Safety and Accident Prevention

Risk communication and the willingness to follow evacuation instructions

in a natural disaster

Sverre Kjetil Rød a,b *, Carl Botan

c and Are Holen

a,d

a Department of Neuroscience, Norwegian University of Science and Technology, Trondheim, Norway;

b Faculty of Media and Journalism, Volda University College, Volda, Norway;

c Department of Communication, George Mason University, Fairfax, VA, USA;

d St. Olavs

University Hospital, Trondheim, Norway

(Received 23 June 2011; final version received 18 October 2011)

Experts have predicted that some time in the future, a rockslide in one of Norway’s narrow fjords will produce a tsunami that may run up above the sea level as high as 82 m (269 feet). This study explores the willingness of the public to follow evacuation instructions before a natural disaster with a high probability, but unknown timeframe. Parts of the studied area were involved in a similar disaster in 1934. For this study, all residents (875) 18 years or older in four threatened communities received a questionnaire; 382 (43.6%) responded. The aim was to determine to what extent socio-demographic variables, individual differences, and the relationships among experts and residents determine a willingness to follow recommended evacuation instructions during emergencies. The survey found that those who reported a willingness to follow instructions lived in areas with a disaster history. Often they had university graduate degrees, good relationships with experts, and found risk information useful. Gender and individual differences did not separately contribute significantly to willingness to evacuate.

Keywords: crisis communication; risk; risk assessment; trust; distrust; evacuation instructions; tsunami

Introduction

Approximately 3000 people live in a tsunami danger zone on the west coast of Norway. The Åknes rockslide site is one area in Western Europe with the highest risk of a tsunami. Lasers, radar, remote video and human staff keep a continuous watch over the site where 18–54 million cubic metres (635 million–1.9 billion cubic feet) of rock may fall into the fjord. In the worst case scenario, a tsunami is predicted to run 82 m (269 feet) above the shoreline in the community of Hellesylt five min after the rockslide. Most of the town, a local school, and a facility for the older people could be submerged. Five minutes later, the tsunami (max. 63 m or 206 feet) will strike the UNESCO-designated tourist community of Geiranger with a wave, submerging at least one school, day care centres and commercial establishments. The potential waves that could affect 10 municipalities are projected to be 13 m (42 feet)

*Corresponding author. Email: [email protected]

Health, Risk & Society

Vol. 14, No. 1, February 2012, 87–99

ISSN 1369-8575 print/ISSN 1469-8331 online

� 2012 Taylor & Francis http://dx.doi.org/10.1080/13698575.2011.641522

http://www.tandfonline.com

in Tafjord and 4m (13 feet) in Fjørå. A similar but less severe calamity called the ‘Tafjord disaster’ struck these two communities in 1934, resulting in 40 fatalities.

This study focuses on how socio-demographic characteristics, individual differences, relationships with experts, i.e. geologists, and expert opinions may influence the attitude of the public and their willingness to accept instructions from those responsible for communication and safety in high-risk situations. The literature on risk information in such situations is reviewed to determine what may influence the willingness to adhere to warnings and instructions relating to disasters.

Responsive to warnings about disasters: a review of relevant literature

At the time of the ‘Tafjord disaster,’ the tendency was to neglect external stressors in psychic trauma (Weisæth 2002), consequently survivors thought the best cure was to let time heal the wounds by not sharing their traumatic stories. Thus, the incident was not widely discussed before 1985 when a local researcher published the first account of the event (Furseth 1985). Nevertheless, awareness of the ‘Tafjord disaster’ must have been quite prevalent in residents’ minds.

Sattler et al. (2000) suggest that there is a relationship between disaster experience and disaster preparedness. People who live in areas where natural disasters occur are more likely to adhere to evacuation warnings than those who may never have been exposed to such risks (Drabek and Boggs 1968, Sattler et al. 2000).

According to the United States Department of Homeland Security (U.S. DHS 2004), evacuation is the ‘organized, phased, and supervised withdrawal, dispersal, or removal of civilians from dangerous or potentially dangerous areas, and their reception and care in safe areas.’ In Norway, there are legal requirements that people follow instructions to move out of dangerous areas when they are instructed to do so. As in the Åknes case, geologists advise the local government and police when to issue instructions to local population to evacuate based on their surveillance of natural hazards.

As noted by Lemyre et al. (2009), most research on risk has taken the form of risk analysis and considered the probability and consequences of hazards, rather than focusing on risk perception; how people assess and respond to those risks, and this focus may lead ‘to an inaccurate belief that events affect all individuals equally’ (p. 207). Risk managers rarely focus on how people with different characteristics process risk information and relate to disasters (Lemyre et al. 2009). This article attempts to bridge the gap between these two perspectives. There is a well-established body of research that indicates ‘at risk’ individuals tend to ignore health warnings (see for example Lindell and Perry 2000, Rohrmann 2000, Heller et al. 2005). Such non- protective responses may be explained by denial, wishful thinking, and fatalism (Grothmann and Reusswig 2006). Alternatively, the reasons may be related to communicator skills, ambiguous risk information, personal characteristics of the public such as socio-demographics, prior experience, individual differences, relation- ships with experts and influence through informal social networks and the news media (Fischhoff 2000, Rohrmann 2000, Slovic 2000, Heller et al. 2005).

Elements of our social identity come from groups that we feel most associated with (Dawnay and Shah 2005). The Social Amplification of Risk Framework is founded on the individual’s social, psychological and cultural network, and is also

88 S.K. Rød et al.

reflected by socio-demographic and individual differences. Kasperson et al. (1988) argue risk can intensify or attenuate, depending on individual personal interpreta- tions of the threat. According to Breakwell (2000), direct experience and communication with others are two major ways of recognising hazards. Thus, the risk may be interpreted individually through psychological, social, and cultural filters (Breakwell 2000).

As noted by Sjöberg (2004), social trust, that is trust in experts or organisations, plays a major role in individual acceptance of risk messages. According to Luhmann (1979, 1988) and Giddens (1990), trust in experts and authorities reduces uncertainty among people. Numerous studies have documented correlations between people’s trust in the government and their willingness to comply with official evacuation orders (West 2001, Granatt 2004, Burton et al. 2006, West and Orr 2007). The more the people trust government officials and evacuation orders, the more willing they are to comply (West and Orr 2007).

When assessing risks, experts typically define it scientifically in terms of probabilities that a hazard of a given magnitude will occur at a given location within a given time (Kaplan and Garrick 1981, Jardine et al. 2009). They also indicate the expected consequences of the hazard on people and infrastructure (Sandman 1993). Experts or governmental agencies in charge of communicating risks focus their efforts on conveying information in ways that they assume will be well received and understood by the public. In the Åknes case, the municipalities and the experts used three means for disseminating the risk information: public meetings, written documentation, and web-based information. Public meetings have been held about once every year in the communities that are likely to be affected, and this has been done since the Åknes/Tafjord project was established in 2004. At the meetings, the participants are given expert information and updates on the situation. They may ask questions or engage in discussions with the experts. Municipality officials usually chair the meetings with support from the experts. Written documentation, that is information given in pamphlets, bulletins and plans, is distributed to the inhabitants by post. Web-based information implies reading, seeing or hearing of information that is frequently updated and presented in print, audio-visually, or digitally at the project’s internet site, www.aknes.no

According to Heller et al. (2005) and Fischhoff (1995), the field of risk communication during the last decades has moved from merely providing the public with facts about what to do, to recognising that personal and social factors require more attention. ‘There is now a recognition that telling people about health and environmental dangers is not likely to lead to risk-reducing behaviours without attention to social influences’ (Heller et al. 2005, p. 400).

According to Slovic et al. (2004), there are two fundamental ways in which humans understand risk. The analytical system uses probability calculus, formal logic, and risk assessment, whereas the experiential system is based on effect — ‘a faint whisper of emotion’ (Slovic et al. 2005, p. 35). It is intuitive, fast, mostly automatic, not characterised by conscious awareness, and ‘encodes reality in images, metaphors, and narratives associated with feelings’ (Slovic et al. 2004, p. 313). These two ways of coping with risk may be related to the filtering process described by the social amplification of risk framework. People may evaluate risk cognitively and through social learning, but they react to it intuitively and emotionally (Loewenstein et al. 2001, Thalmann and Wiedemann 2006). The social amplification of risk framework and Slovic approaches are two different ways of acknowledging the

Health, Risk & Society 89

complexity of risk perception. There is no one single way to process and interpret risk information.

Given the limits of our current understanding of the ways in which target populations respond to warnings and instructions about the measures to protect themselves from natural disasters, this article explored the following question: How do socio-demographic characteristics, individual differences, and relationships with experts relate to the willingness to follow evacuation instructions prior to a natural disaster?

Methodology

Procedure

In February 2008, 875 booklets with 136 items were distributed by post. Ninety items in the questionnaire were taken from established standardised questionnaires (Craig et al. 1984, pp. 79–80, Eysenck et al. 1985, Endresen 1991), addressing some of the variables considered in this study. The remaining 46 items were unique to this study and generated by focus groups run by the first author. The booklet also contained background information about the study. Two reminders were sent at one-month intervals. Those under age 18 were not included in the study, and the questionnaires were treated anonymously in accordance with the policies of the Norwegian Social Science Data Services (NSD).

Participants

A total of 382 people (43.6%) from the four small communities responded, including 53.3% females (n ¼ 200) and 46.7% males (n ¼ 175), adequately reflecting the gender distribution of the population which was 51% females and 49% males. To avoid breaches of anonymity through disaggregation in small communities, participants were placed in four age groups. The response rate of the age group 18–33 was 18.9%, age group 34–49 was 28.5%, age group 50–65 was 24.8% and age group 66 years and above was 27.7%. Of the total responses, 69 were from Geiranger (39.9%), 200 from Hellesylt (42.1%), 40 from Tafjord (40.4%) and 66 from Fjørå (51.6%).

Scoring

The study items were scored on a five-point Likert-type scale ranging from 1 (suits me very poorly) to 5 (suits me very well). Scoring allowed a maximum of one point per item on the scale. The response options are translated from the Norwegian version into English. The original Norwegian term ‘passar’ was translated as ‘suits me’ as the nearest English equivalent, although ‘suits me’ may not be an appropriate response option for another audience, for example, in the UK or US. Some variables were single items, others were the unweighted sum or mean of items with high correlations.

Willingness to adhere to evacuation instructions

To capture the willingness to adhere to evacuation instructions given by authorities, a composite, continuous variable was created from three thematically and

90 S.K. Rød et al.

statistically correlated items (N ¼ 369, M 3.65, SD 0.86, MIC 0.26 [mean inter-item correlation]). The three items were: ‘I will follow the evacuation procedures when the authorities demand it’, ‘When the evacuation warnings are triggered, I will follow my own intuition, not the orders from the authorities, to take care of my family and myself’ (inversely coded), and ‘I will lose faith in the experts if the evacuation starts, but later is cancelled because the rockslide didn’t happen’ (inversely coded). All three items were coded from 1 to 5. In the study, this composite, continuous variable was used as the dependent.

Socio-demographic characteristics

Socio-demographic variables included having children (1 – yes, 0 – no) and gender (1 – female, 0 – male). Residents of geographically, culturally and economically similar communities close to the rockslide site were grouped dichotomously by whether or not their home area had a disaster history, gender, age, level of education and disaster history. The respondents were placed into four groups as dummy variables — 18–33 (N ¼ 71), 34–49 (N ¼ 106), 50–65 (N ¼ 87) and 66 and older (N ¼ 89). Likewise, respondents were assigned to one of five groups depending on their level of education — high school (N ¼ 58), vocational school (N ¼ 148), junior college (N ¼ 64), university undergraduate degree (N ¼ 56), or university graduate degree (N ¼ 41).

Individual differences

Psychological defence mechanisms indicated the respondents’ general relationships to fear, danger and threats, i.e. how likely a person is to respond adequately to threats for psychological reasons. Seven subscales, each with six items, explored the traditional psychodynamic defence mechanisms. They were taken from Plutchik Life Style Index (LSI-R) revised (Endresen 1991), shortened version, where explanations of the various defence mechanisms are also found. Participants responded to each item by answering ‘yes, suits me’ or ‘no, does not suit me’. One of Plutchik’s original eight defence mechanisms was not found relevant for the study and was omitted.

To measure the personality trait neuroticism associated with a person’s persistent worry, anxiety and dysphoric affect, Eysenck’s Personality Questionnaire – Revised (EPQ-R) was applied (Eysenck et al. 1985). The questionnaire originally included 106 items, and 12 items addressed neuroticism. Participants were asked to respond yes or no to those 12 items. A neuroticism score for each participant was calculated based on the sum of positive endorsements.

Externality was measured by the Locus of Control of Behaviour (Craig et al. 1984, pp. 79–80), a scale that was designed to measure the extent to which subjects perceive responsibility for their personal problem behaviour. The questionnaire included 17 Likert-style items, each with six response options. Subjects were asked to indicate their agreement or disagreement to each item. Item scores went from 0 (strongly disagree) to 5 (strongly agree). In accordance with the recommendations of Craig et al. (1984), six items covering the internal locus of control were inverted and added to the sum of those covering the external locus of control. In this way, the total score for all 17 items gave the final externality score with a potential range of 0 to 85. Higher scores indicate external control, which implies that life tends to be seen by the person as determined by fate, coincidence or authority, while lower scores

Health, Risk & Society 91

indicate internal control, implying that the person sees himself as more capable of shaping his destiny.

Relationship with experts

Two variables were included – trust in experts and useful risk information. According to Slovic (1999, p. 697), much of the controversy between the public, industry and risk management professionals is related to trust and to effectiveness of risk communication. The continuous variable trust in experts (N ¼ 374, M 3.48, SD 0.78, MIC 0.34) included five items: ‘I have great trust in the experts of the emergency project,’ ‘The researchers exaggerate the risk of a potential rockslide’ (inversely coded), ‘If the rockslide comes, I think the magnitude will be much less than predicted by the experts’ (inversely coded), ‘The researchers know less about rockslides and tsunamis than people who have lived with such risk in generations’ (inversely coded), and ‘I think the local knowledge about the rockslide and the tsunami risk accumulated over generations is more important than the information from the experts’ (inversely coded).

The other continuous variable, useful risk information (N ¼ 370, M 3.58, SD 0.79, MIC 0.40) included three items: ‘The public meetings about the rockslide risk give me useful information,’ ‘The written documentation about the rockslide risk is useful to me’, and ‘The Åknes/Tafjord web site gives me useful information’. The variable covered how people generally perceived the usefulness of risk communica- tion, face-to-face, written, and through the media. This variable is included in the study to capture how risk information may relate to public willingness to adhere to evacuation instructions.

Data analyses

Point-biserial correlation coefficients mixing dichotomous and continuous variables (Tabachnick and Fidell 2007, p. 914) were calculated for all variables (Table 1). Correlations were used to check for variables unrelated to the other variables in the analyses. Those variables emerging with significant bivariate correlations to the dependent variable were subsequently grouped into three thematic blocks and used in the regression analyses. The analysis examined the importance of socio- demographic characteristics, individual differences, and relationships with experts in predicting the willingness to adhere to evacuation instructions.

According to Pallant (2007), ‘for scales with a small number of items (e.g., less than 10), it is sometimes difficult to get a decent Cronbach alpha value and you may wish to consider reporting the mean inter-item correlation value’ (p. 98). Briggs and Cheek (1986) recommend an optimal range of 0.2 to 0.4 for such scales. Therefore, the variables ‘willingness to adhere to evacuation instructions’ (three items), ‘trust in experts’ (five items), and ‘useful risk information’ (three items) are reported by MIC (mean inter-item correlation).

Findings

Socio-demographic characteristics that had significant correlations with willingness to adhere to instructions were found for those ages 18–33, 34–49, 66 and older, as well as high school graduates, those with university graduate degrees, and for those

92 S.K. Rød et al.

T a b le

1 .

P o in t- b is er ia l in te r- co rr el a ti o n s b et w ee n

v a ri a b le s in

th e a n a ly se s re p re se n ti n g

so ci o -d em

o g ra p h ic

ch a ra ct er is ti cs , in d iv id u a l d iff e re n ce s,

a n d

re la ti o n sh ip s w it h ex p er ts .

V a ri a b le

1 2

3 4

5 6

7 8

9 1 0

1 1

1 2

1 3

1 4

1 5

1 6

1 7

1 8

1 9

2 0

2 1

2 2

2 3

2 4

2 5

1 . W il li n g n es s to

a d h er e

– .0 4

.1 4 * *

.1 0

7 0 .1 2 *

0 .1 9 * *

0 .0 2

7 0 .1 2 *

7 .1 4 * *

7 .0 6

.0 3

.0 5

.2 2 * *

7 .0 6

.0 3

7 .0 9

7 .1 5 * *

7 .1 8 * *

7 .0 1

7 .0 6

7 .2 0 * *

7 .0 5

7 .2 7 * *

.5 6 * *

.3 3 * *

2 . G en d er

– .0 2

.0 5

.0 9

.0 1

7 .0 8

7 .0 2

.0 2

7 .1 0

.1 2 *

7 .0 3

.0 4

7 .1 1 *

7 .1 9 * *

7 .2 3 * *

.1 8 * *

.0 3 9

.0 9

.0 1

.2 0 * *

7 .0 1

7 .0 5

.1 0

7 .0 8

3 . H a v in g ch il d re n

– .0 1

7 .5 1 * *

.1 6 * *

.2 2 * *

.0 6

7 .0 4

.0 5

.0 3

.0 1

7 .0 6

.0 4

7 .0 1

7 .1 2 *

7 .1 5 * *

7 .1 6 * *

7 .0 5

7 .1 7 * *

7 .2 1 * *

.0 5

7 .0 7

.0 7

.1 9 * *

4 . D is a st er

h is to ry

– 7 .1 1 *

7 .1 1 *

.1 7 * *

.0 4

7 .0 0

.0 7

7 .0 2

7 .0 3

7 .0 3

7 .0 1

7 .0 4

7 .0 7

7 .0 5

7 .1 0

7 .0 3

7 .0 8

7 .0 4

7 .0 1

7 .0 5

.0 5

7 .0 9

A g e (y rs ):

5 . 1 8 7 3 3

– 7 .3 0

7 .2 7 * *

7 .2 9 * *

.0 2

7 .1 3 *

.1 3 *

.0 3

.0 5

.0 2

7 .0 9

7 .0 2

.1 5 * *

.2 4 * *

.0 2

.2 7 * *

.1 4 * *

.0 9

.0 3

.0 3

7 .1 7 * *

6 . 3 4 7 4 9

– 7 .3 5 * *

7 .3 8 * *

7 .1 7 * *

7 .1 0 *

.2 5 * *

.0 2

.1 0 *

7 .0 2

7 .1 6 * *

7 .1 5 * *

.0 7

7 .1 1 *

7 .0 4

.1 0

7 .0 4

.1 0

7 .2 7 * *

.1 9 * *

.0 6

7 . 5 0 7 6 5

– 7 .3 5 * *

7 .0 7

.1 1 *

7 .1 4 * *

.0 4

.0 4

7 .0 6

.0 7

.0 7

7 .1 1 *

7 .1 8 * *

7 .0 7

7 .1 5 * *

7 .0 7

7 .1 1 *

7 .0 1

.0 1

.1 0 *

8 . 6 6 4

– .2 3 * *

.1 3 *

7 .2 3 * *

7 .0 7

7 .1 7 * *

.0 6

.1 8 * *

.1 0

7 .0 9

.0 8

.0 9

7 .1 9 * *

7 .0 1

7 .0 8

.2 9 * *

7 .1 7 * *

7 .0 2

E d u ca ti o n :

9 . H ig h sc h o o l

– 7 .3 4 * *

7 .1 9 * *

7 .1 8 * *

7 .1 5 * *

.0 6

.0 2

.0 3

7 .0 2

.1 2 *

.0 7

7 .0 8

.0 9

7 .0 5

.1 7 * *

7 .1 2 *

7 .0 5

1 0 . V o ca ti o n a l sc h o o l

– 7 .3 6 * *

7 .3 3 * *

7 .2 8 * *

7 .1 2 *

.0 5

.1 4 * *

7 .0 1

.0 1

7 .0 9

.0 2

.0 7

7 .0 7

.1 8 * *

7 .1 3 *

7 .0 9

1 1 . Ju n io r co ll eg e

– 7 .1 9 * *

7 .1 6 * *

.0 6

7 .1 7 * *

7 .2 0 * *

.1 3 *

.0 1

7 .0 5

7 .1 1 *

.0 0

.1 0

7 .1 3 *

.0 6

.0 0

1 2 . U n iv er si ty

u n d er g ra d .

– 7 .1 4 * *

7 .0 3

.0 7

7 .0 4

7 .0 6

7 .1 6 * *

.1 3 *

7 .0 9

7 .0 6

7 .0 2

7 .1 5 * *

.1 3 *

.0 6

1 3 . U n iv er si ty

g ra d .

– .0 5

7 .0 5

7 .0 2

7 .0 4

.0 2

7 .0 3

.0 4

7 .1 5 * *

.0 5

7 .1 5 * *

.1 9 * *

.1 3 * *

1 4 . D en ia l

– .1 2 *

7 .0 8

7 .0 9

.2 0 * *

7 .0 5

.0 7

7 .3 2 * *

.3 8 * *

7 .1 3 *

7 .0 6

.1 1 *

1 5 . In te ll ec tu a li za ti o n

– .2 3 * *

7 .0 6

.0 2

.0 7

7 .0 8

.0 2

.0 4

.0 7

7 .0 3

.0 7

1 6 . R ep re ss io n

– 7 .0 1

.0 5

7 .0 0

.0 3

.1 1 *

7 .1 5 * *

.1 7 * *

7 .1 5 * *

7 .0 8

1 7 . R eg re ss io n

– .3 7 * *

.0 9

.3 6 * *

.4 9 * *

7 .0 9

.2 3 * *

7 .1 0

7 .1 0

1 8 . C o m p en sa ti o n

– .0 4

.1 6 * *

.3 5 * *

7 .0 0

.2 8 * *

7 .1 4 *

7 .0 5

1 9 . P ro je ct io n

– .0 2

.1 8 * *

7 .0 0

.0 9

.0 2

7 .0 1

2 0 . D is p la ce m en t

– .3 2 * *

.0 6

.0 4

7 .0 1

7 .0 9

2 1 . N eu ro ti ci sm

– 7 .3 3 * *

.3 9 * *

7 .1 3 * *

7 .1 4 * *

2 2 . S el f- effi

ca cy

– .4 0 * *

.0 1

.0 3

2 3 . E x te rn a li ty

– .2 7 * *

7 .1 0

2 4 . T ru st

in ex p er ts

– .3 7 * *

2 5 . U se fu l ri sk

in fo .

–

M ea n

3 .6 5

.5 3

.7 7

.2 8

.1 6

.2 8

.2 4

.2 7

1 5

.3 9

.1 7

.1 5

.1 1

3 .4 7

3 .5 4

1 .9 9

1 .7 1

1 .4 2

4 .9 6

.9 7

3 .2 6

1 3 .7 9

5 4 .7 1

3 .4 8

3 .5 8

S D

.8 6

.5 0

.4 2

.4 5

.3 9

.4 5

.4 3

4 5

3 6

.4 9

.3 7

.3 5

.3 1

1 .3 9

1 .3 2

1 .4 4

1 .3 9

1 .4 2

1 .1 5

1 .0 4

2 .6 7

3 .1 9

1 1 .9 5

.7 8

.7 9

C o n ti n u o u s/ D ic h o to m o u s ( x )

1 – 5

x x

x x

x x

x x

x x

x x

0 – 6

0 – 6

0 – 6

0 – 6

0 – 6

0 – 6

0 – 6

0 – 1 2

5 – 2 0

0 – 8 5

1 – 5

1 – 5

* * p 5 .0 1 , * p 5 .0 5 .

having children. Of the individual differences, the defence mechanisms regression and compensation were significant, but also the externality from the locus of control measure (Craig et al. 1984) and the neuroticism scale developed by Eysenck (1985).

For relationships with experts, both trust in the experts and useful risk infor- mation correlated significantly with willingness to adhere. All significant variables, except ages 18–33 (p 5 0.005) had correlations of p 5 0.001. The tolerance of variables in the regression analysis was examined. The variables ‘age 34–49’ and ‘junior college’ as a level of education accounted for too little of the unique variance of the dependent variable; the tolerance was less than 0.1. In such cases, the recommendation is that ‘the regression analysis should be conducted again without the violating IV’ (Mertler and Vannatta 2005, p. 194). Therefore, the two variables were excluded from the model.

Table 2 shows that three thematic blocks of predictors accounted for 39.8% of the variance of the willingness to adhere to evacuation procedures, F (16, 316) ¼ 52.773, p 5 0.001. Within this explained variance, the block of socio-demographic characteristics accounted for 13.7%, F (10, 295) ¼ 4.675, p 5 0.001. The block of

Table 2. Willingness to adhere to evacuation instructions as the dependent variable explained by three thematic blocks of independent variables — socio-demographic characteristics, individual differences, and relationship with experts.

Predictor variable b R R2 R2 adj. R2 change F Df

Block 1: Socio-demographic characteristics Gender 0.026

Age (yrs) 18–33 70.080 50–65 70.079 664 70.035

Having children 0.057 Disaster history 0.098* Education High school 70.028 Vocational school 0.039 University undergraduate degree

70.016

University graduate degree

0.117*

After Block 1 0.37 0.137 0.108 4.675** 10, 295

Block 2: Individual differences Regression 70.001 Compensation 70.068 Externality 0.026 Neuroticism 70.069 After Block 2 0.42 0.179 0.139 0.042 4.519** 291

Block 3: Relationship with experts Trust in experts 0.464** Useful risk information 0.116* After all three blocks 0.63 0.398 0.365 0.220 11.958** 289

Notes: F and p represent incremental change by block. R and R 2 are cumulative values. R

2 change explains

each block’s additional change of the variance. **p 5 .001, *p 5 .005.

94 S.K. Rød et al.

individual differences accounted for an additional 4.2%, F (14, 291) ¼ 4.519, p 5 0.001, even though none of the components rose to the level of statistical significance. The block covering relationships with experts, including trust in experts and useful risk information, accounted for an additional 22%, F (16, 289) ¼ 11.958, p 5 0.001. The beta (b) column in Table 2 presents the standardised regression coefficients for the 16 variables of the model. The beta coefficients indicate the relative effect of each variable as a predictor of change in the willingness to adhere, partialling out all others predictor variables. In the final model, only blocks one and three were statistically significant with trust in experts having a higher beta value (b ¼ 0.46, p 5 0.001) than useful risk information (b ¼ 0.12, p 5 0.005), university graduate degree (b ¼ 0.12, p 5 0.005), and disaster history (b ¼ 0.09, p 5 0.005).

Discussion

This study demonstrates that those who report a willingness to evacuate prior to a natural disaster trust the experts, find risk information useful, often have a university graduate degree, and tend to live in areas with a disaster history. Trust in experts appeared as the strongest key determinant. These components are independent and significant contributors to the willingness to follow evacuation instructions. Gender differences and individual differences did not make significant separate contribu- tions. What each individual eventually will do when disaster strikes is impossible to know in advance, but the reported willingness to adhere to instructions is taken as a relevant approximation even though research suggests that behavioural intentions overestimate actual behaviour (Lindell and Perry 2000).

Trust emerges as a key element when people decide whether to accept or reject risk assessments and recommendations by experts, scientists, or authorities (Slovic 1999). This study has documented that the combination of trust in experts and trust in risk information are significant determinants in making people follow evacuation instructions. The learning and reasoning process related to being ready for such adherence may be more important than exploring the most effective evacuation warnings (Dow and Cutter 2000, Trettin and Musham 2000). Even a dreaded ‘false alarm’ may enhance trust between the experts and lay people when the communication is open and honest. ‘People accept that even experts make mistakes sometimes’ (White and Eiser 2006, p. 1200).

The principal rationale for risk communication before, during and after natural hazards is to initiate and direct protective action. Breakwell (2000) claims that risk communication is not a one-way or two-way process, but rather is multi- dimensional. The choice of communication sources through which the risk is interpreted is virtually infinite.

Risk responses may be shaped by psychological, social, institutional and cultural processes that influence risk perceptions, as described by the social amplification of risk framework (Kasperson et al. 1988, Renn 1991, Kasperson 1992, Renn et al. 1992, Pidgeon et al. 2003, Bakir 2005, Hom et al. 2009). This study indicates that individual differences are overridden by trusted relationships with experts and socio- demographic variables. Glik (2007) suggests that situational factors, individual and group level characteristics, and aspects of warning messages all determine whether warnings are heeded. Lay people are likely to be influenced by someone they identify with (Halpern et al. 2004, Dawnay and Shah 2005). Those with higher levels of

Health, Risk & Society 95

education are more likely to have an accurate perception of the risk (Drabek 1986, Faupel et al. 1992, Mileti and Fitzpatrick 1993, Anderson-Berry 2003). The educated are more skilled at acquiring and evaluating information about hazards (Pilisak and Acreddo 1988, Anderson-Berry 2003, Heller et al. 2005).

Living in areas with disaster history

This article has suggested that people who live in areas with a history of natural disasters are inclined to take preventative action; that is to follow evacuation instructions. Loewenstein et al. (2001) argue that emotional reactions to risky situations may drive behaviour, and even dominate over rational analyses. Thus, the emotions related to the past disaster may reinforce protective behaviour that may have been implicitly and tacitly passed on over generations, such as building new homes higher above sea level in the aftermath of the previous disaster. Such behaviour may also have rubbed off on those who merely live in the same area.

Limitations

This is a survey providing a snapshot of a population’s response at the time of the study; it does not show developmental trends or how people will actually behave in a real disaster situation. Nevertheless, the conclusions leave a fair estimate of behaviour that may guide risk communication, but there are limitations on the use of questionnaires; misunderstandings cannot be dealt with on the spot, and certain nuances may go unnoticed.

Even though the response rate (43.6%) seems adequate for this kind of a survey with many of higher age, there are some noticeable issues. The response rates differed between the youngest and older respondents. One possible factor may be media channel preferences. Governmental agencies prioritised traditional written docu- mentation and public meetings which the older age groups are accustomed to, while the younger generation may prefer modern communication platforms such as social media sites (Svoen 2007, Agichtein et al. 2008, Boyd and Ellison 2008). Furthermore, some young adults pursue higher education after high school yet maintain home addresses in the area, but they probably did not receive the questionnaire booklets because they leave their rural homes during the active parts of the academic year. The booklets were distributed in the middle of a semester.

Conclusion

The study suggests a model which contains two independent blocks of determinants of preventative behaviour in relation to a large-scale natural disaster. First and foremost, trusted relationships with experts are essential, but socio-demographics also influence the willingness of the target population to adhere to evacuation instructions, while psychological individual differences do not add anything significant. Because risk is manifested socially, it may prove more useful for governmental agencies to meet people’s needs at both rational and emotional levels, rather than merely conveying facts about risk. One avenue for future research could include further exploration of the effects of risk communication related to emotions, as well as social, psychological and cultural dimensions.

96 S.K. Rød et al.

Acknowledgements

This study has been carried out with the support of Norwegian Public Roads Administration Climate and Transport Project, Norwegian Public Roads Administration Central Region, and the Åknes/Tafjord project.

References

Agichtein, E., Castillo, C., Donato, D., Gionis, A., and Mishne, G., 2008. Finding high- quality content in social media. In Proceedings of the International Conference on Web Search and Web Data Mining, WDSM’08, 2008, 183–194, Stanford, CA.

Anderson-Berry, L.J., 2003. Community vulnerability to tropical cyclones: Cairns, 1996–2000. Natural Hazards, 30, 209–232.

Bakir, V., 2005. Greenpeace v. Shell: Media exploitation and the Social Amplification of Risk Framework (SARF), Journal of Risk Research, 8 (7–8), 678–691.

Boyd, D.M. and Ellison, N.B., 2008. Social network sites: Definition, history, and scholarship. Journal of Computer-Mediated Communication, 13 (1), 210–230.

Breakwell, G.M., 2000. Risk communication: Factors affecting impact. British Medical Bulletin, 56 (1), 110–120.

Briggs, S.R. and Cheek, J.M., 1986. The role of factor analysis in the development and evaluation of personality scales. Journal of Personality, 54, 106–148.

Burton, P.W., Silver, B.D., and Marriott Hotel, P., 2006. The credibility of disaster evacuation warnings: The effects of issue framing and trust in government [online]. Available from: http://35.8.71.4/Documents/SOSSRelated/Burton_Silver_APSA2006_Final_Paper. pdf [Accessed 24 November 2011].

Craig, A.R., Franklin, J.A., and Andrews, G., 1984. A scale to measure locus of control of behaviour. British Journal of Medical Psychology, 57, 173–180.

Dawnay, E. and Shah, H., 2005. Behavioural economics: Seven principles for policy makers. London: New Economics Foundation.

Dow, K. and Cutter, S.L., 2000. Public orders and personal opinions: Household strategies for hurricane risk assessment. Environmental Hazards, 2 (4), 143–155.

Drabek, T., 1986. Human system responses to disasters: An inventory of sociological findings. New York: Springer-Verlag, 509.

Drabek, T.E. and Boggs, K., 1968. Families in disaster: Reactions and relatives. Journal of Marriage and the Family, 30, 443–451.

Endresen, I.M., 1991. A Norwegian translation of the Plutchik questionnaire for psycho- logical defense. Scandinavian Journal of Psychology, 32 (2), 105–113.

Eysenck, S.B.G., Eysenck, H.J., and Barrett, P., 1985. A revised version of the psychoticism scale. Personality and Individual Differences, 6 (1), 21–29.

Faupel, C.E., Kelley, S.P., and Petee, T., 1992. The impact of disaster education on household preparedness for Hurricane Hugo. International Journal of Mass Emergencies and Disasters, 10 (1), 5–24.

Fischhoff, B., 1995. Risk perception and communication unplugged: Twenty years of process. Risk Analysis, 15, 137–145.

Fischhoff, B., 2000. What forecasts (seem to) mean. In: T. Connolly, R.H. Arkes, and K.R. Hammond, eds. Judgment and decision making: an inter-disciplinary reader. 2nd ed. New York: Cambridge University Press, 353–377.

Furseth, A., ed. 1985. Dommedagsfjellet, Tafjord 1934 (2. Utg. A.F. 1994 ed.). Oslo Norway: Gyldendal.

Giddens, A., 1990. The consequence of modernity. Oxford: Oxford Polity Press. Glik, D.C., 2007. Risk communication for public health emergencies. Annual Review of Public

Health, 28, 33–54. Granatt, M., 2004. On trust: Using public information and warning partnerships to support the

community response to an emergency. Journal of Communication Management, 8 (4), 354– 365.

Grothmann, T. and Reusswig, F., 2006. People at risk of flooding: Why some residents take precautionary action while others do not. Natural Hazards, 38 (1), 101–120.

Halpern, D., et al., 2004. Personal responsibility and changing behaviour: The state of knowledge and its implications for public policy. London: Cabinet Office.

Health, Risk & Society 97

Heller, K., et al., 2005. Social and personal factors as predictors of earthquake preparation: The role of support provision, network discussion, negative affect, age, and education. Journal of Applied Social Psychology, 35 (2), 399–422.

Hom, A.G., Plaza, R.M., and Palmén, R., 2009. The framing of risk and implications for policy and governance: The case of EMF. Public Understanding of Science. Sage Publications 1: 1–15.

Jardine, C., Boyd, A., and Furgal, C., 2009. Gender and place influences on health risk perspectives in Northern Canadian aboriginal communities. Gender, Place & Culture: A Journal of Feminist Geography, 16 (2), 201–233.

Kaplan, S. and Garrick, B.J., 1981. On the quantitative definition of risk. Risk Analysis, 1 (1), 11–27.

Kasperson, R.E., 1992. The social amplification of risk: Progress in developing an integrative framework of risk. In: S. Krimsky and D. Golding, eds. Social theories of risk. New York: Praeger, 153–178.

Kasperson, R.E., et al., 1988. The social amplification of risk: A conceptual framework. Risk Analysis, 8 (2), 177–187.

Lemyre, L., et al., 2009. Emergency preparedness for higher risk populations: Psychosocial considerations. Radiation Protection Dosimetry, 134 (3–4), 207–214.

Lindell, M.K. and Perry, R.W., 2000. Household adjustment to earthquake hazard. A review of research. Environment and Behavior, 32 (4), 461–501.

Loewenstein, G.F., et al., 2001. Risk as feelings. Psychological Bulletin, 127 (2), 267–286. Luhmann, N., 1979. Trust and power. Chichester, UK: John Wiley & Sons. Luhmann, N., 1988. Familiarity, confidence, trust: Problems and alternatives. In: D.

Gambetta, ed. Trust: Making and breaking cooperative relations. Oxford: Basil Blackwell, 94–108.

Mertler, C.A. and Vannatta, R.A., 2005. Advanced and multivariate statistical methods: Practical application and interpretation. 3rd ed. Glendale, CA: Pyrczak Publishing.

Mileti, D.S. and Fitzpatrick, C., 1993. The great earthquake experiment: Risk communication and public action. San Francisco: Westview Press, 149.

Pallant, J., 2007. SPSS survival manual: A step by step guide to data analysis using SPSS (3rd ed.). Philadelphia: Open University Press.

Pidgeon, N., Kasperson, R.E., and Slovic, P., eds. 2003. The social amplification of risk. Cambridge: Cambridge University Press.

Pilisak, M. and Acreddo, C., 1988. Fear of technological hazards: One concern or many? Social Behavior, 13 (17), 110–127.

Renn, O., 1991. Risk communication and the social amplification of risk. In: R.E. Kasperson and P.J. Stallen, eds. Communicating risks to the public: International perspectives. Amsterdam and New York: Kluwer Academic, 287–324.

Renn, O., et al., 1992. The social amplification of risk: Theoretical foundations and empirical applications. Journal of Social Issues, 48, 137–160.

Rohrmann, B., 2000. A socio-psychological model for analyzing risk communication processes. Australasian Journal of Disaster and Trauma Studies, 4 (2). Available from: http://www.massey.ac.nz/%7Etrauma/issues/2000-2/rohrmann.htm [Accessed 24 Novem- ber 2011].

Sandman, P.M., 1993. Responding to community outrage: Strategies for effective risk com- munication. Fairfax, VA: American Industrial Hygiene Association.

Sattler, D.N., Kaiser, C.F., and Hittner, J.B., 2000. Disaster preparedness: Relationships among prior experience, personal characteristics, and distress. Journal of Applied Social Psychology, 30 (7), 1396–1420.

Sjöberg, L., 2004. Explaining individual risk perception: The case of nuclear waste. Risk Management: An International Journal, 6 (1), 51–64.

Slovic, P., 1999. Trust, emotion, sex, politics, and science: Surveying the risk-assessment battlefield. Risk Analysis, 19 (4), 689–701.

Slovic, P., 2000. Perceived risk, trust, and democracy. In: T. Connolly, R.H. Arkes, and K.R. Hammond, eds. Judgment and decision making: An inter- disciplinary reader. 2nd ed. New York: Cambridge University Press, 500–513.

Slovic, P., et al., 2004. Risk as analysis and risk as feelings: Some thoughts about affect, reason, risk, and rationality. An International Journal, 24 (2), 311–322.

98 S.K. Rød et al.

Slovic, P., et al., 2005. Affect, risk, and decision making. Health Psychology, 24, S35–S40. Svoen, B., 2007. Consumers, participants, and creators: Young people’s diverse use of

television and new media. Computers in Entertainment (CIE), 5 (2), 5. Tabachnick, B.G. and Fidell, L.S., 2007. Using multivariate statistics. Boston: Pearson/Allyn

and Bacon. Thalmann, A. and Wiedemann, P., 2006. Belief and emotionality in risk appraisals. Journal of

Risk Research, 9 (5), 453–466. Trettin, L. and Musham, C., 2000. Is trust a realistic goal of environmental risk communica-

tion? Environment and Behavior, 32 (3), 410–426. U.S. Department of Homeland Security, 2004. National response plan [online]. Available

from: http://www.dhs.gov.xlibrary/assets/NRPbaseplan.pdf [10 July 2009]. Weisæth, L., 2002. The European history of psychotraumatology. Journal of Traumatic Stress,

15 (6), 443–452. West, D., 2001. The rise and fall of the media establishment. Boston: Bedford/St. Martin’s

Press. West, D.M. and Orr, M., 2007. Race, gender, and communications in natural disasters. Policy

Studies Journal, 35 (4), 569–586. White, M.P. and Eiser, J.R., 2006. Marginal trust in risk managers: Building and losing trust

following decisions under uncertainty. Risk Analysis, 26 (5), 1187–1203.

Health, Risk & Society 99

Copyright of Health, Risk & Society is the property of Routledge and its content may not be copied or emailed

to multiple sites or posted to a listserv without the copyright holder's express written permission. However,

users may print, download, or email articles for individual use.