Content uploaded by Jessica Liddell
Author content
All content in this area was uploaded by Jessica Liddell on Nov 01, 2019
Content may be subject to copyright.
Cumulative disaster exposure, gender and the protective action
decision model
Jessica L. Liddell
a,
⁎, Leia Y. Saltzman
b
, Regardt J. Ferreira
b,c
, Amy E. Lesen
d
a
City, Culture and Community-PhD Program, Tulane University School of Social Work, United States of America
b
Tulane University School of Social Work, United States of America
c
Department of Social Work, University of the Free State, South Africa
d
Tulane ByWater Institute, Tulane University, United States of America
ABSTRACTARTICLE INFO
Article history:
Received 19 March 2019
Received in revised form 11 September 2019
Accepted 12 September 2019
Available online 29 October 2019
The relationship between gender, disaster exposure, and the Protective Action Decision Model (PADM) is explored
through a survey administered to 326Gulf Coast residents followingthe Deep-Water Horizon oil spill. Structural Equa-
tion Modeling was used to find that disaster exposure demonstrated a significant negative effect on PADM, such that
greater exposure was associated with lower scores (g = −3.09, p< .001). Similarly, gender was a significant covar-
iate in the model, such that beingfemale was associatedwith an increase in scores(g = 0.33, p< .05).This work high-
lights the relationships between gender, cumulative disaster exposure, and the PADM.
Keywords:
Protective Action Decision Model
Technological disaster
Disaster recovery
1. Introduction
Between the years 2005 and 2015, disasters caused over US $1.3 trillion
in damages and displaced millions of people [1]. These disasters are com-
prised of both naturaldisasters (disasters that are either geophysical, mete-
orological, hydrological, climatological, or biological) and technological
disasters [2]. Technological disasters are disasters that are man-made and
that are generally caused by some type of accident (vehicular, structure col-
lapse, exposure, fire, or chemical) [3]. The United States is considered a
“hot-spot”for technological disasters, and ranks #3 in overall occurrences
of technological disasters, number #4 in number of deaths, number #5 in
number of injuries, number #6 in number of affected people and number
#1 in economic damages [4]. Though the Deepwater Horizon Oil Spill is
the largest and most significant oil spill in the scope of damage it caused,
there have been at least 44 oil spills since the 1969 oil well blowout in
Santa, Barbara California [5]. Other notable spills have included the 1989
Exxon Valdez Spill, the 1979 Ixtoc Spill, the 1994 Morris J. Berman Spill,
the 1971 Texaco Oklahoma Spill, and the 1977 Hawaiian Patriot, among
others [5]. There is evidence that natural and technological disasters may
differ in their short and long-term impacts on individual mental and
physical health, and in their impact on community infrastructure [6–8],
making continued research on technological disasters especially important.
The 2010 Deepwater Horizon oil rig explosion was one of the largest
technological disaster events in United States history [9]. In addition to kill-
ing 11 crewmembers, the explosion caused an estimated 4.9 million barrels
of oil to pour into the Gulf of Mexico, negatively impacting communities
throughout Alabama, Florida, Louisiana, and Mississippi, and causing eco-
nomic damage of at least US $36.8 billion [9]. Because many individuals
living in this region depend on coastal areas for both social, recreational,
and economic resources, the impact of the oil spill on local communities
was immense. It is estimated that the loss to the seafood industry was at
least $4.36 billion dollars [9]. Tourism is an additional important area of
revenue for these areas and is estimated to have cost local economies be-
tween $7.6 billion to $22.7 billion dollars [9]. Loss of income was fre-
quently associated with increased mental distress, and high rates of
anxiety and depression in areas impacted by the spill [10]. Other re-
searchers have noted the increase in “community corrosion”following
the disaster, with communities offering less emotional and instrumental
support to fellow members [11]. However, much of the long-term impact
of this disaster on coastal communities has still yet to be fully determined
[8,12,13].
Coastal areas are especially vulnerable to natural disasters such as hur-
ricanes and flooding, in addition to being at high risk for technological di-
sasters due to their proximity to oil rigs and processing facilities [14]. In
the United States, coastal communities in the Gulf South have been
Progress in Disaster Science 5 (2020) 100042
⁎Corresponding author at: School of Social Work, Tulane University, 127 Elk Place, New
Orleans, LA 70112-2627, United States of America.
E-mail address: jliddell@tulane.edu.(J.L.Liddell).
http://dx.doi.org/10.1016/j.pdisas.2019.100042
2590-0617/©2019 The Authors. Published by Elsevier Ltd. This is an open access
article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Contents lists available at ScienceDirect
Progress in Disaster Science
journal homepage: www.elsevier.com/locate/pdisas
impacted by numerous disasters in the past decade, including the Deepwa-
ter Horizon oil spill and Hurricanes Katrina, Rita, Gustav, Ike, Isaac, and
Harvey [15]. Encouragingly, there is a growing body of research exploring
the factors that impact resilience and coping for individual and communi-
ties following disasters ([10,16–20]). However, additional research is
needed, particularly in the context of technological disasters such as oil
spills.
The increased exposure of individuals living on the U.S. Gulf Coast to
multiple instances of, and types of disasters, also warrants further explora-
tion. Though previous research has tended to focus on individual or com-
munity exposure to a single disaster event, exploration of the impact of
exposure to multiple disasters on disaster recovery and preparedness is
needed [21]. Similarly, few studies of disasters have focused on the rela-
tionship between gender and recovery, risk perception, and future disaster
preparedness in post-disaster settings, and even fewer have explored these
topics following technological disasters [22–24]. There is evidence that fe-
males may be especially vulnerable following disasters, in large part due to
family responsibilities and economic constraints, and susceptibility to do-
mestic violence [25–31], making research needs in this area even more
pressing.
1.1. PADM
The Protective Action Decision Model (PADM) offers researchers a tool
for analyzing three types of respondent perceptions: threat perceptions,
protective action perceptions, and stakeholder perceptions. The PADM
can be used to investigate these perceptions in response to a disaster or en-
vironmental hazard, or in terms of how they informdecision-making before
threats occur [32]. The PADM was developed in part as a way to offer an
alternative explanation for human behavior following disasters in contrast
to protection motivation theory and the theory of planned behavior [33].
The PADM model combines information from an individual's specificsoci-
etal and environmental context, available social information, and personal
experience with a hazard (frequently, some type of disaster) [21]. Individ-
uals indicate how they perceive threats, if they need additional information
about thethreat and where to get that information, ifthere is a need for pro-
tective action, what type of protective action can be taken, and when that
action should be taken [21].
1.2. Gaps in the literature: PADM and disaster exposure
Most previous PADM research has focused on natural disasters [34–36].
The PADM has not often been used in the context of technologicaldisasters
[34] and has not been used to look at the impact of the combination of ex-
posure to multiple hurricane events and oil spills. Though previous research
with the PADM has indicated that disaster experience increases threat per-
ception, this relationship has not always been consistent [21] and associa-
tion with demographic factors has been variable, indicating that there is a
need for further exploration of the relationship between demographic var-
iables and the PADM [35,36]. Though prior studies have explored the rela-
tionship between previous hurricane exposure and decision-making (e.g.
[24]) and exposure to technological disasters and decision-making [34], lit-
tle has explored the combined effects of these on the PADM [21]. To date,
the PADM has been infrequently applied to the Deepwater Horizon oil spill
[37] and the impactof prior exposure to multiple disasterevents has not yet
been researched among those impacted by the DeepwaterHorizon oilspill.
1.3. Gaps in the literature: gender
Though gender in disaster contexts is increasingly being addressed,
there is relatively scant research that has specifically focused on how indi-
viduals of each gender may be differently impacted in their perception of
disaster risk and decision-making, and the important implications this
may have on disaster response and prevention [24,38]. The studies that
do exist tend to focus on natural disasters, leaving a gap in the literature
concerning gender differences following a technological disaster event.
This gap is concerning since the research that does exist suggests that
women and men are often impacted differently by disasters [22,27].
These differences may exist because women are frequently the care-givers
for both immediate and extended family members in post-disaster settings
and often have increased domestic and family responsibilities before and
following a disaster, their frequent economic dependence on male bread-
winners, limited economic opportunities, and a wage gap which penalizes
women [25,28,29]. These factors are compounded by decreased access to
social services and compromised social networks following a disaster
[25,28,39–41].
Following Hurricane Katrina, women reported a significant increase in
both emotional and physical victimization, which was not found for men
(Schmacher et al., 2010). In a study looking at health outcomes following
the Deepwater Horizon oil spill,women were more at risk for negative men-
tal health outcomes [27]. In lower SES (socioeconomic status) countries,
there is evidence that women receive less aid following disasters, and in
one study they were seven times more likely to die in the post-disaster re-
covery period than their male counterparts [42]. This is the first research
to look explicitly at the role of gender and the PADMfollowing the Deepwa-
ter Horizon oil spill.
1.4. Current study
Precautionary measures taken before disasters occur foster prepared-
ness for future traumatic events and promote resilience [16,19,20,43].
This study specifically seeks to investigate the preparedness measures—
and predictors of disaster preparedness—taken by individuals and house-
holds in the U.S. Gulf Coast region to mitigate the impact of a future hydro-
carbon event, such as the 2010 Deepwater Horizon oil spill. The aims of this
study were to 1. Explore the impact of cumulative exposure to natural and
oil spill disasters on the PADM and 2. Explore gender differences on the
PADM following the Deepwater Horizon oil spill. This study provides a
unique opportunityto explore the relationships between the cumulative ef-
fect of trauma exposure and the PADM by gender. This research makes an
important contribution to the body of literature investigating post-disaster
recovery and preparedness and has significant implications for program-
ming and interventions used by practitioners.
2. Methods
This study uses a cross-sectional design, and asks participants questions
related to risk perception, preparedness measures, and belief in the likeli-
hood of future natural and oil spill events on the U.S Gulf Coast. Individual
participants were each asked a series of questions during a 60-min in-
person survey. The survey was carried out in Galliano (Lafourche Parish)
and Port Sulphur (Plaquemines Parish) in southeastern Louisiana, and in
Bayou La Batre (Mobile County), Alabama, areas which have all experi-
enced multiple disasters in the past decade. Surveys were administered by
trained data collectors in these three communities between June 2017
and November 2017.
2.1. Participants
The target sample consisted of individuals residing in areas surrounding
Port Sulphur and Galliano, Louisiana and Bayou La Batre, Alabama on the
U.S. Gulf Coast because of the high prevalence of disaster exposure (and ex-
posure to multiple types of disasters) in these areas. Within each sampling
site the goal was to collect a minimum of 100 in-person surveys. Survey
subjects were recruited for participation in the study through a mixture of
snowball sampling and the use of an existing database maintained by an
outside vendor. Recruitment packages for prospective participants were
sent by the outside vendor and included a recruitment letter containing in-
formation about the study and both mail-in instructions and a web-link for
scheduling an appointment to participate. The research team did not have
access to the full names or contact information for any of the study partici-
pants. All adults over the age of 18 years, residing in the areas in and
J.L. Liddell et al. Progress in Disaster Science 5 (2020) 100042
2
surrounding Galliano and Port Sulphur, Louisiana and Bayou La Batre, Ala-
bama were eligible to participate in the study.
2.2. Procedure
Both the online and mail-in scheduling recruitment forms listed poten-
tial times for participating in a 60 min in-person survey at a specified
local site. Once prospective participants submitted a response via postal
mail or through the weblink, participants were contacted by the vendor
to schedule an appointment with the research team to participate in the
in-person survey. The research team received a schedule from the vendor
for carrying out the in-person surveys, whichcontained only a 4-digitiden-
tifier and last name for each participant. The research team did not have ac-
cess to any other information about the participants, and the schedule was
destroyed after data collection was completed to maximize participant pri-
vacy. The outside vendor destroyed any identifying records linking partici-
pants to the study after data collection ended.
All members of the data collection team completed Institutional Review
Board requirements for conducting research (e.g. Collaborative Institu-
tional Training Initiative (CITI) training). The data collectors—PhD and
masters students—also completed trainings in cultural sensitivity and
data collection before data collection began. Survey administration was su-
pervised by the research team PI and Co-PI's, with a minimum of one super-
visor being present at a given data collection event to monitor data
collectors at all times. Verbal informed consent was obtained for all partic-
ipants before beginning the survey, and all participants were informed that
they did not have to answer any question they did not want to, and that
they were free to end the survey at any time. Participants were also pro-
vided with an information sheet containing the contact information for
study PIs and additional information about the study. Though the magni-
tude of risk for participation in the study was minimal, data collectors
were trained in assisting participants with accessing resources if they re-
ported any discomfort as a result of study participation. Participants were
providedwith a $50.00Walmart giftcard immediately following their par-
ticipation in the survey.
Data was collected through use of either a handheld tablet computer or
a paper-based version depending on the preference of the participant.
When using the online, tablet-based version of the survey, data collectors
inputted the survey question responses into the tablet for the participant.
The final sample included 326 individuals. IRB approval was obtained
from XXX (omitted for blind review) University (IRB # 16-997431U)
prior to initiating data collection. For further discussion of the follow-up
procedures following data collection please refer to (omitted for blind
review).
2.3. Measures
Participants were asked the questions described below in Table 1.(See
Tables 2–5.)
2.4. Analysis approach
Paired-samples t-tests were conducted to compare scores on the sepa-
rate subscales of the PADM (Hazard Adjustment Perceptions, Perceived
Event Likelihood, Perceived Consequences and Cumulative PADM scores)
between men and women. Paired-samples t-tests were also conducted to
compare scores on the separate components of the PADM (Hazard Adjust-
ment Perceptions, Perceived Event Likelihood, Perceived Consequences
and Cumulative PADM scores) between those exposed to disaster, and
those not exposed to disaster.
A Structural Equation Model (SEM) was estimated in Stata Version 13.1
using maximum likelihood estimation to test the relationship between ex-
posure to disasters and the PADM (n= 218). List-wise deletion was used
as the method for addressing missing data, reducing the sample from 326
to 218. The model was re-estimated based on the modification indices to
allow covariation between error terms in the measurement components
of the model. These modifications were allowed to improve overall model
fit. The final model included six indicators of the latent exogenous variable
exposure and seven indicators of the endogenous latent variable PADM. In
Table 1
Questions and response options of measure items.
Measure questions Response options (coded as X)
PADM questions
“Have you ever prepared for a natural
disaster such as a hurricane or a flood?”
*If respondents answer yes, they are
then asked the following questions.
•Yes
•No
“Do you think your preparations are
effective for protecting the safety of you
and members of your household for a
future natural disaster?”
•Very ineffective (0)
•Ineffective (1)
•Effective (2)
•Very effective (3)
•Don't know/NA
“How effective do you think your
preparations are in limiting the
negative financial impact of a future
natural disaster?”
*Same as above
“How likely is it that in the next five
years another large oil spill disaster like
the Deepwater Horizon oil spill will
occur in the Gulf of Mexico?”
•Very unlikely (0)
•Unlikely (1)
•Likely (2)
•Very Unlikely (3)
•Don't know/NA
“How likely is it that in the next five
years another major natural disaster
like a hurricane will occur in the Gulf of
Mexico?”
*Same as above
“How worried are you now about any
ongoing impacts of the Deepwater
Horizon oil spill on the physical health
of you or any member of your
immediate family”
•Not at all worried (0)
•A little worried (1)
•Moderately worried (2)
•Very worried (3)
•Don't know/NA
“How worried are you now about any
ongoing impacts of the Deepwater
Horizon oil spill on the economy in your
community?”
*Same as above
“How worried are you now about any
ongoing impacts of the Deepwater
Horizon oil spill on the relationships
with family and friends for you or any
member of your immediate family?”
*Same as above
Gender •Female (0)
•Male (1)
Race •White
•Black or African American
•Native American/Alaska Native
•Asian Indian
•Chinese
•Filipino
•Japanese
•Korean
•Vietnamese
•Cambodian
•Native Hawaiian
•Other Pacific Islander
•Other*Because of the relatively small
numbers of non-White participants, for
purposes of this analysis, participants
were collapsed into those identifying as
White (coded as 1) and those identify-
ing as non-White (coded as 0).
Hispanic/Latino/Spanish origin •Yes
•No
* Those who reported not identifying as
Hispanic, Latino or Spanish were coded
as 1 and those who did as 0.
Exposure to previous disasters:
Had participant been in the region during:
•Yes (1)
•No (0)
Hurricane Katrina *Same as above
Hurricane Rita *Same as above
Hurricane Gustav *Same as above
Hurricane Ike *Same as above
The Deepwater Horizon Oil Spill *Same as above
Hurricane Isaac *Same as above
J.L. Liddell et al. Progress in Disaster Science 5 (2020) 100042
3
addition, the covariate gender was added to account for the differences in
preparedness across genders. Model fit was determined based on a series
of goodness of fit indices including chi-square (χ
2
), root mean squared
error of approximation (RMSEA), comparative fit index (CFI), Tucker-
Lewis fit index (TLI), and coefficient of determination (CD). The model
was estimated and compared to the following indices of fit: χ
2
/df <2,
RMSEA <0.05, CFI >0.90, TLI >0.90, CD >0.80 [44].
3. Results
The final sample for this analysis was 326 participants. The average age
of participants was 55.05 years (SD = 15.80) with 61.0% of participants (n
= 199) identifying as female and 39. 0% (n= 127) as male. Over half of re-
spondentsidentified as White (n= 160, 51.6%) with 150 (48.4%) identify-
ing as non-White. The majority of participants reported not being Hispanic
or Latino (n= 313, 96.0%) with 13 (4.0%) identifying as Hispanic or
Latino.
Many (n= 141, 64.7%) participants reported being in the region for all
of the six disaster experiences they were questioned about. For Hurricane
Katrina 204 (62.6%)were present (n= 122, 37.4% not present); for Hurri-
cane Rita, 170 (52.1%) were present, (n= 156, 47.9% not present); for
Hurricane Gustav, 167 (51.2%) were present, (n= 159, 48.8% not pres-
ent); for Hurricane Ike, 166 (50.9%) were present, (n= 160, 49.1% not
present); for Hurricane Isaac, 172 (52.8%) were present, (n=160,
47.2%, not present); and for the Deepwater Horizon oil spill, 207 individ-
uals (63.5%) reported being in the area (n= 119, 36.5% not in the area).
For the Hazard Adjustment Perception questions, respondents reported
ameanof3.47(SD=0.72)(scaleof1–4) when asked “Do you think your
preparations are effective for protecting the safety of you and members of
your household for a future natural disaster?”and a mean of 2.94 (SD =
0.85) (scale of 1–4) when asked “How effective do you think your
preparations are in limiting the negative financial impact of a future natural
disaster?”For Perceived EventLikelihood questions, participants reported a
mean of 2.51 (SD = 1.05) (scale of 1–4) for “How likely is it that i n the next
five years another large oil spill disaster like the Deepwater Horizon oil spill
will occur in the Gulf of Mexico?”and a mean of 3.54 (SD 0.68) (scale of
1–4) for “How likely is it that in the next five years another major natural
disaster like a hurricane will occur in the Gulf of Mexico?”For Perceived
Oil Spill Disaster questions, respondents had a mean of 2.54 (SD = 1.23)
(scale of 1–4) for “How worried are you now about any ongoing impacts
of the Deepwater Horizon oil spill on the physical health of you or any
member of your immediate family”, a mean of 3.02 (SD = 1.13) for
“How worried are you now about any ongoing impacts of the Deepwater
Horizon oil spill on the economy in your community?”and a mean of
2.11 (SD = 1.23) when asked “How worried are you now about any ongo-
ing impacts of the Deepwater Horizon oil spill on the relationships with
family and friends for you or any member of your immediate family?”
The results of the structural equation model suggest that the model fit
the data moderately well (χ
2
(df) = 143.85(73), p< .001). The Root
mean square error of approximation (RMSEA) was greater thanthe desired
cut point of 0.05 (RMSEA = 0.067, CI 0.05–0.08, p= .05). However, the
CFI, TLI, and CD all indicated that the model fit the data well (CFI =
0.94, TLI = 0.92, and CD = 0.93). Within the path model, exposure had
a direct and negative relationship with PADM (β=−3.09, p< .001). Sec-
ondly, being female had a direct andpositive relationship PADM (β= 0.33,
p< .05) (Fig. 1).
To more explicitly explore the differences identified in the SEM model,
an independent samples t-test was performed to explore differences be-
tween men and women on the PADM. An additional independent samples
t-test was performed to assess differences between those exposed to disas-
ters and those not exposed. There was not a significant different in the av-
erage Hazard Adjustment Perceptions.
scores for men (M=2.23,SD = 0.66) and women (M=2.22,SD =
0.63); t(286) = 0.21, p= .57. There was also not a significant different
in the mean Perceived Event Likelihood scores for men (M=1.98,SD =
0.76) and women (M=2.06,SD = 0.74); t(309) = −0.91, p= .18. How-
ever, there was a significant different in the mean of Perceived Conse-
quences scores for men (M= 1.33, SD = 0.99) and women (M= 1.71,
SD = 0.96); t(317) = −3.31, p= .00, such that women had higher aver-
age scores of perceived consequences. There was also a significant different
in the mean Cumulative PADM scores for men (M= 1.75, SD = 0.60) and
women (M= 1.95, SD =0.55);t(323) = −3.05, p=.00,againwith
women reporting higher average scores on the cumulative PADM as com-
pared to men.
Table 2
Descriptive characteristics of study sample, n= 326.
n m/% SD Min–Max
Demographics
Age (in years) 326 55.05 15.80 18–89
Gender 0.41 1–2
Female 199 61.0
Male 127 39.0
Race 0.52 1–2
Non-White 150 48.4
White 160 51.6
Hispanic/Latino 0.76 1–2
Hispanic/Latino 13 4.0
Not Hispanic/Latino 313 96.0
Table 3
Model variables, n= 326.
n m/% SD Min–Max
Disaster exposure (Yes)
Hurricane Katrina 204 62.6 0.27 1–2
Hurricane Rita 170 52.1 0.45 1–2
Hurricane Gustav 167 51.2 0.45 1–2
Hurricane Ike 166 50.9 0.46 1–2
Hurricane Isaac 172 52.8 0.45 1–2
DWH oil spill 207 63.5 0.24 1–2
PADM
Hazard adjustment perceptions
Effective for safety 326 3.47 0.72 1–4
Effective for financial 326 2.94 0.85 1–4
Perceived event likelihood
Future oil spill 326 2.51 1.05 1–4
Future nat. disaster 326 3.54 0.68 1–4
Perceived consequences
Worry health 326 2.54 1.23 1–4
Worry economy 326 3.02 1.13 1–4
Worry relationships 326 2.11 1.23 1–4
Table 4
Relationship of gender and the PADM, n=326.
Men Women
N and mean SD N and mean SD T-test
Hazard adjustment perceptions 113(2.23) 0.66 175(2.22) 0.63 0.21
Perceived event likelihood 121(1.98) 0.76 190(2.06) 0.74 −0.91
Perceived consequences 124(1.33) 0.99 195(1.71) 0.96 −3.31***
Cumulative PADM score 126(1.75) 0.60 199(1.95) 0.55 −3.05***
*p<.05,**p<.01,***p<.001.
Table 5
Relationship of disaster exposure and the PADM, n=326.
Exposure No Exposure
N and Mean SD N and Mean SD t-test
Hazard adjustment perceptions 277(2.23) 0.65 11(2.23) 0.51 0.01
Perceived event likelihood 298(2.05) 0.74 13(1.62) 0.74 −2.06*
Perceived consequences 308(1.55) 0.98 11(1.68) 1.03 0.43
Cumulative PADM score 312(1.87) 0.61 13(1.76) 0.72 −0.65
*p<.05,**p<.01,***p<.001.
J.L. Liddell et al. Progress in Disaster Science 5 (2020) 100042
4
There was not a significant different in the mean Hazard Adjustment
Perceptions scores for those exposed (M= 2.23, SD = 0.65) and those
not exposed to disaster (M= 2.23, SD = 0.51); t(286) = 0.01, p= .53.
There was not a significant different in the average Perceived Consequences
scores for those exposed (M=1.55,SD = 0.98) and those not exposed to
disaster (M=1.68,SD = 1.03); t(317) = 0.43, p= .67. There was not a
significant different in the average Cumulative PADM scores for those ex-
posed (M=1.87,SD = 0.61) and those not exposed to disaster (M=
1.76, SD = 0.72); t(323) = −0.65, p= .26. However, there was a signifi-
cant different in the mean Perceived Event Likelihood scores for those ex-
posed (M=2.05,SD = 0.74) and not exposed to disaster (M=1.62,SD
= 0.74); t(309) = −2.06, p= .02; such that those exposedreported higher
average perceived likelihood of future disasters.
4. Discussion
The key findings from our study are outlined in Fig. 2.Thisstudy
found that being female was positively associated with PADM scores.
This finding is congruent with PADM research finding female gender
had positive correlations with perceived storm characteristics, the im-
portance of official warnings and social cues, expected personal impact
of the disaster and evacuation decisions [24,38]. It is also congruent
with previous studies showing that in some disaster contexts, females
maybemoreresilientthantheirmalecounterparts[25,30,39,45,46].
These differences have not been widely explored, thus adding an ex-
plicit analysis of the role of gender to this model offers an important
contribution to the body of literature looking at risk perception and di-
saster recovery following disasters. The results of the paired-samples t-
test indicate that women, compared to men, are more likely to report
continuing concern about the possible negative impact following on a
disaster on their health, the local economy and the relationships in
their community. Women also had statistically significantly higher
scores overall on the PADM, lending support to previous research that
women and men interpret and respond to disasters in different ways
[25,30,39,45,46] and that there is a need for gender-sensitive disaster
response and preparedness services [26,31]. It is also interesting to
note that we di d not observe statistically significant differences between
men and women related to their belief of the efficacy of their actions to
protect the physical safety and financial security of their families (Haz-
ard Adjustment Perceptions) or in their belief in the likelihood of a fu-
ture natural disaster or oil spill (Perceived Event Likelihood).
The finding that increased disaster exposure was negatively associated
with the PADM has significant implications for policy and programmatic in-
terventions since it is expected that the number and type of disasters expe-
rienced by any one individual or community is expected to rise [15]. It is
also notable that manyof these participants had experienced multiple disas-
ters. These findings are congruent with research by Lindell and Hwang
[21], suggesting that multi-hazard exposure has a unique impact on the
PADM and that direct exposure to disasters does not always lead to an in-
crease in protective actions or belief in the efficacy of those actions. It is es-
sential that future research further explore the unique role that multi-
hazard exposure has on risk perception and preparedness. In the case of
technological disasters such as oil spills, the impact of the disaster may in
some instances be delayed, with some impacts only visible months, or
even years after the initial disaster [13], further complicating this relation-
ship. The term“corros ive community”has been used to describethese long-
term negative impacts, and in the case of oil spills may frequently be related
to stress surrounding litigation and the negative impact this has on
Fig. 1. SEM model.
Overall respondents exposed to disasters were statistically significantly more likely to report that they
believed a future oil spill or natural disaster was likely.
Being female was positively associated with PADM scores.
Women had statistically significantly higher scores overall on the PADM.
Women as compared to men are more likely to report:
Continuing concern about the possible negative impact following a disaster on their health.
Continuing concern about the local economy and the relationships in their community.
We did not observe statistically significant differences between men and women related to:
Their belief of the efficacy of their actions to protect the physical safety and financial security
of their families.
Their belief in the likelihood of a future natural disaster or oil spill.
Fig. 2. Key findings from the study.
J.L. Liddell et al. Progress in Disaster Science 5 (2020) 100042
5
individuals, the community as a whole, and relationships within the com-
munity [8]. Though caution must be used because of the small number of
individuals not exposed to any disaster in these communities, it is interest-
ing that the only differences observed between those not exposed, and
those exposed to disasters relates to the Perceived Event Likelihood. The
finding of the t-test indicates that those exposed to disasters were statisti-
cally significantly more likely to report that they believed a future oil spill
or natural disaster was likely. This has important implications for education
and outreach since this suggests that those who have not yet been exposed
to a disaster may not take future action because they feel like a disaster is
unlikely to impact them, even though they live in a region where disasters
are frequent, and are becoming increasingly more common.
Additional implications of these findings include the need for increased
attention to the information provided by government agencies before, dur-
ing, and following disasters, and how this information may be perceived
and utilized differently by gender. In a study exploring the informational
needs of women following the Deepwater Horizon oil spill, 50% reported
lacking the information they needed to make informed decisions about sea-
food consumption, and reported general high levels of distrust of govern-
ment officials [47]. Accurate information about disaster risk perception
and preparedness is essential for ensuring adequate government and NGO
responses prior to, during, and following future disasters. Because of the in-
creased risk of both natural and technological disasters in coastal areas,
continued research on the impact of technological disasters, such as the
Deepwater Horizon oil spill, on recovery, risk perception, and response is
essential.
4.1. Limitations
This dataset has some limitations which should be noted. Because this
data comes from cross-sectional surveys and was collected post-disaster,
we cannot make any claims about causality because we do not know the
levels of disaster risk perception prior to the oil spill. Exploring changes
in the PADM, particularly as it relates to gender differences, is an important
area of future research. Participants were also interviewed seven years after
the Deepwater Horizon oil spill, which may impact their recall and risk per-
ception. Additionally, the findings in this study may not be representative
of all individuals living in the Gulf South. In this data set there is limited
representation of some minorities. A large portion of the sample (51.6%)
identified as white, with 18.3% identifying as African American, and only
4.1% reporting being of Hispanic or Latino origin. Though other articles
have explored the topic of race within this sample (omitted for blind re-
view), this is an important limitation of our study. Additional limitations in-
clude the dichotomous categorizing of individuals into either male or
female categories. Though these categories were self-defined by individ-
uals, and individuals were given the option of listing a gender other than
male or female, no one in this sample identified outside of this binary and
therefore this study doesn't include an analysis of difference across gender
identity outside of male and female. This is an important gap in the litera-
ture that future studies should address [48]. Another important item to
note is that this model does not account for covariates other than gender
that influence the relationship between repeated disaster exposure and
the PADM.
In this study participants were asked about both their exposure to natu-
ral and oil spill disasters. Future research could explore if there are differ-
ences between increased exposure to each of these types of disasters and
the PADM, for example, exposure to multiple oil spills. Additional studies
could also explore differences among sub-groups of men and women:
here we focus on broad differences between men and women and it is ex-
pected that this may vary by occupation, race and other factors. Finally, a
qualitative exploration of gender differences in preparedness and disaster
recovery would explicate specific differences in preparedness beliefs and
motivations following a technological disaster. It is hoped and anticipated
that this study will encourage future research in this area that will explore
these factors more in depth.
5. Conclusion
This study contributes to our understanding of the relationship between
disaster exposure, gender, and the PADM, and provides support for previ-
ous research that has investigated the PADM following disasters [21, 29,
30 32, 33]. This study has important implications for how researchers
and practitioners conceptualize the impact of exposure to multiple disaster
events and the PADM, and adds to the scant literature that applies a gender
lens to understanding the PADM. The results of this study indicate that in-
creased disaster exposure is negatively associated with the PADM. How-
ever, being female had a positive impact on the PADM.
Though there is an increasing body of research investigating the rela-
tionship between both natural disasters and technological disasters (like
oil spills) on the PADM, the relationship of exposure to both natural disas-
ters and technological ones has been less explored. This researchlends sup-
port to findings that the impact of exposure to multiple disaster events
negatively impactsthe PADM. In contrast toprevious research which solely
looks at the impact of a single disaster event on the PADM, these findings
contribute to our understanding of the increasing likelihood of exposure
to multiple disaster events and types of disasters and its impact in post-
disaster contexts.
Of particular interest is the finding that being female was positively as-
sociatedwith the PADM. Men maybe more vulnerablein certain areas post-
disaster, perhaps because they may have fewer pre-existing social support
mechanisms and because of stigma surrounding help-seeking behavior
[45,49]. These findings provide practitioners with some baseline knowl-
edge about the impact of exposure to multiple disaster events and the role
of gender on the PADM following the Deepwater Horizon oil spill. It is
hoped that this research will aid researchers, practitioners, and policy-
makers in focusing their interventions on the areas where it is most needed
in order to empower resilient individuals and communities, and to identify
the areas where there may be current gaps in services. This manuscript of-
fers a unique perspective as it highlights gender differences in the relation-
ship between cumulative disaster exposure and the PADM model.
Acknowledgements
This research was made possible by a grant from The Gulf of Mexico Re-
search Initiative [231501-00]. Data are publicly available through the Gulf
of Mexico Research Initiative Information &Data Cooperative (GRIIDC) at
https://data.gulfresearchinitiative.org (doi: https://doi.org/10.7266/n7-
h9ty-ce44). The authors wish to thank our collaborators on the Consortium
for Resilient Gulf Communities project, as well as our Tulane University
data collection team. Special thanks to our partner organizations in the
three communities where we collected data, and to the residents of Louisi-
ana and Alabama who participated in this research.
References
[1] United Nations (UN). Sendai framework for Disaster Risk Reduction 2015–2030′,3rd
United Nations Conference on Dis aster Risk Reduc tion. availab le at: http://www.
wcdrr.org/upload s/Sendai_Framework_for_Disaster _Risk_Reduction_2015-2030.pdf;
2015, Accessed date: 14 September 2015.
[2] Guha-Sapir D, Vos F, Below R, Ponserre S.Annual Disaster Statistical Review 2011: The
Numbers and Trends. Centre for Research on the Epidemiology of Disasters (CRED);
2012 Available from: http://lib.riskreductionafrica.org/bitstream/handle/123456789/
1141/annual%20disast er%20statistical%20 review%202011.%20t he%20numbers%
20and%20trends.pdf?sequence=1, Accessed date: 15 August 2018.
[3] Shaluf IM. An overview on the technological disasters. Disaster Prevention and Manage-
ment: An International Journal 2007;16(3):380–90.
[4] Shen G, Hwang SN. Revealing global hot spots of technological disasters: 1900–2013.
Journal of Risk Research 2018;21(3):361–93.
[5] NOAA. Large st Oil Spills Af fecting U.S. Wat ers Since 1969. available at : https://
response.res toration.noaa .gov/oil-and -chemical-spi lls/oil-spills/largest-o il-spills-
affecting-us-waters-1969.html; 2019, Accessed date: 11 September 2019.
[6] Arata CM, Picou JS, Johnson GD, McNally TS. Copingwith technologicaldisaster: an ap-
plication of the conservation of resources model to theExxon Valdez oil spill. Journal of
Traumatic Stress: Official Publication of The International Society for Traumatic Stress
Studies 2000;13(1):23–39.
J.L. Liddell et al. Progress in Disaster Science 5 (2020) 100042
6
[7] Green BL,Solomon SD. The mental health impact of natural and technological disasters.
In: Freedy JR, Hobfoll SE, editors. Traumatic Stress: From Theory to Practice. Boston,
MA: Springer; 1995. p. 163–80.
[8] Picou JS, Marshall BK, Gill DA. Disaster, litigation, and the corrosive community. Soc
Forces 2004;82(4):1493–522.
[9] Smith L, Smith M, Ashcroft P. Analysis of environmental and economic damages from
British Petroleum's Deepwater horizon oil spill. Albany Law Rev 2011;74:563–85.
[10] Morris Jr JG, Grattan LM, Mayer BM, Blackburn JK. Psychological responses and resil-
ience of people and communities impacted by the Deepwater Horizon oil spill. Transac-
tions of the American Clinical and Climatological Association 2013;124:191.
[11] Mayer B, Running K, Bergstrand K. Compensation and community corrosion: perceived
inequalities, social comparisons, and competition following the Deepwater Horizon oil
spill. Sociological Forum 2015;30(2):369–90.
[12] Cope MR, Slack T, Blanchard TC, Lee MR. Does time heal all wounds? Community at-
tachment,natural resource employment, and health impacts in the wakeof the BP Deep-
water Horizon disaster. Soc Sci Res 2013;42:872–81.
[13] Gill DA, Picou JS, Ritchie LA. The Exxon Valdez and BP oil spills:a comparison of initial
social and psychological impacts. American Behavioral Scientist 2012;56(1):3–23.
[14] CutterSL, Emrich CT. Moral hazard, social catastrophe: the changing face of vulnerabil-
ity along the hurricane coasts. Ann Am Acad Pol Soc Sci 2006;604(1):102–12.
[15] FEMA. Gulf of Mexico coastal information. available at: https://www.fema.gov/gulf-
mexico-coastal-information; 2018, Accessed date: 5 May 2018.
[16] Buckingham-Howes S, Sreekumar P, Morris G, Grattan LM.Resilience after the Deepwa-
ter Horizon oil spill. Disaster Prevention and Management 2017;26(5):597–610.
[17] Connor KM, Davidson RT. Development of a new resilience scale: the Connor Davidson
resilience scale. Depress Anxiety 2003;18:76–82.
[18] Cutter SL. The landscape of disaster resilience indicators in the USA.Nat Hazards 2016;
80(2):74 1–58.
[19] Grattan LM, Brumback B , Roberts SM, Bucking ham-Howes S, Toben AC, Morris G.
Bouncing back after the Deepwater Horizon oil spill. Disaster Prevention and Manage-
ment: An International Journal 2017;26(2):122–33.
[20] NorrisFH, Stevens SP, Pfefferbaum B, Wyche KF, Pfefferbaum RL. Community resilience
as a metaphor, theory, set of capacities, and strategy for disaster readiness. Am J Com-
munity Psychol 2008;41(1–2):127–50.
[21] Lindell MK, Hwan g SN. Households ' perceived pers onal risk and responses in a
multihazard environment. Risk Analysis: An International Journal 2008;28(2):539–56.
[22] Enarson E, Meyreles L. International perspectives on gender and disaster: differences
and possibilities. Int J Sociol Soc Policy 2004;24(10/11):49–93.
[23] Ginige K, Amaratunga D, Haigh R. Mainstreaming gender in disaster reduction: why and
how? Disaster Prevention and Management: A n International Jour nal 2009;18(1):
23–34.
[24] Huang SK, Lindell MK,Prater CS, Wu HC, Siebeneck LK. Household evacuation decision
making in response to Hurricane Ike. Natural Hazards Review 2012;13(4):283–96.
[25] Enarson E, Morrow BH. The Gendered Terrain of Disaster. Westport, CT: Routledge;
1998.
[26] Eriksen C. Gendered risk engagement: challenging the embedded vulnerability, social
norms and power relations in conventional Australian bushfire education. Geographical
Research 2014;52(1):23–33.
[27] Fan AZ, Prescott MR, Zhao G, Gotway CA, Galea S. Individual and community-level de-
terminants of mental and physical health after the Deepwater Horizon oil spill: findings
from the Gulf States Population Survey. J Behav Health Serv Res 2014:23–41.
[28] Jenkins P, Phillips B. Battered women, catastrophe, and the context of safety after Hur-
ricane Katrina. NWSA Journal 2008:49–68.
[29] Lauve-Moon K, Ferreira RJ. An exploratory investigation: post-disaster predictors of in-
timate partner violence. Clinical Social Work Journal 2017;45(2):124–35.
[30] NorrisF. Disasters and domestic violence. NationalCenter for PTSD; 2014.http://www.
ptsd.va.gov/professional/trauma/d isaster-terrorism/disasters-do mestic-violence.asp
[Accessed 5 May 2016].
[31] Whittaker J, Eriksen C, Haynes K. Gendered responses to the 2009 Black Saturday bush-
fires in Victoria, Australia. Geographical Research 2016;54(2):203–15.
[32] Lindell MK, Perry RW. The protective action decision model: theoretical modifications
and additional evidence. Risk Analysis: An International Journal 2012;32(4):616–32.
[33] Lindell MK, Perry RW. Communicating Environmental Risk in Multiethnic Communi-
ties. Thousand Oaks, CA: Sage; 2004.
[34] Heath RL, Lee J, Palenchar MJ, Lemon LL. Risk communication emergency response
preparedness: contextual assessment of the protect ive action decisi on model. Risk
Anal 2018;38(2):333–44.
[35] LindellM. North American cities at risk: household responses to environmental hazards.
In: Joffe E, Rossetto T, Adams J, editors. Cities at Ri sk. Dordrecht: Spring er; 2013.
p. 109–30.
[36] Rickard LN, Yang ZJ, Schu ldt JP, Eosco GM, Scherer CW, Daziano RA. Sizing up a
superstorm: exploring the role of recalled experience and attribution of responsibility
in judgments of future hurricane risk. Risk Anal 2017;37(12):2334–49.
[37] Rung AL, Gaston S, Oral E, Robinson WT, Fontham E, Harrington DJ, et al. Depression,
mental distress, and domestic conflict among Louisiana women exposed to the Deepwa-
ter Horizon oil spill in the WaTCH study. Environ Health Perspect 2016;124(9):1429.
[38] Lindell MK, Lu JC, Prater CS. Householddecision making and evacuation in response to
Hurricane Lili. Natural Hazards Review 2005;6(4):171–9.
[39] Enarson E, Scanlon J. A case study in Canada's Red River valley. Applied Behavioral Sci-
ence Review 1999;7(2):103–24.
[40] Myers DG, Wee DF. Disaster Mental Health Services: A Primer for Practitioners. New
York and Hove: Brunner-Routledge; 2005.
[41] Wisner B, Blaikie P, Cannon T, Davis I. At Risk: Natural Hazards,People's Vulnerability
and Disasters. London and New York: Routledge; 2004.
[42] Bradshaw S, Fordham M. Women, girls and disasters: a review for DIFD. available at:
https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/
236656/women-girls-disasters.pdf; 2013, Accessed date: 12 January 2015.
[43] Levy J, GopalakrishnanC. Promoting ecological sustainability and community resilience
in the US Gulf Coast after the 2010 Deepwater Horizon oil spill. Journal of Natural Re-
sources Policy Research 2010;2(3):297–315.
[44] Hu LT, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: con-
ventional criteria versus new alternatives. Struct Equ Model Multidiscip J 1999;6(1):
1–55.
[45] Liddell J, Ferreira RJ. Predictors of individual resilience characteristics among individ-
uals ages 65 and older in post-disaster settings. Disaster Med Public Health Prep 2018:
1–9.
[46] Morrow BH, Phillips B. ‘What's gender got to do with it’? Int J Mass Emerg Disasters
1999;17(1):5–13.
[47] Simon-Friedt BR, Howard JL, WilsonMJ, Gauthe D, Bogen D, Nguyen D, et al. Louisiana
residents' self-reported lack of information followingthe Deepwater Horizon oil spill: ef-
fects on seafood consumptionand risk perception. J Environ Manage 2016;180:526–37.
[48] Gaillard JC, Sanz K, Balgos BC, Dalisay SNM, Gorman-Murray A, Smith F, et al. Beyond
men and women: a critical perspective on gender and disaster. Disasters 2017;41(3):
429–47.
[49] Adams ML, Ford JD, Dailey WF. Predictors of help seeking among Connecticut adults
after September 11, 2001. Am J Public Health 2004;94(9):1596–602.
J.L. Liddell et al. Progress in Disaster Science 5 (2020) 100042
7
