Risk Analysis, Vol. 31, No. 10, 2011 DOI: 10.1111/j.1539-6924.2011.01616.

x
Emotions, Trust, and Perceived Risk: Affective and
Cognitive Routes to Flood Preparedness Behavior
Teun Terpstra

Despite the prognoses of the effects of global warming (e.g., rising sea levels, increasing river
discharges), few international studies have addressed how ood preparedness should be stim-
ulated among private citizens. This article aims to predict Dutch citizens ood preparedness
intentions by testing a path model, including previous ood hazard experiences, trust in pub-
lic ood protection, and ood risk perceptions (both affective and cognitive components).
Data were collected through questionnaire surveys in two coastal communities (n = 169,
n = 244) and in one river area community (n = 658). Causal relations were tested by means
of structural equation modeling (SEM). Overall, the results indicate that both cognitive and
affective mechanisms inuence citizens preparedness intentions. First, a higher level of trust
reduces citizens perceptions of ood likelihood, which in turn hampers their ood prepared-
ness intentions (cognitive route). Second, trust also lessens the amount of dread evoked by
ood risk, which in turn impedes ood preparedness intentions (affective route). Moreover,
the affective route showed that levels of dread were especially inuenced by citizens nega-
tive and positive emotions related to their previous ood hazard experiences. Negative emo-
tions most often reected fear and powerlessness, while positive emotions most frequently
reected feelings of solidarity. The results are consistent with the affect heuristic and the
historical context of Dutch ood risk management. The great challenge for ood risk man-
agement is the accommodation of both cognitive and affective mechanisms in risk communi-
cations, especially when most people lack an emotional basis stemming from previous ood
hazard events.
KEY WORDS: Affect heuristic; emotions; experience; ood hazard; risk perception; trust
1. INTRODUCTION
Floods have devastating effects all over the
globe, both in terms of material damage
(1)
as well
as in lost lives.
(2)
Because global warming will put
low-lying, densely populated deltas and river area
communities at even greater risk,
(3)
the improvement
of public ood protection as well as citizens ood
preparedness will become a worldwide challenge.
Global warming is one of the driving forces behind
the debate on ood risk management in the Nether-

Address correspondence to Teun Terpstra, HKV Consultant,

The Netherlands; [Link]@[Link] and [Link]@[Link].
lands, which is on the verge of shifting from a primar-
ily prevention-based approach toward a broader risk
management approach that includes citizens disas-
ter preparedness.
(4)
Unfortunately, however, only a
few international studies have addressed the issue of
citizens ood preparedness behavior.
(5)
This article focuses exclusively on Dutch citi-
zens ood preparedness intentions. Without the ex-
isting river and sea ood defenses, two-thirds of
the Netherlands would be permanently ooded. As
these areas are inhabited by 60% of the population
(about 9 million people), many citizens depend heav-
ily on the ood defenses for their safety. Due to its
high ood protection standards, the Netherlands is
1658 0272-4332/11/0100-1658$22.00/1
C
2011 Society for Risk Analysis
Affective and Cognitive Routes to Flood Preparedness Behavior 1659
currently also one of the best protected deltas in the
world.
(6)
Since the last severe ood occurred over
55 years ago (in 1953, claiming 1,836 casualties), most
citizens have no personal recollection of ooding. It
is often assumed that the high levels of public ood
protection and the lack of ood experiences have
lowered citizens ood risk perceptions, which in turn
keeps them from preparing for potential ood dis-
asters.
(7)
Thus far, however, these assumptions have
never been tested. This article therefore investigates
citizens ood preparedness intentions and focuses
on three underlying determinants: citizens ood risk
perceptions, their trust in public ood protection, and
their previous ood hazard experiences.
2. THEORY AND EXPECTATIONS
2.1. Effects of Hazard Experiences
The personal experience of a disaster makes
people uniquely aware of their vulnerability to a
disasters consequences. Disaster experience is there-
fore often assumed to increase peoples risk percep-
tions.
(8)
For instance, both Grothmann and Reuss-
wig
(5)
and Siegrist and Gutscher
(9)
came to this
conclusion in regards to ood hazards. However,
studies in other hazard domains have shown oppo-
site effects.
(10)
To resolve these inconsistencies, Lin-
dell and Perry
(11)
suggested that the effect of hazard
experience depends on how people interpret their
experiences or what they have learned from them.
Indications of how people interpret their ood ex-
periences were recently presented by Siegrist and
Gutscher.
(12)
These authors found that roughly 20
35% of the ood victims interviewed mentioned feel-
ings of uncertainty and insecurity, fear and shock,
and helplessness as the worst outcomes of their ood
experience, while among nonvictims, hardly any re-
spondents anticipated these emotions as the worst
outcomes of future ooding. According to Siegrist
and Gutscher,
(12)
these negative emotions were a key
factor in explaining why ood victims had taken sub-
stantially more precautionary actions against future
oods than nonvictims. These results are generally
consistent with the ndings of two recent experimen-
tal studies. Keller, Siegrist, and Gutscher
(13)
aimed to
increase negative affect by presenting subjects with
pictures of ooded houses. Although no manipula-
tion checks were performed, these authors suggested
that the presentation of emotion-laden images con-
tributed to a higher level of perceived ood risk in
the subjects. Vastfjall, Peters, and Slovic
(14)
manip-
ulated negative affect associated with the 2004 East
Asia Tsunami. By reminding their Swedish subjects
of the tsunami, they indeed elicited negative affect,
which in turn resulted in more pessimistic expecta-
tions about the future. Interestingly, none of these
subjects had been directly involved in the disaster,
but the Swedish media covered the tsunami exten-
sively, since 600 Swedish tourists were killed or lost.
The notion that negative emotions increase risk
perceptions (and potentially adaptive behavior) is
consistent with research on affect and decision mak-
ing
(15)
(the affect heuristic) and is coined the
risk-as-feelings hypothesis.
(16)
In other words, peo-
ple may interpret their hazard experiences differ-
ently, depending on whether these experiences evoke
negative emotionsor as Slovic, Finucane, Peters,
and MacGregor
(17)
described negative affect, (sub-
tle) feelings of badness (p. 312). If negative af-
fect is experienced when making risk judgments, the
perceived risk should increase. Moreover, accord-
ing to the affect heuristic, positive affect (feelings of
goodness) should decrease perceived risk. For in-
stance, experiments by Finucane, Alhakami, Slovic,
and Johnson
(18)
showed that an increased negative
affect associated with nuclear technology resulted in
higher risk judgments, while positive affect had the
opposite effect.
Unfortunately, few studies have investigated
positive emotions associated with disasters. Rather,
studies have focused on negative emotions and ad-
verse health impacts. These studies have found that
the intensity of negative emotions increases with
the severity of personal disaster consequences (e.g.,
property losses, injuries, stresses in social relations).
Interestingly, such negative emotions may develop
into more severe health problems (e.g., posttrau-
matic stress disorder), particularly when people lack
the economic, social, and psychological resources
needed to cope with these experiences.
(19)
Moreover,
in the eld of trauma research, increasingly more
studies have shown that stronger coping abilities and
resilience against health impairments are related to
the experience of positive emotions.
(20,21)
According
to the broaden-and-build theory,
(22)
positive emo-
tions contribute to the ability to cope with stress
and negative life experiences because they stimulate
thought and increase the number of perceived cop-
ing behaviors, thereby adding to ones physical, in-
tellectual, social, and psychological resources.
(23)
Re-
garding disasters, Fredrickson, Tugade, Waugh, and
Larkin
(24)
supported the link between positive emo-
tions (e.g., gratitude, interest) and resilience (e.g.,
1660 Terpstra
life satisfaction, optimism) in the context of the
9/11 terrorist attacks. Moreover, increases in cop-
ing resources may be enduring, which makes people
more resilient when dealing with future events.
(25)
Vazquez, Cervellon, Perez-Sales, Vidales, and Ga-
borit
(26)
investigated positive emotions among earth-
quake survivors in refugee camps in El Salvador. In
addition to negative emotions, almost 75% of the in-
terviewees recalled moments of happiness that could
be attributed to either being alive or feeling ac-
companied. Community and social activities largely
contributed to positive emotions and the ability to
cope with the difcult circumstances.
(27)
Moreover,
the majority of survivors indicated that they had
gained something positive (e.g., regarding social rela-
tions and personal skills) and felt better able to cope
with future events. Similar ndings were reported
by Tang,
(28)
who investigated posttraumatic growth
among Thai survivors of the 2004 East Asia Tsunami.
Regression analyses showed that the seeking and/or
receiving of social support were the most important
predictor of positive adaptation (perceived positive
changes in, for instance, relations with others and ap-
preciation of life).
In summary, the affect heuristic predicts that
positive and negative emotions attached to natural
hazard experiences should inuence risk perceptions
and, possibly, preparedness behavior. Indicators of
the effects of negative emotions have recently been
shown by Siegrist and Gutscher.
(12)
Positive emo-
tions have been investigated in the context of trauma
(e.g., abuse, violence) and crisis (e.g., 9/11 attacks)
in relation to peoples coping abilities and resilience
in dealing with future events. However, at the time
of this research we are unaware of any studies that
investigated the effect of positive emotions on the
perceived risk of natural hazards and subsequent pre-
paredness behavior. Moreover, in addition to experi-
encing positive and negative emotions after disasters,
people may also be left emotionally unimpressed.
This article will explore whether it is useful to dis-
tinguish between people who experience positive af-
fect, those who experience negative affect, and those
who are left emotionally unimpressed when evalu-
ating the effects of ood hazard experiences in the
Netherlands.
2.2. Trust in Public Flood Protection
Despite the high ood protection standards in
the Netherlands, no ood protection system is per-
fect. Relatively recently, in 1993 and 1995, high river
discharges unexpectedly threatened several upriver
communities that nearly escaped ood disaster. Typ-
ically, citizens lack the expert knowledge to judge
the uncertainties that cause such unexpected events.
Only when citizens trust the risk experts will they
be able to tolerate these uncertainties and live rel-
atively unconcerned behind the ood defenses.
(29)
Trust serves to reduce the complexity of a situa-
tion.
(30)
These authors argued that, especially when
people lack knowledge about a hazard, their risk
judgments are based on the degree to which they
trust the responsible risk managers. That is, when
trust is high, risk perception is low, and vice versa.
(31)
Trust shares conceptual similarity with affect.
Affective responses are generated quickly and au-
tomatically and are experienced as a feeling state,
dening whether something is perceived as good or
bad.
(15)
Thus, both trust and affect reduce the com-
plexity of risk judgments because neither requires a
consideration of all of the pros and cons related to
the risks. Poortinga and Pidgeon
(32)
suggested that
trust and affect share similarities because they re-
ect more general attitudes toward risk, which in
turn drive more specic risk judgments. Although
we expect that trust and affect both predict more
specic attitudes toward ood preparedness, we also
foresee that more cognitive evaluations of ood risk
play a role. First, this would be consistent with re-
search showing that many decisions are informed by
the integrated effect of affective and cognitive pro-
cesses.
(17)
Second and less obvious, however, com-
pared to other types of risks, ooding risk may be
a special case. For most technological risks (e.g., a
nuclear power plant, genetically modied food), one
cannot actually see from the outside what constitutes
the risk and what has been done precisely to reduce
the likelihood of a mishap. Therefore, positive feel-
ings and trust in risk managers are highly important
because these are virtually the only means through
which people accept their own lack of knowledge
about a risk. However, ood probability is reduced
by the construction of ood defenses. These ood de-
fenses (e.g., a dike or dam) are grossly visible in the
landscape and thus provide some information about
the quality of risk management. In the Netherlands,
people can physically walk on dikes and dams, and
characteristics such as their magnitude, height, and
state provide a yardstick for estimating the provided
protection level. Thus, although lay people lack the
expertise needed to calculate the actual protection
level provided by ood defenses, they may derive
ood likelihood based on the amount of trust that is
Affective and Cognitive Routes to Flood Preparedness Behavior 1661
inspired by their observations. Moreover, public and
media communications often boast about the qual-
ity of the Dutch ood defenses. Therefore, trust may
not only relate to affective responses to risk (e.g.,
dread), but trust may also predict how people as-
sess ood likelihood (a more cognitive evaluation
of ooding risk). Grothmann and Reusswig
(5)
con-
rmed this relation among citizens from the German
city of Cologne, which is located about 175 km up-
stream from the Dutch border along the river Rhine.
They found that those citizens with more faith in pub-
lic ood protection revealed lower ood risk percep-
tions (a conjoint measure of perceived ood proba-
bility and severity) and had taken less precautionary
measures. We too expect to nd that perceptions of
ood risk (especially ood likelihood) are decreased
by a high level of trust in the ood defenses, which in
turn keeps citizens from preparing for potential ood
disasters.
2.3. Aims and Expectations
This article investigates the predictors of ood
preparedness among Dutch citizens. We focus specif-
ically on how affect associated with previous ood
hazard experiences, trust in public ood protection,
and cognitive and affective evaluations of ood risk
shapes ood preparedness intentions. Fig. 1 presents
the expected relations among these variables.
First, the model reects our view that higher risk
perceptions increase citizens ood preparedness in-
tentions (H1a, H1b, and H1c). Second, we expect
that higher levels of trust have a direct negative im-
pact on citizens risk perceptions (H2a, H2b, and
H2c) and their ood preparedness intentions (H2d).
Finally, we predict that citizens affective interpre-
tations of their ood hazard experiences directly
inuence their ood risk perceptions (H3a, H3b, and
H3c), their trust in ood protection (H3d), and their
ood preparedness intentions (H3e). Specically, we
predict that positive affect increases trust and de-
creases perceived risk and preparedness intentions
due to an optimistic bias. Negative affect results in
the opposite effect. In addition to these direct effects,
we will also test for mediated effects. Our model pro-
poses two mediating variablestrust in ood pro-
tection and ood risk perceptions. Although direct
empirical evidence is scarce, Grothmann and Reuss-
wig
(5)
provided indirect evidence by investigating
citizens ood damage mitigation behavior through
hierarchical regression analyses. Flood experience
contributed most to explaining the variance in pre-
cautionary behavior since it was entered rst in the
regression analyses. In subsequent separate steps,
risk perception and trust added little unique explana-
tory power (36%). However, because these vari-
ables did show signicant, mediocre correlations with
precautionary behavior (0.210.39), the almost com-
plete absence of unique effects of trust and perceived
risk on precautionary behavior suggests partial and
possibly full mediation. We will therefore test for
the potential mediation effects of trust and perceived
risk. The model displayed in Fig. 1 will be tested us-
ing structural equation modeling (SEM). SEM is cur-
rently regarded as the most sophisticated method for
evaluating causal relations among multiple depen-
dent variables based on cross-sectional data, due to
its incorporation of both latent and observed vari-
ables and the inclusion of measurement errors.
(3335)
In addition, we will validate a three-way classi-
cation of emotions (negative emotions, neutral emo-
tions, and positive emotions) by comparing these
three groups to citizens who lack any ood hazard
experiences. In line with the affect heuristic, citizens
who are left emotionally unimpressed (neutral emo-
tions) should reveal the same levels of perceived risk,
Preparedness
intention
Trust in flood
protection
Perceived
likelihood
Perceived
dread
Perceived
consequences
H3d
H3a
H3b
H3c
H2b
H2a
H2c
H2d
H3e
H1a
H1c
H1b
Feelings associated
with previous
experience
Fig. 1. Path model of predicted causal
effects.
1662 Terpstra
trust, and preparedness intentions as inexperienced
citizens. Citizens with negative emotions should re-
veal higher risk perceptions, less trust, and lower pre-
paredness intentions, as compared to the inexperi-
enced respondents, while the opposite is expected for
citizens with positive emotions.
2.4. The Present Studies
Predictions were tested in two studies. Study 1
was conducted in a coastal community that recently
experienced a heavy storm. Study 2 was performed
to replicate causal relations in two different ood risk
contextsin the impact area of the 1953 ood disas-
ter at the Dutch coast and in the river area that came
close to experiencing oods in 1993 and 1995. More-
over, Study 2 aimed to gain further insight into which
emotions are evoked by previous ood hazard expe-
riences.
3. STUDY 1: THE EFFECTS OF STORM
EXPERIENCES
3.1. Method
3.1.1. Sample and Procedures
Data were collected in the northern end of the
Netherlands at the Wadden Sea coast. Early Novem-
ber 2006, about 1 month before data collection, a
heavy storm (force 10 on the Beaufort-scale) caused
storm water levels near the study area that, according
to statistics, occur 15 times every 1,000 years
1
(1.5%
probability per year). Despite the relatively high wa-
ter levels, there was no imminent risk of ooding be-
cause the existing dike protects against water levels
reached during storms that occur once in every 4,000
years (0.025% probability per year). Still, this event
gained much (inter)national media attention, primar-
ily because a hundred horses in an unprotected ood
plain needed rescue.
A random sample of 7,000 household addresses,
gathered from a telephone book, received a let-
ter explaining our research and inviting them to
participate in our Internet survey.
2
The letter con-
tained the Internet address and a password for tak-
ing the questionnaire. Data were collected between
1
From the website of the Royal Dutch Meteorological Institute,
[Link], accessed on November 4, 2006.
2
In 2008, 86% of Dutch households were connected to the Inter-
net, which is among the highest rates of Internet penetration in
Europe (Statistics Netherlands, 2008).
December 8, 2006 and January 10, 2007. Respon-
dents were thanked for participation in the invita-
tion letter, on the website, and again at the end of
the survey. In total, 826 citizens participated in the
survey (response rate 11.8%). For the purpose of
this study, we selected a subsample of 244 respon-
dents who recalled a personal experience of storm
events in the study area (but who reported no other
ood hazard experiences). In addition, we also se-
lected respondents who failed to recall any ood
hazard experiences (n = 228). This latter, inexperi-
enced group serves exclusively as a reference group
for comparison with the experienced respondents. Of
the 472 respondents (244 with experience of storm
events, and 228 without ood hazard experiences),
338 (72%) respondents were males, and the mean
age was 50.4 years (SD = 12.8).
3.1.2. Measurements
Emotions related to storm events were measured
by asking those who reported personal experiences
of stormevents in the study area: Could you indicate
the type of feelings you experience now, when re-
calling what you experienced at that time? We pro-
vided ve choices: very negative feelings, rather
negative feelings, neither negative nor positive
feelings, rather positive feelings, and very pos-
itive feelings. All other questionnaire items were
labeled using the prexes not at all, hardly,
somewhat, quite, and very (ve-point Likert-
type scales). In addition, the questionnaire mea-
sured trust in ood protection (four items, Cronbachs
alpha = 0.90), risk perceptionsby three constructs,
perceived dread (three items, Cronbachs alpha =
0.94), perceived ood likelihood (one item), and per-
ceived ood consequences (three items, Cronbachs
alpha = 0.79)and ood preparedness intentions
(three items, Cronbachs alpha = 0.86). Table I
presents all item contents.
3.1.3. Analyses
The proposed path model of Fig. 1 was evalu-
ated using SEM in AMOS 17 (SPSS Inc., Chicago,
IL, USA), including both latent and observed (i.e.,
indicator) variables. Prior to analysis, the data were
checked for normality. The subsequent analysis was
performed in two steps (for another recent ap-
plication of this procedure see, for instance, Pe-
ters
(36)
). First, to ensure that the questionnaire items
measured their intended constructs correctly, the
Affective and Cognitive Routes to Flood Preparedness Behavior 1663
Table I. Latent Variables, Items, and Factor Loadings in Study 1
Latent Variable Item Loading
Emotions related to storm
events
Could you indicate the type of feelings you experience now, when recalling what you
experienced at that time?

Trust in ood protection I am condent that the ood defenses along Wadden Sea coast are maintained well. 0.81
I have condence in the technological skills of ood risk managers. 0.88
I have condence in the strength of the ood defenses along Wadden Sea coast. 0.89
I am condent that there are sufcient, properly qualied people working with the water
management authorities.
0.74
Perceived dread Could you indicate how you feel now, at this moment, when you think of the risk of ooding in
your area?
I feel frightened. 0.90
I feel worried. 0.83
I feel restless. 0.97
Perceived ood likelihood How likely do you nd major ooding in your area within the next 10 years?
Perceived ood How severely do you feel a oods consequences will affect you personally? 0.61
consequences In case of ooding, how likely are you to become a victim (damage to possessions/wounded)? 0.87
How likely does it seem to you that a ood will have fatal consequences for you personally? 0.69
Flood preparedness To what extent are you interested in information about ood preparedness? 0.76
intentions To what extent do you intend to search for information about ood preparedness? 0.84
To what extent do you intend to prepare for ooding? 0.80
measurement model was validated by performing
rst-order conrmatory factor analysis (CFA). Based
on goodness-of-t indexes, factor loadings, and mod-
ication indexes it was decided whether the model
needed adjustment.
(33)
Second, if the measurement
model was satisfactory, AMOS was subsequently
used to estimate the regression paths of Fig. 1, simul-
taneously.
Mediation effects were evaluated following
Brown.
(37)
Imagine a model predicting Y1 from
X1 and the potential mediator Y2. Mediation then
requires compliance with three conditions. There
should be a signicant direct effect of X1 on the me-
diating variable Y2 (condition 1) and a signicant di-
rect effect of the mediator Y2 on Y1 (condition 2).
Third, when controlling for the mediating variable
Y2, the signicant direct effect of X1 on Y1 should
become less signicant (partial mediation) or disap-
pear (full mediation). In AMOS, condition 3 can be
veried by assessing the total effect that is composed
of a direct and indirect effect. In case the total effect
of X1 on Y1 is signicant, but the direct effect be-
comes less signicant or drops below signicance due
to a signicant indirect effect, then there is evidence
of partial or full mediation, respectively.
A problem often encountered in performing
SEM is a lack of consensus on the criteria for accept-
ing or rejecting a model. Based on state-of-the-art
knowledge, Kline
(34)
recommends reporting at least
the (1) model chi-square, (2) the Steiger-Lind root
mean square error of approximation (RMSEA) with
its 90% condence interval (90% CI), (3) the Bentler
comparative index (CFI), and (4) the standardized
root mean square residual (SRMR). The model chi-
square is often divided by the model degrees of free-
dom (
2
/df), which is less sensitive to uctuations in
sample size. This test statistic should be below 5 and
preferably below 2 in order to accept the model. A
CFI value (which varies from 0 to 1) above 0.90 and
preferably above 0.95 indicates a model t. RMSEA
values should be as low as 0.08 and preferably 0.06 or
lower in order to accept the model. Finally, SRMR
values below 0.10 are considered as favorable.
The classication of respondents in three emo-
tions groupsnegative, nonspecic, and positive
emotionswas veried by comparing their scale
means to those of the inexperienced respondents.
To this end, we performed a multivariate analysis of
variance and post hoc tests, using SPSS 15 (SPSS Inc.,
Chicago, IL, USA).
3.2. Results
3.2.1. Analysis of the Proposed Model (Respondents
with Flood Experience)
Measurement model. Because emotions asso-
ciated with previous storm weather experiences
could only be measured for respondents who indi-
cated such experiences, the subsequent analysis only
1664 Terpstra
Table II. Model Fit Indexes Study 1
Model Analysis Items/Constructs n
2
df
2
/df RMSEA (90% CI) CFI SRMR
1 Measurement model
(rst-order CFA)
13/4 244 102.02 59 1.73 0.05 (0.030.07) 0.98 0.05
2 Structural model (path
analysis)
13/4 + 2 single indicator
variables
244 134.35 77 1.74 0.06 (0.040.07) 0.97 0.04
applies to those respondents (n = 244). Using a rst-
order CFA, the measurement model estimated the
extent to which the observed items loaded onto their
respective latent variables. To this end, all latent con-
structs but no observed error variances were allowed
to covary with one another. Constructs were scaled
by xing one factor loading to one of each of the
four latent variables, which ensured the identication
of the model. All other factor loadings were left un-
constrained. The CFAincluded four latent constructs
and 13 measured variables with accompanying er-
ror termstrust (four items), dread (three items),
perceived consequences (three items), and prepared-
ness intention (three items). Note that emotions at-
tached to previous experiences and perceived ood
likelihood were omitted from this validation proce-
dure, since these variables were measured by one
item only. Results indicated satisfactory factor load-
ings on all constructs (see Table I) and a good t (Ta-
ble II, Model 1), indicating that questionnaire items
loaded on the intended latent constructs while cross-
loadings were absent.
Structural model. Next, we evaluated the pro-
posed path model of Fig. 1. In addition to the
four latent variablestrust (four items), dread (three
items), perceived consequences (three items), and
preparedness intention (three items)the model in-
cluded emotions attached to previous experiences
and perceived ood likelihood. The causal structure
was specied by drawing the regression paths iden-
tical to Fig. 1. The model revealed a good t (see
Table II, Model 2). The results of this procedure are
depicted in Fig. 2.
First, the analysis supported the predicted effects
of perceived dread (H1a; = 0.20, p < 0.05) and
perceived likelihood (H1b; = 0.46, p < 0.001) on
preparedness intention, but rejected the predicted ef-
fect of perceived consequences on preparedness in-
tention (H1c; = 0.09, ns).
Second, the model supported that trust was
a signicant predictor of perceived dread (H2a;
= 0.29, p < 0.001) and perceived ood likeli-
hood (H2b; = 0.46, p < 0.001)but the model
rejected that trust predicted the perceived conse-
quences (H2c; =0.12, ns). In addition, the AMOS
output indicated that the total effect of trust on pre-
paredness intention was signicant ( = 0.18, p <
0.05), but there was no support for a direct effect of
trust on preparedness intention (H2d; = 0.08, ns).
Rather, the indirect effect was signicant ( =0.26,
Preparedness
intention
Trust in flood
protection
Perceived
likelihood
Perceived
dread
.31***
-.24***
-.16**
-.22**
-.46***
-.29***
-.12
.08
.03
.20*
-.09
.46***
Feelings associated
with previous
experience
Perceived
consequences
(.09)
(.18)
(.08)
(.28)
(.26)
Notes: 1. Oval variables reect latent variables; rectangular variables reect single indica-
tor (i.e., observed) variables. Reported numbers are standardized regression coefcients ()
indicating direct effects. Indirect (mediated) effects are explained in the results section.
2. Explained variances (R
2
) are provided in parentheses. 3. n = 244. 4.

p < 0.001,

p < 0.01,

p < 0.05.
Fig. 2. Path modeling Study 1 results.
Affective and Cognitive Routes to Flood Preparedness Behavior 1665
p < 0.001)that is, the signicant effects of per-
ceived dread and perceived likelihood on prepared-
ness intention together with the signicant effects of
trust on perceived dread and perceived likelihood
supported that perceived dread and perceived like-
lihood fully mediated the effect of trust on prepared-
ness intentionwith perceived likelihood accounting
for about 80% of the mediation effect.
Third, the results supported the predicted di-
rect effects of emotions attached to previous
storm weather experiences on the perceived risk
variablesmore positive (and less negative) emo-
tions reduced perceived dread (H3a; = 0.24,
p < 0.001), perceived likelihood (H3b; = 0.16,
p < 0.01), and perceived consequences (H3c; =
0.22, p < 0.001). Also, more positive emotions in-
creased trust in ood protection (H3d; = 0.31,
p < 0.001). Moreover, the AMOS output also in-
dicated a signicant indirect effect of emotions on
perceived likelihood ( = 0.14, p < 0.05), which
supported that the total effect of emotions on per-
ceived likelihood was partly mediated by trust in
ood protection. Finally, the total effect of emo-
tions on preparedness intention consisted of a non-
signicant direct effect, which rejects H3e ( = 0.03,
ns), and a signicant indirect effect ( = 0.15,
p < 0.05). There were four mediation paths from
emotions to preparedness intentionmediation by
the combinations trust/perceived likelihood (about
30%), trust/perceived dread (10%), and mediation
by perceived dread (about 35%) and by perceived
likelihood (20%). The remaining 5% of the total ef-
fect was mediated by nonsignicant paths.
3.2.2. Comparison to the Inexperienced Respondents
Nonspecic (neutral) emotions and negative and
positive emotions were reported by 137 (56%),
67 (27%), and 40 (16%) respondents, respectively.
These groups did not differ in terms of gender distri-
bution (Kruskal-Wallis test,
2
(3)
=7.37, ns) or mean
age (F
3,448
= 1.43, ns). To validate the emotion clas-
sication of nonspecic (neutral), negative, and pos-
itive emotions, we performed a multivariate analy-
ses of variance, using the group without ood hazard
experiences as a reference group (n = 228). Indeed,
the MANOVA was signicant (F
12,1401
= 5.45

) and
Table III indicates that there were signicant differ-
ences among groups.
Comparing inexperienced respondents to the re-
spondent groups with negative, nonspecic, and posi-
tive emotions on each of ve variables required eval-
uating 15 (three times ve) hypotheses. Table III
indicates that 11 of them (73%) were supported by
the data. Specically, in line with the affect heuris-
tic, respondents who indicated neither negative nor
positive emotions (i.e., nonspecic emotions) should
reveal the same levels of trust, perceived risk, and
preparedness intentions as inexperienced respon-
dents. This prediction was conrmed on all but the
perceived consequences variable. Second, respon-
dents with negative emotions should reveal less trust,
higher risk perceptions, and higher preparedness in-
tentions, as compared to the inexperienced respon-
dents. This prediction was conrmed for all variables,
but on the trust variable the difference failed to reach
signicance. Finally, respondents with positive emo-
tions should reveal higher trust, lower risk percep-
tions, and lower preparedness intentions, as com-
pared to the inexperienced respondents. This was
conrmed regarding trust and dread measurements,
but not on the variables measuring perceived conse-
quences and preparedness intentions.
3.3. Discussion of Study 1
Respondents with heavy storm experience not
only reported negative feelings (27%) but also non-
specic (neutral, 56%) and positive (16%) feelings.
A comparison of these groups to respondents with-
out any ood hazard experiences largely supported
this three-way classication.
Table III. Analysis of Study 1 Scale
Means (SD) using Manova and
Post Hoc Tests
No Experience Nonspecic Negative Positive
(Reference Group, Feelings Feelings Feelings
Coastal Area n = 228) (n = 137) (n = 67) (n = 40)
Trust in ood protection 3.78 (0.71) 3.78 (0.60) 3.65 (0.62) 4.21

(0.61)
Perceived dread 2.14 (0.98) 2.21 (0.90) 2.66

(0.99) 1.70

(0.87)
Perceived likelihood 2.16 (1.01) 2.28 (0.95) 2.61

(0.97) 1.70

(0.79)
Perceived consequences 2.91 (1.03) 3.28

(0.89) 3.69

(0.88) 2.93 (1.01)

Preparedness intention 2.74 (0.92) 2.86 (0.88) 3.14

(0.92) 2.71 (1.07)

Asterisks indicate signicant deviations from the No experience group;

p < 0.01,

p <
0.05.
1666 Terpstra
Path analysis showed, consistent with our pre-
dictions based on the affect heuristic,
(15)
that nega-
tive feelings decreased trust in ood protection, and
increased risk perceptions. Positive feelings had the
opposite effect. Additionally, the effects of emotions
on preparedness intention were indirectmediation
followed four paths: (1) through perceived dread, (2)
through perceived likelihood, (3) via trust through
perceived dread, and (4) via trust through perceived
likelihood. Thus, trust in ood protection played a
central role because it had both indirect effects on
preparedness intention and partly mediated the ef-
fects of emotions. Remarkably, perceptions of ood
consequences failed to predict ood preparedness in-
tentions.
A major shortcoming of this study lies in the fact
that it is unclear which feelings respondents meant
when they reported negative and positive feelings. In
addition, one could argue that the results of affect
were found because the data were collected only 1
month after a storm. Finally, the response rate was
lower than desired. The next study aims to overcome
these shortcomings.
4. STUDY 2: THE EFFECTS OF (NEAR)
FLOOD EXPERIENCES
4.1. Method
4.1.1. Study Areas and Samples
Study 2 included two samples. First, a sample was
drawn from the impact area of the 1953 ood disas-
ter at the North Sea coast in the southwestern part
of the Netherlands (coastal community). The second
sample was drawn from two communities in the river
area (river area communities) that experienced near
oods in 1993 and 1995 (i.e., extremely high river dis-
charges but no dikes were breached). The procedures
for data collection were identical to those in Study
1. However, in order to increase the response rate,
reminders were sent 3 and 5 weeks after the invita-
tion letter. All invitations were sent out on April 1,
2008, and the Internet survey was closed on May 31,
2008. We selected random samples of 5,000 house-
hold addresses from the coastal community and in
total 11,000 household addresses from the two river
area communities. Despite the two reminders, re-
sponse rates were again lower than desired; between
9.6 and 12.9%. This issue will be addressed in the -
nal discussion.
In the coastal community, we selected a subsam-
ple of 428 respondentsincluding 169 respondents
with personal experience of the 1953 ood disaster
and 259 respondents without any ood hazard ex-
periences. Overall, 69% were males, and the mean
age was 56.1 years (SD= 13.3; 4 persons failed to re-
port their age). Because both river area samples were
similar with respect to gender (Mann-Whitney U,
Z = 0.80, ns) and mean age (t
850
= 0.61, ns), we
treated them as one homogeneous group. This was
done to increase the reliability and statistical power
of the path model. The selected subsample consisted
of 861 respondents658 respondents personally ex-
perienced the 1993/1995 high river discharges, while
203 lacked any ood hazard experiences. Overall,
69% were males, and the mean age was 52.5 years
(SD = 13.0; 9 persons failed to report their age).
4.1.2. Measurements
Emotions related to ood hazard experiences.
Respondents indicated the type of feelings they
sensed when thinking about their personal experi-
ences. Response labels were exclusively negative
feelings, more negative than positive feelings, no
(dominance of) positive or negative feelings, more
positive than negative feelings, and exclusively
positive feelings. As Study 1 failed to investigate
which particular feelings were experienced on recall,
respondents were invited to provide verbal descrip-
tions of their negative and positive feelings. Addition-
ally, as in Study 1, the questionnaire measured trust
in ood protection (three items), risk perceptions
by three constructs, perceived dread (two items with
a negative connotation and two items with a positive
connotation), perceived ood likelihood (one item),
perceived ood consequences (four items)and ood
preparedness intentions (four items). Response labels
were tailored to the item contents. Items that were
intended to form a multi-item scale were presented
in a randomized order. Cronbachs alpha was larger
than 0.80 on all scales. Table IV presents all item
contents.
4.1.3. Analysis
Identical to Study 1.
4.2. Results
4.2.1. Analysis of the Proposed Model (Respondents
with Flood Experience)
Measurement model. As in Study 1, we applied
rst-order CFA to estimate the extent to which the
observed items loaded onto their respective latent
Affective and Cognitive Routes to Flood Preparedness Behavior 1667
Table IV. Latent Variables, Items, and Factor Loadings in Study 2
Coastal River
Latent Variable Item Area Area
Emotions related to prior
experiences
Could you indicate the type of feelings you experience now, when recalling what
you experienced at that time?

Trust in ood protection How condent are you . . .
. . . the strength and height of the ood defenses in [area] is based on a thorough and
sound risk analysis?
0.88 0.87
. . . that the ood defenses [area] are maintained properly? 0.81 0.73
. . . that the water management authorities in your area have sufcient knowledge
about ood protection?
0.82 0.75
Perceived dread Could you indicate how you feel now, at this moment, when you think of the risk of
ooding in your area?
I feel frightened. 0.94 0.88
I feel restless. 0.86 0.92
I feel relaxed.
I feel at ease.
Perceived ood likelihood How likely do you nd major ooding in your area within the next 10 years?
Perceived ood consequences Imagine there will be a ood in your area. How likely do you nd the following?
Substantial damage to public facilities (roads, parks, etc.) in your city. 0.92 0.84
Substantial damage to your house or possessions. 0.89 0.84
You and/or your family will face a life-threatening situation. 0.76
Your daily life (job and other daily routines) will be disrupted for a long time. 0.82 0.80
Flood preparedness intentions To what extent do you intend to do the following in the near future?
Assemble an emergency kit (including water, food, a battery powered radio, a rst
aid kit, etc.).
0.78 0.71
Collect information about ood consequences, evacuation routes, and safe/high
locations.
0.86 0.83
Making a to-do list that is helpful in case of an evacuation or ood (household plan). 0.94 0.94
Making agreements with family, friends, and neighbors on how to help each other
in case of evacuation/ooding.
0.84 0.83
Table V. Model Fit Indexes Study 2
Analysis Items/Constructs n
2
df
2
/df RMSEA (90% CI) CFI SRMR
Coastal Area
Models
1 Measurement model
(rst-order CFA)
15/4 169 213.14 84 2.54 0.10 (0.08 0.11) 0.92 0.06
2 Measurement model
(rst-order CFA)
13/4 169 90.18 59 1.53 0.06 (0.030.08) 0.98 0.05
3 Structural model (path
analysis)
13/4+ 2 single indicator
variables
169 114.71 77 1.49 0.05 (0.030.07) 0.98 0.05
River Area
Models
1 Measurement model
(rst-order CFA)
15/4 658 641.49 84 7.64 0.10 (0.090.11) 0.90 0.05
2 Measurement model
(rst-order CFA)
12/4 658 66.23 48 1.38 0.02 (0.000.04) 1.00 0.02
3 Structural model (path
analysis)
12/4+ 2 single indicator
variables
658 85.91 64 1.34 0.02 (0.010.03) 0.99 0.02
variables. Thus, all latent variables but no observed
error variances were allowed to covary with one an-
other. Both in the coastal community (1953 impact
area, n = 169) and the river area community (n =
658) the initial model consisted of four latent, un-
observed variables and 15 observed variables with
accompanying error termstrust in ood risk man-
agement (three items), dread (four items), perceived
consequences (four items), and preparedness inten-
tion (four items). Table V reveals a moderate t
1668 Terpstra
-.00
-.12*
Preparedness
intention
Trust in flood
protection
Perceived
likelihood
Perceived
consequences
.10
.14***
-.35***
-.14***
-.10
-.15***
-.20*
.00
-.41***
-.25***
.16
.01
.10
.01
.31**
.15***
.09
.20***
.32***
.19***
Feelings associated
with previous
experience
(.26)
(.15)
(.35)
(.06)
(.01)
(.02)
(.19)
(.09)
(.04)
(.01)
-.44***
-.19***
Perceived
dread
Notes: 1. Upper numbers relate to the coastal area (1953 ood disaster, n = 169) and lower numbers to the river area (1993/1995 high
river discharges, n = 658). 2. Oval variables reect latent variables; rectangular variables reect single indicator (i.e., observed) variables.
Reported numbers are standardized regression coefcients () indicating direct effects. Indirect (mediated) effects are explained in the
results section. 3. Explained variances (R
2
) are provided in parentheses. 4.

p < 0.001,

p < 0.01,

p < 0.05.
Fig. 3. Path modeling Study 2 results.
of the initial model in both areas for (Model 1).
Modication indexes (MIs) revealed that the er-
ror terms of the two positively formulated dread
items shared a substantial amount of variance. More-
over, as these two items also showed poor fac-
tor loadings, we removed them from the models.
In addition, the river area model could be further
improved by removing a perceived consequences in-
dicator that showed poor performance. Factor load-
ings of the nal models were satisfactory (see Table
IV), and goodness-of-t statistics were within the de-
sired ranges (see Table V).
Structural model coastal area (1953 ood disas-
ter). Fig. 3 presents the results of the path analysis.
The analysis supported the predicted effects of per-
ceived dread (H1a; =0.31, p < 0.01) and perceived
likelihood (H1b; = 0.32, p < 0.001) on prepared-
ness intention, but rejected the predicted effect of
perceived consequences (H1c; = 0.09, ns). In ad-
dition, the model supported that trust was a signi-
cant predictor of perceived dread (H2a; = 0.44,
p < 0.001) and perceived ood likelihood (H2b;
= 0.41, p < 0.001), but rejected the effect of trust
on the perceived consequences (H2c; = 0.00, ns).
The AMOS output indicated a counterintuitive but
nonsignicant direct effect of trust on preparedness
intention (H2d; = 0.16, ns). Rather, the indirect
effect was signicant ( = 0.27, p < 0.01). Per-
ceived dread and perceived likelihood fully mediated
the effect of trust on preparedness intention, with ei-
ther path accounting for about 50% of the mediation
effect.
In addition, results supported the predicted di-
rect effects of emotions on perceived dread (H3a;
= 0.35, p < 0.001) and perceived consequences
(H3c; = 0.20, p < 0.05), but the analysis rejected
the proposed direct effect of emotions on perceived
likelihood (H3b; = 0.10, ns). Moreover, the anal-
ysis also failed to support the effect of emotions on
trust in ood protection (H3d; = 0.10, ns). There-
fore, trust failed to function as a mediator between
emotions and perceived risk.
Finally, the total effect of emotions on prepared-
ness intention consisted of a nonsignicant direct ef-
fect, which rejects H3e ( = 0.10, ns), and a signi-
cant indirect effect ( = 0.17, p < 0.05). This latter
indirect effect was mediated by perceived dread only,
because the direct paths from emotions to perceived
dread (condition 1) and from perceived dread to pre-
paredness intention (condition 2) were both signif-
icant. Perceived dread accounted for about 65% of
the mediation effectthe remaining 35% was trans-
mitted through nonsignicant paths.
Structural model river area (1993/1995 high river
discharges). Fig. 3 also presents the results for the
path model in the river area. The analysis supported
the predicted effects of perceived dread (H1a; =
0.15, p < 0.001), perceived likelihood (H1b; =
0.19, p < 0.001), and perceived consequences (H1c;
= 0.20, p < 0.001) on preparedness intention. The
Affective and Cognitive Routes to Flood Preparedness Behavior 1669
results also supported that trust was a signicant pre-
dictor of perceived dread (H2a; = 0.19, p <
0.001), perceived likelihood (H2b; = 0.25, p <
0.001), and perceived consequences (H2c; =0.12,
p < 0.05). However, the results rejected a direct ef-
fect of trust on preparedness intention (H2d; =
0.01, ns), but the indirect was just signicant ( =
0.10, p <0.05). This indirect effect was mediated by
all three risk perception variablesperceived dread
(about 25%), perceived likelihood (about 50%), and
perceived consequences (about 25%). In addition,
the results supported the predicted direct effects of
emotions on perceived dread (H3a; = 0.14, p <
0.001) and perceived likelihood (H3b; = 0.15,
p < 0.001), but the effect on perceived consequences
(H3c; = 0.00, ns) was rejected. There was support
for a direct effect of emotions on trust in ood protec-
tion (H3d; =0.14, p <0.001), but there was no sup-
port for a direct effect of emotions on preparedness
intention (H3e; = 0.01, ns). Moreover, none of the
emotions effects were mediated by trust or perceived
risk because none of the indirect effects reached sig-
nicance.
4.2.2. Comparisons to the Inexperienced
Respondents
In the coastal area, nonspecic (neutral) feel-
ings and negative and positive feelings were reported
by 107 (63%), 47 (28%), and 15 (9%) respondents,
respectively. In the river area, the distribution was
similarnonspecic (neutral), negative, and positive
feelings were reported by 433 (65%), 149 (23%), and
76 (12%) respondents, respectively. As in Study 1,
we validated this classication through two separate
multivariate analyses of variance, using the groups
without ood hazard experiences as reference groups
(coastal area, n = 269; river area, n = 203). Because
in both areas the four groups (three emotion groups
plus the reference group) differed in gender distribu-
tion (Kruskal-Wallis test; coastal area,
2
(3)
= 22.77;
river area,
2
(3)
= 12.09; p < 0.01) and in mean age
(coastal area, F
3,420
= 74.80; river area, F
3,848
= 6.31;
p < 0.01), these variables were entered as covariates.
Table VI presents the results.
Both in the coastal and river area, signi-
cant multivariate effects were obtained for feelings
associated with previous ood hazard experience
(coastal area, F
15,1398
= 4.55, p < 0.01; river area,
F
15,2532
= 4.19, p < 0.01)that is, after controlling
for the effects of gender (coastal area, F
5,414
= 8.04,
p < 0.01; river area, F
5,842
= 13.84, p < 0.01) and age
(coastal area, F
5,414
= 1.89, ns; river area, F
5,842
=
5.94, p < 0.01). Overall, Table VI supports the three-
way classication, especially regarding the distinction
between negative and nonspecic feelings. Positive
emotions generally showed somewhat smaller effects
than negative emotions.
More specically, both the coastal and the river
area respondents with nonspecic (neutral) emotions
did not deviate in their scale means from the inex-
perienced respondents, except on the dread scale in
the river area. Second, all negative emotion effects
occurred in the expected directions. Due to a small
Table VI. Analysis of Study 2 Scale
Means (SD) using Manova and
Post Hoc Tests
No Experience Nonspecic Negative Positive
(Reference Group, Feelings Feelings Feelings
Coastal Area n = 259) (n = 107) (n = 47) (n = 15)
Trust in ood protection 3.72 (0.71) 3.76 (0.83) 3.49 (0.96) 3.67 (0.95)
Perceived dread 1.78 (1.00) 1.51 (0.78) 2.26

(1.25) 1.40 (0.51)

Perceived likelihood 1.98 (0.95) 1.87 (0.83) 2.28

(1.08) 2.00 (0.76)

Perceived consequences 3.82 (0.98) 3.74 (1.07) 4.05

(0.89) 3.38 (0.98)

Preparedness intention 2.60 (0.98) 2.55 (0.98) 2.90 (1.10) 2.68 (1.12)
No Experience Nonspecic Negative Positive
(Reference Group, Feelings Feelings Feelings
River Area n = 203) (n = 433) (n = 149) (n = 76)
Trust in ood protection 3.56 (0.81) 3.52 (0.73) 3.32

(0.90) 3.68 (0.86)

Perceived dread 1.93 (1.00) 1.70

(0.85) 2.15

(1.03) 1.61 (0.79)

Perceived likelihood 2.27 (0.97) 2.33 (1.00) 2.74

(1.10) 2.05 (0.95)

Perceived consequences 3.42 (1.02) 3.43 (1.06) 3.45 (1.11) 3.33 (1.11)
Preparedness intention 2.64 (0.92) 2.64 (0.92) 2.71 (0.92) 2.53 (0.95)
Asterisks indicate signicant deviations from the No experience group;

p < 0.01,

p <
0.05.
1670 Terpstra
group size in the coastal area (n = 47), however, the
lower level of trust in ood protection and the higher
preparedness intention rating were not signicant.
In the river area, the effects on the perceived con-
sequences and preparedness intentions scales were
marginal. Finally, positive emotions failed to show
any signicant effects. In the coastal area, this is
partly explained by the small group size (n = 15).
Still, clear deviations were observed in the expected
directions on the dread and perceived consequences
scales in the coastal area, and on the dread scale in
the river area.
4.3. Verbal Descriptions of Negative and
Positive Feelings
In addition to rating their feelings, respondents
were invited to provide verbal descriptions of their
feelings, with a maximum of three words for negative
and three words for positive feelings. A total of 292
respondents (35% of all respondents with ood haz-
ard experiences) provided verbal descriptions; 109
respondents lled out only negative verbal descrip-
tions, 82 provided only positive verbal descriptions,
and 102 respondents lled out both negative and pos-
itive verbal descriptions.
To validate our measurements, we veried
whether the respondents had lled out description
boxes that were potentially inconsistent with how
they rated their feelings. One respondent rated his
feelings as negative but inconsistently lled out a pos-
itive description box only (helpful). Among the
540 respondents who indicated no (dominance of)
positive or negative feelings, 30 respondents (6%)
inconsistently lled out either a negative or a posi-
tive verbal description box only. However, because
all of the words that were lled out in the negative
description box had a true negative connotation (e.g.,
fear) and because all of the words lled out in the
positive description boxes also had a true positive
connotation (e.g., solidarity), we kept these words
for further analysis.
All of the positive words (324) and negative
words (381) were classied into broader categories
(see Fig. 4). Classes were based on words that
were used most frequently. For instance, the word
fear was counted 67 times. Fear-related words
(e.g., dread, horried) were subsequently in-
cluded in this class. In addition, 45 negative words
and 50 positive words (13.5%) could not be classi-
ed because these words were expressed only once
or twice and did not properly t into the other classes
(e.g., waking up, pain, change, courage).
Regarding the negative feelings, fear and feel-
ings of powerlessness and helplessness were fre-
quently mentioned. Together, these two classes ac-
counted for 31% of all expressed negative feelings.
Ten percent of the negative words were focused on
the amount of damage, casualties, and difculties en-
countered during the recovery period, while another
10% reected poor risk and crisis management on
the part of the government and local water board
(e.g., poor maintenance of dikes, poor organiza-
tion of evacuation, mismanagement, warnings
given too late). In addition, worries and stresses
(8%), feelings of uncertainty (7%), and pity and sad-
ness (6%) were mentioned relatively frequently.
Among the expressions of positive feelings, sol-
idarity (togetherness and unity) was by far men-
tioned the most frequently (32%). The word sol-
idarity itself was mentioned 60 times (19%). The
care/aid/help dimension (21%) was closely related,
but contained words that were more directed to the
act of helping itself (e.g., helping evacuated friends
and family members, caring for others, helpful-
ness, etc.). Other recurring words reected feelings
of sensation and relief (12%) and accounts of feel-
ing impressed by the beauty and force of nature
(7%).
Thus, the vast majority of respondents who rated
their feelings as negative also described their feelings
with words that clearly had a negative connotation,
while the opposite was found for respondents who
rated their feelings as positive. These ndings sup-
port the validity of our measurements and provide
insight into the origin and meaning of negative and
positive feelings.
5. GENERAL DISCUSSION
Dutch ood risk management is on the verge of
shifting from a primarily prevention-based approach
toward a broader risk management approach that in-
cludes citizens disaster preparedness.
(4)
This article
aimed to predict citizens ood preparedness inten-
tions from their risk perceptions, their trust in ood
protection, and their previous ood hazard experi-
ences. Moreover, we aimed to expand empirical ev-
idence of the role of affect associated with previ-
ous experiences, by focusing on negative and positive
feelings evoked by these experiences.
Results showed that citizens experience not only
negative feelings when recalling their previous ood
hazard experiences but also positive feelings. How-
ever, positive feelings seem to be evoked less fre-
quently than negative feelingson average, in Study
Affective and Cognitive Routes to Flood Preparedness Behavior 1671
20 40 60 80 100 120
45 (11.8)
5 (1.3)
5 (1.3)
6 (1.6)
7 (1.8)
8 (2.1)
12 (3.1)
21 (5.5)
26 (6.8)
32 (8.4)
38 (10.0)
41 (10.8)
45 (11.8)
74 (19.4)
16 (4.2)
104 (32.1)
69 (21.3)
38 (11.7)
24 (7.4)
15 (4.6)
12 (3.7)
7 (2.2)
5 (1.5)
50 (15.4)
Positive feelings
(324 classified words)
Negative feelings
(381 classified words)
solidarity / togetherness / compassion
care / aid / help
relief / sensation
beauty / force of nature
safety / flood defenses
alert / watchful / prepared
cheerful
information / communication
unclassified
danger / unsafety
pity / sadness
uncertainty
worriedness / stresses
bad risk / crisis management
damage / casualties / recovery
powerless / helpless
fear
unclassified
ignorant
care / aid / help
anger
chaos
astonished / bewildered
(power of) nature
120 100 80 60 40 20
Fig. 4. Classication of positive and
negative feelings (combined for the
coastal and river area). Numbers are
word counts within classes (percentages
in parentheses)
1 and 2, positive and negative feelings were reported
by 12% and 25% of the respondents, respectively.
The majority (63%), however, reported that they
were not emotionally affected by their experiences.
The validity of this three-way classication was gen-
erally supported when comparing their scale means
to those of citizens without ood experiences. How-
ever, there were also some unexpected ndings. For
instance, in the river area, groups hardly differed in
the perception of ood consequences. Although this
study assessed emotions as the sole factor channel-
ing the effects of experience, more cognitive evalu-
ations may play a role as well. For instance, having
witnessed (or not witnessed) a oods consequences
provides knowledge that can be applied to estimate
consequences of future oods.
Verbal expressions supported the validity of how
feelings were measured. Study 2 showed that posi-
tive feelings most often reected solidarity and unity
and were related to helping one another during the
1953 ood disaster and during the 1993/1995 near
river oods. In addition, respondents mentioned feel-
ings of sensation and relief, and some were impressed
by the beauty and force of nature (e.g., water ows,
views). Furthermore, Vazquez et al.
(26)
reported a
sense of relief (being alive) and feelings of sol-
idarity (feeling accompanied) among earthquake
survivors in El Salvador. Such positive feelings are
important because they add to peoples social and
psychological resources, which are necessary for cop-
ing with the negative consequences of disasters.
(22,23)
Negative feelings most often reected fear, power-
lessness and helplessness, worries, feelings of uncer-
tainty, and sadness. Moreover, respondents some-
times focused on the amount of damage and number
of casualties and on poor risk and crisis management
1672 Terpstra
Preparedness
intention
Perceived
likelihood
Perceived
dread
Perceived
consequences
-
-
-
+
+
Feelings associated
with previous
experience
Trust in flood
protection
Note: Solid lines are supported in all three areas. Dashed lines reect inconsistent results between studies.
Fig. 5. Summary of path modeling results in Study 1 and 2.
on the part of the government and local water board
(e.g., poor maintenance of ood defenses, poor orga-
nization of the evacuation).
SEM was performed to test the causal relations
among variables. Fig. 5 provides a summary of the
path analysis results (two coastal communities and a
river area community). Although the majority of our
expectations were conrmed, three relations had lit-
tle to no support. First, emotions attached to prior
ood hazard experiences failed to have signicant,
direct effects on ood preparedness intentions. The
absence of an effect of emotions would be in contrast
with Siegrist and Gutscher,
(12)
who argued that nega-
tive emotions explained why ood victims had taken
more precautionary measures than nonvictims. How-
ever, in the two coastal communities emotions inu-
enced preparedness intentions indirectly. That is, in
Study 1 emotions attached to experience with a re-
cent storm were mediated by trust, perceived dread,
and perceived ood likelihood. In Study 2 emotions
attached to experience with the 1953 ood disaster
failed to affect levels of trust in ood protection,
but emotions inuenced preparedness intentions in-
directly through perceived dread only. However, in
the river area (Study 2) emotions attached to ex-
periences of the 1993/1995 high river discharges re-
vealed only direct effects on trust, perceived dread,
and perceived ood likelihood. Thus, the effects of
emotions in terms of the paths through which they af-
fected trust, perceived risk, and preparedness inten-
tions differed between the three study areas. This dis-
crepancy may be explained by the severity of disaster
consequences combined with the time at which the
emotions were assessed. A review on health impair-
ments after disasters showed that high impact disas-
ters lead to more severe health impairments, as com-
pared to moderate or low impact disasters, while the
symptoms of health impairments generally decrease
with time.
(19)
Because Siegrist and Gutscher
(12)
in-
vestigated negative emotions only 9 months after an
actual ood event, the intensity of those emotions
may have caused a more direct and larger effect on
preparedness behavior, as compared to the indirect
effects that we found in our studieswhich were
performed 2 months after a heavy storm in 2006,
about 15 years after the near river oods in 1993
and 1995, and 55 years after the 1953 ood disaster.
Moreover, there may be other qualitative differences
between hazards that inuence how (near) mishaps
are perceived; for example, whether one is exposed
voluntarily
(38)
or whether a hazard is perceived as
natural or manmade
(39)
which are both relevant
dimensions in the context of ood risk management
in the Netherlands.
Second, trust in ood protection failed to have
signicant, direct effects on ood preparedness inten-
tions. However, in the two coastal areas the effects
of trust were mediated by perceived dread and per-
ceived ood likelihood and in the river area also by
the perceived ood consequences. Thus, trust played
an important, central role because it reduced risk
perceptions, had mediated effects causing lower pre-
paredness intentions, and functioned as a mediator in
itself for emotions.
Third, in contrast to claims that perceived con-
sequences are the most important factor in risk
perceptions,
(40)
our results showed that the percep-
tions of ood consequences play a marginal role in
ood preparednessthat is, there were few consis-
tent paths to and from perceived consequences (see
Fig. 5). More consistent, in all three areas trust in
ood protection lessened perceptions of ood like-
lihood and dread, and these in turn predicted pre-
paredness intentions. These results should be viewed
Affective and Cognitive Routes to Flood Preparedness Behavior 1673
in the context of Dutch ood risk management. His-
torically, the Dutch have relied primarily on the
maintenance of strong and sizeable ood defenses.
By observing the magnitude of these ood defenses
(e.g., personally, on television, or on the Internet),
people are provided with a yardstick for estimat-
ing the protection level, most likely inspiring trust in
ood protection. In addition, government risk com-
munications during the previous decades have pri-
marily emphasized the strength of the Dutch ood
defenses. Since risk communications have almost
completely failed to inform citizens about the poten-
tially large consequences of ooding, it is not surpris-
ing that citizens perceptions of ood consequences
have little effect on their attitudes toward private
ood preparedness.
Another way to view the results is to sum-
marize the mechanisms in two parallel paths: an
affective route and a cognitive route, from percep-
tions to behavior. Both paths include multiple depen-
dent variables and indicate how people combine af-
fect and reason in order to respond to risk.
(41)
The
affective route is reected in the chain emotions
attached to previous experience/trust in ood pro-
tection perceived dread preparedness inten-
tion. Affect is an important resource for judging
risks for two reasons. First, because lay people lack
the knowledge needed to make expert judgments,
they require a more intuitive mechanism for mak-
ing risk judgments.
(30)
Our path models showed that
negative feelings decreased trust in ood protection
and increased ood risk perceptions, while positive
feelings had the opposite effect. Because we tapped
these feelings at the end of the questionnaire, prim-
ing effects were impossible. Rather, we argue that
these ndings should be interpreted in the light of
the affect heuristic.
(15)
It seems plausible that by con-
fronting citizens with questions about ood risk, their
memory was triggered to search for relevant infor-
mation. Among the retrieved information were pos-
itive and negative feelings that were associated with
the subjects experiences. Likely, these available af-
fective feelings subsequently guided the respondents
perceptions of ood riskin particular, the amount
of dread evoked by ood risk. In addition, trust in
ood protection was also inuenced by positive and
negative feelings associated with previous ood haz-
ard experiences. This supports the idea that trust and
affect share similarities
(32)
and that both may serve as
a quick guide for judging risks because both reduce
the complexity of judging risk. Consistent with this
view, Poortinga and Pidgeon
(42)
found that peoples
feelings about genetically modied food guided their
trust in risk regulation of genetically modied food.
In addition, Metlay
(43)
found that trust in the U.S.
Department of Energy was partly explained by affec-
tive beliefs about this institution. Second, affective,
intuitive feelings are also important because they are
essential for guiding more cognitive, rational judg-
ments.
(4446)
The more cognitive route is reected in the chain
trust perceived likelihood preparedness inten-
tion. Although we place trust in the affective route,
we also believe that it is informed by logical think-
ing. Research that has been devoted to distinguishing
dimensions of trust has often found that perceived
competence or expertise of risk management institu-
tions underpinned trust ratings.
(31,47)
The perceived
expertise of risk management authorities can be re-
garded as a more cognitive component of trust be-
cause it can be judged based on these authorities
past behavior. In general, Dutch ood risk manage-
ment authorities are highly regarded for their tech-
nical skills.
3
Moreover, the expertise and quality of
ood risk management can also be judged by person-
ally observing the ood defenses. Indeed, our model
supported the idea that beliefs about ood likeli-
hood are lessened by high levels of trust in ood pro-
tection. Low perceptions of ood likelihood in turn
hampered citizens ood preparedness intentions.
We nd it important to shed light on three lim-
itations. First, both surveys had low response rates
(approximately 10%), which might raise questions
about the extent to which the sample is representa-
tive of the Dutch population. Indeed, comparison of
the respondents from each risk area to the NRM2004
database
4(48)
showed the samples overrepresented
males and older residentsjust as in U.S. sam-
ples.
(49,50)
This could be problematic for two rea-
sons. First, overrepresentation of some demographic
categories could produce biases in the item means
of psychological variables, but only to the degree
the latter are correlated with demographic variables.
Such correlations are generally weak, however.
(50)
In
this study, all correlations of demographic variables
3
According to a nationwide opinion poll, knowledge of water
management was the number one source of national pride among
the Dutch (WaterForum Online, May 4, 2006).
4
The NRM2004 database provides information about the demo-
graphic characteristics of the Dutch population on the level of
zip codes within predened ood risk areas (so-called dike rings).
The primary source of NRM are the demographic population
characteristics from Statistics Netherlands, which is the responsi-
ble organization in the Netherlands for collecting and processing
data in order to publish statistics to be used in practice by policy-
makers and for scientic research.
1674 Terpstra
with psychological variables were around 0.10 or
lower, except for gender, which in some cases showed
correlations up to 0.20 (e.g., males perceived less
dread, more trust in ood protection, and smaller
ood likelihood). We did not correct for this because
the sample means in this article were nonrepresenta-
tive of the Dutch population by denition. That is, we
were aiming to assess the effects of a specic popu-
lation characteristicthat is, the effects of emotions
attached to previous ood hazard experiences. Sec-
ond, Lindell and Perry
(51)
argued that low response
rates would affect correlations only if the item vari-
ances were severely restricted by severe overrepre-
sentation of respondents at one end of the response
distribution, which was not the case. Because corre-
lation coefcients are resistant to mean bias the re-
sults can be taken at face value. A second limitation
is that this studylike all cross-sectional designs
has limited ability to draw conclusive causal infer-
ences. Strictly, the cross-sectional data cannot iden-
tify temporal orderthat is, whether, if two variables
are correlated, A caused B or vice versa.
(49,52)
Al-
though hypotheses about causality were carefully de-
rived from the literature, additional research will be
required to provide conclusive evidence; for instance,
by applying longitudinal designs and laboratory ex-
periments. A third limitation concerns the different
itemwordings used to measure the constructs in stud-
ies 1 and 2. For example, the items used to measure
ood preparedness intentions in study 1 focus more
on seeking information and general intentions to pre-
pare, while in study 2 the items concern very concrete
actions. Differences in item wordings may have had
small effects on the construct means and their corre-
lations. This was also the reason that the two studies
could not be analyzed simultaneously. Nevertheless,
in both studies validation procedures suggested that
the expressed emotions, risk perception, and behav-
ioral intentions were reliably measured and captured
by our proposed causal models. Future studies are
needed to challenge our ndings.
In summary, the goal of this article was to gain
insight about the mechanisms that inuence citizens
ood preparedness intentions, with a specic focus
on the role of emotions, trust, and perceived risk. The
results showed that both affective and cognitive pro-
cesses together inuenced citizens behavioral inten-
tions. This nding is consistent with much of the work
that has been conducted on affect and decision mak-
ing.
(16,46)
Moreover, in addition to negative emotions,
positive emotions also play a role. Not only do ac-
tual disasters seem to evoke such emotions, but near
misses and milder threats also seem to contribute
to peoples cognitive and affective responses to
risk.
We recommend that measures of emotions be
improved in future studies, for instance, by using the
verbal descriptions to develop multi-item scales. In
addition, the current model can be extended by in-
cluding more factors that explain preparedness be-
havior, for instance, by including peoples beliefs
about the efcacy of preparedness measures (e.g., an
emergency kit) and their abilities to adopt such mea-
sures. The results further suggest that risk communi-
cations should accommodate both the affective and
cognitive routes for motivating disaster prepared-
ness behavior. The current study showed that pre-
vious experiences fuel the affective route with emo-
tions. However, because ood probability is low in
the Netherlands, the challenge is therefore how to
infuse needed doses of feeling into circumstances
where lack of experience may otherwise leave us too
coldly rational (p. 320).
(17)
We believe that the
role of positive emotions in promoting preparedness
among citizens could be a valuable direction for fu-
ture research.
ACKNOWLEDGMENTS
The author thanks Jan Gutteling at the Univer-
sity of Twente for his helpful comments on earlier
drafts of this article. The research was carried out
at the University of Twente (Psychology and Com-
munication of Health and Risk) in the framework of
the PROMO project in the Netherlands. The project
was supported by the Dutch Living with Water pro-
gram (project reference number P3062), and national
and provincial governments. None of the conclusions
expressed here necessarily reect views other than
those of the author.
REFERENCES
1. Linnerooth-Bayer J, Amendola A. Introduction to special is-
sue on ood risks in Europe. Risk Analysis, 2003; 23(3):537
543.
2. Jonkman SN. Global perspectives on loss of human life
caused by oods. Natural Hazards, 2005; 34(2):151175.
3. IPCC. Summary for policymakers. In Solomon S, Qin D,
Manning M, Chen Z, Marquis M, Averyt KB, Tignor M,
Miller HL (eds). Climate Change 2007: The Physical Science
Basis. Contribution of Working Group I to the Fourth As-
sessment Report of the Intergovernmental Panel on Climate
Change. Cambridge, UK and New York, USA: Cambridge
University Press; 2007.
4. Terpstra T, Gutteling JM. Households perceived responsi-
bilities in ood risk management in the Netherlands. Inter-
national Journal of Water Resources Development, 2008;
24(4):555565.
5. Grothmann T, Reusswig F. People at risk of ooding: Why
Affective and Cognitive Routes to Flood Preparedness Behavior 1675
some residents take precautionary action while others do not.
Natural Hazards, 2006; 38(12):101120.
6. Ten Brinke WBM, Bannink BA. Dutch Dikes and Risk
Hikes: A Thematic Policy Evaluation of Risks of Flooding in
the Netherlands. Bilthoven, The Netherlands: RIVM, 2004.
7. Terpstra T, Lindell MK, Gutteling JM. Does communicating
(ood) risk affect (ood) risk perceptions? Results of a quasi-
experimental study. Risk Analysis, 2009; 29(8):11411151.
8. Peacock WG, Brody SD, Higheld W. Hurricane risk percep-
tions among Floridas single family homeowners. Landscape
and Urban Planning, 2005; 73(23):120135.
9. Siegrist M, Gutscher H. Flooding risks: A comparison of lay
peoples perceptions and experts assessments in Switzerland.
Risk Analysis, 2006; 26(4):971979.
10. Halpern-Felsher BL, Millstein SG, Ellen JM, Adler NE,
Tschann JM, Biehl M. The role of behavioral experience in
judging risks. Health Psychology, 2001; 20(2):120126.
11. Lindell MK, Perry RW. Communicating Environmental Risk
in Multiethnic Communities Thousand Oaks, CA: Sage Pub-
lications, Inc., 2004.
12. Siegrist M, Gutscher H. Natural hazards and motivation for
mitigation behavior: People cannot predict the affect evoked
by a severe ood. Risk Analysis, 2008; 28(3):771778.
13. Keller C, Siegrist M, Gutscher H. The role of the affect and
availability heuristics in risk communication. Risk Analysis,
2006; 26(3):631639.
14. Vastfjall D, Peters E, Slovic P. Affect, risk perception and fu-
ture optimism after the tsunami disaster. Judgment and Deci-
sion Making Journal, 2008; 3(1):6472.
15. Slovic P, Finucane ML, Peters E, MacGregor DG. The affect
heuristic. European Journal of Operational Research, 2007;
177(3):13331352.
16. Loewenstein GF, Weber EU, Hsee CK, Welch N. Risk as
feelings. Psychological Bulletin, 2001; 127(2):267286.
17. Slovic P, Finucane ML, Peters E, MacGregor DG. Risk as
analysis and risk as feelings: Some thoughts about affect, rea-
son, risk, and rationality. Risk Analysis, 2004; 24(2):311322.
18. Finucane ML, Alhakami A, Slovic P, Johnson SM. The affect
heuristic in judgments of risks and benets. Journal of Behav-
ioral Decision Making, 2000; 13(1):117.
19. Norris FH, Friedman MJ, Watson PJ, Byrne CM, Diaz E,
Kaniasty K. 60,000 disaster victims speak: Part I. An empir-
ical review of the empirical literature, 19812001. Psychiatry-
Interpersonal and Biological Processes, 2002; 65(3):207239.
20. Agaibi CE, Wilson JP. Trauma, PTSD, and resilience: A
review of the literature. Trauma Violence & Abuse, 2005;
6(3):195216.
21. Folkman S, Moskowitz JT. Positive affect and the other side
of coping. American Psychologist, 2000; 55(6):647654.
22. Fredrickson BL. The role of positive emotions in positive psy-
chology: The broaden-and-build theory of positive emotions.
American Psychologist, 2001; 56(3):218226.
23. Tugade MM, Fredrickson BL. Resilient individuals use posi-
tive emotions to bounce back from negative emotional expe-
riences. Journal of Personality and Social Psychology, 2004;
86(2):320333.
24. Fredrickson BL, Tugade MM, Waugh CE, Larkin GR. What
good are positive emotions in crises? A prospective study of
resilience and emotions following the terrorist attacks on the
United States on September 11th, 2001. Journal of Personality
and Social Psychology, 2003; 84(2):365376.
25. Tedeschi RG, Calhoun LG. Posttraumatic growth: Concep-
tual foundations and empirical evidence. Psychological In-
quiry, 2004; 15(1):118.
26. Vazquez C, Cervellon P, Perez-Sales P, Vidales D, Gaborit
M. Positive emotions in earthquake survivors in El Salvador
(2001). Journal of Anxiety Disorders, 2005; 19(3):313328.
27. Perez-Sales P, Cervellon P, Vazquez C, Vidales D, Gaborit
M. Post-traumatic factors and resilience: The role of shelter
management and survivours attitudes after the earthquakes
in El Salvador (2001). Journal of Community & Applied So-
cial Psychology, 2005; 15(5):368382.
28. Tang CSK. Positive and negative postdisaster psychological
adjustment among adult survivors of the Southeast Asian
earthquake-tsunami. Journal of Psychosomatic Research,
2006; 61(5):699705.
29. Earle TC, Cvetkovich GT. Social Trust. Towards a Cos-
mopolitan Society. London: Praeger, 1995.
30. Siegrist M, Cvetkovich GT. Perception of hazards: The role
of social trust and knowledge. Risk Analysis, 2000; 20(5):713
719.
31. Siegrist M, Gutscher H, Earle TC. Perception of risk: The in-
uence of general trust, and general condence. Journal of
Risk Research, 2005; 8(2):145156.
32. Poortinga W, Pidgeon NF. Trust in risk regulation: Cause or
consequence of the acceptability of GM food? Risk Analysis,
2005; 25(1):199209.
33. Byrne MB. Structural Equation Modeling with AMOS. Ba-
sic Concepts, Applications, and Programming. Mahwah, NJ:
Lawrence Erlbaum Associates, 2001.
34. Kline RB. Principles and Practice of Structural Equation
Modeling (2nd edition). New York: Guilford Press, 2005.
35. Schumacker RE, Lomax GL. A Beginners Guide to
Structural Equation Modeling (2nd edition). Mahwah, NJ:
Lawrence Erlbaum, 2004.
36. Peters O. A social cognitive perspective on mobile communi-
cation technology use and adoption. Social Science Computer
Review, 2009; 27(1):7695.
37. Brown RL. Assessing specic mediational effects in com-
plex theoretical models. Structural Equation Modeling, 1997;
4(2):142156.
38. Slovic P. Perception of risk. Science, 1987; 236:280285.
39. Sjoberg L. Factors in risk perception. Risk Analysis, 2000;
20(1):111.
40. Sj oberg L. Will the real meaning of affect please stand up?
Journal of Risk Research, 2006; 9(2):101108.
41. Finucane ML, Holup JL. Risk as value: Combining affect and
analysis in risk judgments. Journal of Risk Research, 2006;
9(2):141164.
42. Poortinga W, Pidgeon NF. Prior attitudes, salient value sim-
ilarity, and dimensionality: Toward an integrative model of
trust in risk regulation. Journal of Applied Social Psychology,
2006; 36(7):16741700.
43. Metlay D. Institutional trust and condence: A journey into
a conceptual quagmire. Pp. 100116 in Cvetkovich GT, Lofst-
edt RE (eds). Social Trust and the Management of Risk. Lon-
don: Earthscan, 1999.
44. Damasio AR. Descartes Error: Emotion, Reason, and the
Human Brain. New York: Avon, 1994.
45. Epstein S. Integration of the cognitive and the psychodynamic
unconscious. American Psychologist, 1994; 49(8):709724.
46. Slovic P, Peters E, Finucane ML, MacGregor DG. Af-
fect, risk, and decision making. Health Psychology, 2005;
24(4):S35S40.
47. Poortinga W, Pidgeon NF. Exploring the dimensionality of
trust in risk regulation. Risk Analysis, 2003; 23(5):961972.
48. Goudappel Coffeng. NRM-basisbestand 2004. Delft: Dienst
Verkeer en Scheepvaart, 2004.
49. Lindell MK, Hwang SN. Households perceived personal risk
and responses in a multihazard environment. Risk Analysis,
2008; 28(2):539556.
50. Lindell MK, Arlikatti S, Prater CS. Why people do what
they do to protect against earthquake risk: Perceptions of
hazard adjustments and their attributes. Risk Analysis, 2009;
29(8):10721088.
51. Lindell MK, Perry RW. Household adjustment to earthquake
hazard. A review of research. Environment and Behavior,
2000; 32(4):461501.
52. James LR, Mulaik SA, Brett JM. Causal Analysis: Assump-
tions, Models and Data. Beverly Hills, CA: Sage, 1982.