-taking behavior (p
.001). Living near a river is associated with a .43 increase in
risk-taking ( p .06) If the respondents reported that they experienced
a prior flood in the area, they are predicted to take an additional
.53
risks (p .02). Taken together, respondents who lived in the flood
plain, who were near a river that flooded in the past were more likely
(p 05) to take risks then other respondents.
However, place of residence was not the only indicator of increased
risk taking behavior on the part of respondents. Sandbagging was
strongly related to increased risk taking behavior; other types of
volunteers also apparently took on additional risks. Simply visiting
flooded areasDa behavior that 64 percent of the respondents didDalso
is
associated with increased risk taking behavior.
Men appeared to take more risks than women, (p .001); caucasians
appeared to take more risks than non-caucasians ( p .02) and those with
lower levels of education also appeared to take more risks ( p
.04).
Each of these variables, in addition to income, also may be related
to
other variables in the model as well as to housing patterns, which
also
were associated with property damage and other forms of flood risk
in
1993.
It is important to note that other measures were not associated with
increased risk taking. Information was not associated with increased
risk taking behavior and income also appeared not to be associated
with
increased risk taking. Knowledge about how to treat polluted water
for
decontamination, how to maintain personal hygiene, and where to
obtain
potable water after the flood also did not predict risk taking
behavior.
In summary, the model examined four components simultaneously. Among
the individual factors examined, only income was not a contributor
to
risk taking.
Geographic proximity to rivers was a strong predictor of risk activity,
particularly for those living in the flood plain or those who lived
near
waterways that had flooded previously. Sandbaggers, visitors to
flooded
areas, volunteers and those who talked about the flood were much
more
likely to take risks. This effect is mediated by the assignment of
blame for the predicament of flood victims. If one believes that a
victim is responsible for his or her own fate, the likelihoodof taking
risks declines.
Thus, risk taking during the flood was relatively straightforward. If
people lived near water or, if, in an effort to help others they put
themselves near water, then risk taking behaviors escalated.
Putting yourself at risk: Driving in flood waters
One of the most common experiences Missourians had during the flood of
1993 was driving over water-covered streets and roads. Of those
polled,
more than 45 percent said they recalled driving on flooded streets
at
least once.
To get a better sense of what that may have meantDand how much residents
knew about the risks of that particular behaviorDrespondents were
presented with a scenario. The question was as follows: "A young man
leaves his house at night to drive home. He gets about half way to
his
destination when he sees water flowing over the road. How do you
think
the young man was feeling in this situation?"
Of those responding, 24.8 percent said they would feel scared or
frightened, and an additional 20.9 percent said they would be
"concerned". Miscellaneous responses, including those who did not
comment or did not provide an emotional response, accounted for 16.3
percent of the responses. About 9.9 percent of the respondents said
they would be curious or ask questions, and an additional 6.1 percent
said they would feel helpless/stranded, while 5.2 percent said they
would be shocked/or stunned. Other responses, each accounting for less
than 5 percent of the total, included: angry; confused/puzzled;
sad/depressed; don't know; stupid/foolish, and calm/unconcerned.
The scenario continued with respondents being told that "...actually,
the young man drives his vehicle into the water where it stalls.
How do
you think he is feeling now?" Responses generally followed the
same
pattern, although many more people--41 percentDreported that they
would
feel scared/frightened. In addition, 14.9 percent said they would
feel
foolish/stupid, 12.6 percent provided miscellaneous responses or did
not
elaborate on their response to the first part of the scenario, 11
percent said they would be shocked/stunned, 6.9 percent reported they
would be concerned and 6.4 percent said they would be angry/upset.
Fewer than 3 percent of the respondents said they would feel
helpless/stranded, confused or puzzled, curious, sad/depressed, and
calm/unconcerned.
The scenario concluded by asking respondents what they would do in the
young man's situation. Of those responding, only 14.7 percent
volunteered the "right answer"Dthat is, they never would have driven
into the water in the first place. At the other end of the continuum,
the largest single subgroup of those responding--36.9 percent-- said
they would engage in the riskiest of behaviorsDleaving or getting out
of
the truck. An additional 11 percent said they would attempt to get
on
top of the truck while another 21 percent said they would remain in
the
truck. Another 11 percent provided miscellaneous responses. About
2.6
percent of those responding said they did not know what they would
do
and about 1.7 percent said that, in this particular situation, they
would "pray".
Although projective techniques have their clear limitations, responses
to this set of questions coupled with the risk taking model confirm
the
risk perception literature, particularly in situations where people
believe they have some control and where risks are not considered
immediately life threatening. One of the best documented contradictions
in risk perception is that which surround automobiles. Although
statistically, the chance of being involved in an injury-producing
accident during the average lifetime is one in three, most people rate
the risk of driving as much lower than, for example, being struck by
lightening or traveling in an airplane. Slovic (1987) and others have
linked this inability to evaluate risk with the individual belief
that a
good driver (always the person who is responding to the question)
has
control over a situation and hence won't "get into trouble". In
addition, most people do not conceive of auto accidents as routinely
life-threatening, at least when compared with risks such as living
near
a nuclear power plant. This lack of "dread" about auto accidents
also
tends to support individual evaluations of them as less frequent and
less serious than the statistics indicate.
A similar sort of logic seems to permeate the scenario responses.
People seemed to evaluate the risks of driving on flooded roads as
rather less than more "dreadful" and they also appeared to believe that
they had a significant amount of individual control in the
situation.
Furthermore, almost half of those polled said they had driven on
flooded
streets, thus lending support to their "hypothetical responses".
When
confronted with a situation "gone sour", people responded
emotionally
(and many would argue with appropriate emotions) but lacked
information
about the best strategy for mitigating the risk. Furthermore,
while
the scenario was presented in terms of a 500-year event, flash floods
are common in Missouri, and each year Missourians die as a result of
them.
Information seeking behavior
Considering the duration and long-term salience of the event, the
literature would predict that Missourians would be interested in finding
out about the flood for reasons that ranged from issues of personal
safety to curiosity. More than 70 percent of those responding said they
had followed "what was going on with the flood" most of the time.
Respondents indicated they had done much of this "following" through
the mass media.
The broadcast media, specifically television and radio, were named as
the "primary source of information about the 93 flood" by more than
89
percent of those surveyed. Furthermore, what Missourians expected
to
find in media accounts was what disaster researchers would describe
as
information about the quick onset and recovery phases of the event.
People relied on the broadcast media for information about local flood
crests (88 percent), weather forecasts (88 percent), bridge, highway
and
street closings (93 percent), and information about "what to do if
you
were exposed to flood water" (87 percent).
Respondents were also asked to rank the importance of certain kinds of
flood news to them. People rated as very important news stories
about
"how to remain healthy and safe" (60 percent), how to protect
yourself
(59 percent), how and where to get help (53 percent), pictures of
flood
damage (47 percent), rainfall predictions (45 percent).
A second category of news account, information that would be more
likely to deal with long-term recovery and mitigation issues, was less
sought after but still considered important by a significant
minority of
those surveyed. People said they wanted news about the economic
impact
of the flood (32 percent), information about what government was
doing
to help (28 percent), about inter-agency conflicts (27 percent) and
information about who to blame for the disaster (12 percent .
This less sustained interest in the deeper causes of the flood was
reflected in another set of responses. When asked about flood
prevention measures, 60 percent of those responding agreed with the
state that, "once it starts raining, there's little you can do to
prevent a flood, " while 65 percent of the respondents said they believe
"there is no way to avoid the possibility of another great flood".
By integrating and synthesizing the component parts of this study with
the previously existing literature, the following conclusions stand
out:
y Midwestern residents, that is those who are used to dealing with
large events such as a 500-year flood only infrequently, appear to
lack
usable information about how to cope with a variety of flood related
hazards that occur much more frequently in the mid-west, specifically
flash flooding.
y Geography coupled with housing and other patterns appears to place
certain portions of the population more at risk. In state such as
Missouri, however, local geography--particularly the presence of small
creeks and streams--make such risk more widely shared but still
unacknowledged.
y Many Missouri residents say they would engage in risk behavior
surrounding floods. Their self-reports of actions taken during the
flood of 1993 appear to confirm this view.
y Information about the flood is salient to Missourians. In addition,
they specifically desire information directed at health and safety.
y Missourians can be expected to turn to the broadcast media for
coping and mitigation information.
y Government is viewed as at least a moderately trustworthy and
helpful source of information relating to the flood.
In conclusion, if social learning theory is accepted, then a risk
communication campaign that models appropriate behaviors in predictable
situations may have some modest chance for success. Such a campaign
should be highly mediated, particularly emphasizing the broadcast
media.
Such an effort would represent a clear departure for the state of
Missouri, although other states, for example Colorado and Arizona
(1992), have begun to implement such campaigns. Based on this study and
previous literature, the following campaign outline is suggested:
The proposed campaign: Modeling how to deal with flash floods
Based on the foregoing analysis, the following elements of a risk
communication campaign seem relatively clear.
Missouri should develop a risk communication campaign around issues of
flash floodingDa common occurrenceDsimilar to campaigns that now
begin
around the advent of tornado season:
The campaign should rely on the broadcast media, specifically
television, which is where people now expect to get hazard information;
To be effective, the campaign needs to explore situations that people
are likely to encounter in time of flash flooding. As indicated by
the
survey responses, these situations should include living near creeks
or
streams that may flood or have flooded in the past and traveling to
visit flooded areas or to provide help and assistance to those who are
affected by the flood.
A potentially effective strategy is one of modeling appropriate
behavior rather than merely telling audience members what they should in
a more factual and less dramatic approach. Using dramatic scenarios
may
allow individuals more ready access to risk heuristics that employ
both
rational and emotional forms of dealing with risk. Linking flash
floods
with the flood of 1993 may help viewers build appropriate cognitive
structures to help insure that the information can be retained and
accessed.
Any such campaign would be more effective if coupled with other forms
of communication. Further, certain communities of interestDfor
example
sandbaggers or churches that provided volunteer help during the
floodDwould be appropriate foci of interpersonal information campaigns
on the same issue.
Any risk communication campaign should include a strong element of
evaluation research to track initial effectiveness and potential
longevity.
While providing state and federal policy makers with a way of
mitigating the hazards of flash flooding will not solve the nation's
hazard mitigation problem, it does provide an important step in
integrating knowledge about risk across disciplines and across hazards.
A 500-year flood is a rare occurrence; flash floods are common.
Using
what scholars know about the rare event to inform actions about more
frequent events provides one element of a national mitigation effort.
Footnotes
1. For a much more complete account of the origins and impact of the
1993 mid-west flood, please consult: Stanley A. Changnon, "The Great
Flood of 1993". Boulder, Co.: Westview Press, in press.
2. The gender distribution of survey respondents was planned. Most
contemporary telephone surveys result in about 60 percent female
respondents. To avoid this sampling problem, the survey was designed to
oversample for male respondents.
Bibliography
Amemiya, T. 1985. Advanced econometrics. Cambridge: Harvard
University Press.
Anderson, D., Lepper, M. and Ross, L. 1980. Perseverance of social
theories: the role of explanation in the persistence of discredited
information. Journal of personality and social psychology 39:
1037-1049.
Arizona floodplain management association. 1992. "Flash floods: A
warning to beware," video tape.
Boyce, R. R., Brown, T. D. McClelland, G H. Peterson, G. L. and Schulze,
W. D. 1992. "An experimental examination of intrinsic values as a
source of the WTA-WTP disparity," American Economic Review 82:
1376-1372.
Changnon, S. A. 1995. The Great Flood of 1993. Boulder, Co.:
Westview Press, in press.
Doble, J., and Richardson, A. 1992. "You don't have to be a rocket
scientist," Technology Review, January, pp. 51-54.
Douglas, M., and Wildavsky, A. 1982. Risk and culture. Berkeley, CA.:
University of California Press.
Dunwoody, S. and Griffin, R. 1993. Journalistic strategies for
reporting long-term environmental issues: a case study of three
Superfund sites. In The mass media and environmental issues. (ed) A.
Hansen. Leicester, U.K.: Leicester University Press.
Drabek, T. 1986. Human system responses to disasters: An inventory of
sociological findings. New York: Springer Verlag.
Ferguson, M.A., and Valenti, J.M. 1991. Health Education Quarterly 18
(3): 303-318.
Ferguson, M.S., and Valenti, J. M., and Melwani, G. 1991.
"Communicating with risk takers: A public relations perspective," in
Grunig, J.E. and Grunig, L. A. (eds.): Public Relations Research
Annual
3 . Hillsdale, N.J.: Lawrence Erlbaum Associates, pp. 195-224.
Fischhoff, B., P. Slovic, and S. Licktenstein. 1982. "Lay fables and
expert foibles in judgments about risk," The American Statistician
36
: 240-255.
Fischhoff, B., Bostrom, A., Quadrel, M J. 1993. "Risk perception and
communication," Annual Review of Public Health 14, 183-203.
Fischhoff, B. 1994. "Risk perception and communication unplugged:
Twenty years of process." A paper commissioned by the Center for
Environmental Communication, Rutgers University.
Fox, J. 1985. Linear statistical models and related methods. New
York: John Wilye and Sons.
Granberg, D. and Brown, T. A. in press (1995). The Monty Hall dilemma.
Bulletin of personality and social psychology.
Gregory, R. 1991. "Risk perceptions as substance and symbol," In Risky
business: Communicating issues of science, risk and public policy.
Wilkins, L., and Patterson, P. eds. Westport, CT.: Greenwood Press.
Hornig, S. 1992. "Framing risk: Audience and reader factors,"
Journalism Quarterly 69: 679-689.
Interagency floodplain management review committee. 1994. Sharing the
challenge: Floodplain management into the 21st century.
Washington, D.
C.: U.S. Government Printing Office.
Iyengar, S. 1994 . Is anyone responsible: How television frames
political issues.
Chicago: University of Chicago Press.
Krimsky, S., and Plough, A. 1988. Environmental hazards:
Communicating risks as a social process. Dover, Mass.: Auburn House.
Kahneman, D., and Tversky, A. 1982. The simulation heuristic. In ,
Judgment under uncertainty: Heuristics and biases. D. Kahneman, P.
Slovik and A. Tversky (eds) New York: Cambridge University Press, pp.
201-208.
Lowery, S., and DeFleur, M. 1988. Milestones in mass communication
research. 2nd edition. New York: Longman.
Maccoby, N., and Farquhar, J. W. 1975. Communication for health:
unselling heart disease. Journal of Communication 25 (3): 114-126.
National Research Council, Committee on Risk Perception and
Communication. 1989. Improving Risk Communication. Washington D. C.:
National Academy Press.
Perrow, C. 1984. Normal accidents: Living with high risk technologies
. New York: Basic Books.
Perry, R., and Lindell, M. 1989. Communication threat information for
volcano hazards. In Bad Tidings: Communication and catastrophe,
eds.
Walter, L. Wilkins, L. and Walters, T. New York: Lawrence Erlbaum.
Priest, S. H. 1995. Information equity, public understanding of
science and the biotechnology debate. Journal of Communication 45:
39-54.
Quarantelli, E. L. 1989. The social science study of disasters and
mass communication. In Bad Tidings: Communication and catastrophe,
eds. Walter, L. Wilkins, L. and Walters, T. New York: Lawrence
Erlbaum.
Singer, E., and Endreny, P. M. 1992. Reporting on risk. New York:
Russell Sage Foundation.
Slovic, P. 1987. "Perception of Risk," Science, 36, 280-285.
Smith, C. 1992. Media and apocalypse: News coverage of the
Yellowstone Forest Fires, Exxon Valdez oil spill, and Loma Prieta
earthquake. Westport, CT.: Greenwood Press.
Stocking, H. S. 1992. "Packaging risk: Lessons for students in PR and
journalism." Journalism Educator 47 (2): 26-31.
Times-Mirror. 1994 (October). "The people, the press and politics:
The new political landscape."
Turner, R. H. 1986. Waiting for disaster. Berkeley: University of
California Press.
Vaughn, E. 1994. "The Significance of socioeconomic and ethnic
diversity for the risk communication process," a paper commissioned by
Center for Environmental Communication, Rutgers University.
Walters, L., Wilkins, L., and Walters, T. eds. 1989. Bad tidings:
Communication and Catastrophe. New York: Lawrence Erlbaum.
White, G. F. and Myers, M. F., eds. 1994. "Coping with the flood: The
next phase," Water Resources Update 94-95.
Wilkins, L. 1987. Shared vulnerability: The media and American
perception of the Bhopal disaster. Westport, CT.: Greenwood Press.
Wright, J. M. 1994. "Mitigating flood losses on a national scale: A
critical review," in The great flood of '93 . eds., The American
Meteorological Society. Boston, Mass.
|
