Community and psychological barriers to tsunami preparation

Tsunami risk was investigated as a platform for helping to understand the impact of community and psychological factors in natural disaster preparedness. Residents who lived within four meters of sea level in high and moderate tsunami risk areas of Vancouver Island were queried about possible influences on their preparation. Both community-level and psychological factors played significant roles. More community participation and fewer justifications for not preparing (the dragons of inaction) directly predicted reported preparation. The relation between sense of community and preparation was mediated by the level of community participation, and the relation between positive outcome expectancy and preparation was mediated by the dragons of inaction. Together, the resulting model accounted for 21% of the variance in reported tsunami preparation. Policy implications are discussed.


Introduction
Eighty percent of tsunamis occur around the Pacific Ocean "ring of fire, "mainly as a result of shifts in the earth's tectonic plates. Although the tsunami risk at Pacific-wide and regional scales is relatively low, they have a great destructive potential. The last historical tsunami to cause significant damage on the west coast of North America occurred on March 27, 1964, from a great (magnitude 9.2) earthquake in southern Alaska. It was the strongest earthquake recorded in North American history, and the second-strongest recorded in world history (Clague et al. 2000;Clague and Orwin 2005).
The destructive effects of a local tsunami generated from a large earthquake along the Cascadia subduction zone should not be overlooked (Fig. 1). In the event of a tsunami, very little can often be done to provide warnings for the coastal areas because of their often short travel time. This highlights the importance of effective education initiatives to improve residents' level of tsunami preparedness.
Preparation is the most important phase of the emergency management cycle. Gaining knowledge about typical householders' levels of preparedness, and the factors that influence it, are crucial for emergency management organizations. Obviously, greater levels of awareness and preparedness before the occurrence of a natural hazard will result in significant reductions of human and financial losses. However, despite the existence of numerous education programs, levels of residents' participation and disaster preparedness remain low (Ballantyne et al. 2000;Lindell and Hwang 2008;Paton 2006;Paton et al. , 2006a).
An important reason that these programs have not been more effective is that they have not been sufficiently grounded in resident attitudes and behaviors toward natural disasters. The ineffectiveness of traditional public education programs that are primarily focused on disseminating hazard information through websites and pamphlets has been identified in many studies (Lindell and Whitney 2000;Paton et al. 2000Paton et al. , 2006b. We are unaware of any published research on tsunami preparedness in southwestern British Columbia, and therefore, one purpose of the present study was to understand tsunami preparedness in that region so as to assist in regional preparedness planning, which of course may well be useful in other tsunami-threatened regions around the globe.  (Clague and Orwin 2005) However, at a more theoretical, the main objective of this study was to examine the roles that community-and psychological-level factors might play in a useful disaster preparedness model. To do so, the present study began with Paton et al. (2008a) model, but considered possible enhancements to it, so as to better understand who prepares, who does not, and why. Besides providing useful guidance to organizations mandated with disaster emergency management, an expanded model should also aid in the organizing and implementation of programs aimed at increasing community participation and encouraging residents to engage in more behavioral risk mitigation.

Literature review
Several broader theories of attitude-behavior relations, such as social cognitive theory (SCT; Bandura 1986;McIvor et al. 2009;, the protective action decision model (PADM; Lindell and Perry 2012), protective motivation theory (PMT ;Bandura 1977;Rogers 1975Rogers , 1983Rogers and Prentice-Dunn 1997;Paton and McClure 2013), the health belief model (HBM ;Becker 1974;Rosenstock 1974;Akompab et al. 2013), the person-relative-to-event (PrE) model (Duval and Duval 1985;Duval and Mulilis 1989), the theory of planned behavior (TPB; Ajzen 1985), and critical awareness theory  have been applied to predict natural hazard preparedness. These theories have investigated individual-level factors such as coping, self-efficacy, response efficacy and control beliefs to explain decision-making and behavior under various conditions of risk and uncertainty in order to predict both individuals' intentions to prepare and actual preparedness. Several studies (e.g., Duval and Mulilis 1999;Lindell and Whitney 2000;Perry and Lindell 2008) have also identified the role played by intra-individual factors on levels of preparedness.
One early comprehensive social-cognitive model of natural hazard preparedness was proposed by Paton (2003), who based it on studies of the adoption of health risk behaviors (e.g., Abraham et al. 1998;Bennett and Murphy 1997;Duval and Mulilis 1999). The model includes three phases, each influenced by a specific set of variables. The first contains antecedent factors that motivate individuals including critical risk awareness, risk perception and hazard anxiety. The second includes variables that link the initial motivation to intention formation, such as outcome expectancy, self-efficacy, response efficacy, and problem-focused coping. The third focuses on the relation between intentions and preparation, including perceived responsibility, time of hazard activity, sense of community, response efficacy, and normative factors such as trust and empowerment. Johnston et al. (2005) adopted this 2003 model to examine tsunami preparedness in coastal Washington State and found that preparedness levels were low-to-moderate despite reasonable success in disseminating hazard information. The authors conclude that different strategies should be used for each of the three stages of the model: motivate individuals to prepare (that is, precursor variables), facilitate the formation of intentions (that is, intention formation variables), and influence the conversion of intentions into preparedness.
Other studies have focused on the role of community context factors as predictors of natural disaster preparedness (e.g., Lindell and Perry 2004;Paton 2013;Paton et al. , 2008a). One such model (Paton et al. 2008b) describes how beliefs and their social context influence levels of hazard preparedness. In it, individuals' interpretations under uncertainty in natural hazards commence with their beliefs about the efficacy of protective actions and interact with 1 3 social context factors including community participation, collective efficacy, empowerment and trust to influence levels of tsunami preparedness.
In sum, the present study adopted Paton et al.'s (2008a) model as a starting point because it is most clearly relevant to our objectives, but it includes, in particular, psychological barriers to action as a potentially important influence. The variables in the present study's model are described next.

Sense of community
The sense of belonging to one's community and attachment for local individuals and places, sometimes called community bondedness (e.g., Turner 1983;Turner et al. 1986), can be an important factor (e.g., Bishop et al. 2000;Carver et al. 1989;Duval and Mulilis 1999). Few studies have investigated the influence of sense of community on community participation, although their connection appears to be relatively strong (e.g., Paton et al. 2008a). This connection might be expected to strengthen the relation between public participation and actual preparation.

Community participation
Individuals tend to share their interests, values and expectations with others when they are dealing with uncertain events and when they lack information they need (Earle 2004;Lion et al. 2002;Paton and Bishop 1996;Poortinga and Pidgeon 2004). Involvement in community activities provides individuals with a broader range of information and collective knowledge from other community members while sharing their interests, values and expectations with others.
Some studies report a significant relation between community participation and preparation (Paton et al. 2008b, which supports the hypothesis that interacting with other community members fosters preparation.

Outcome expectancy
Outcome expectancy is an intra-individual variable that concerns how cost and benefit beliefs influence levels of preparedness when one is faced with complex and uncertain events (Bennett and Murphy 1997). For individuals who hold negative outcome expectancy (NOE) beliefs, personal actions would not make any difference to safety because of the catastrophic effects of tsunamis, so that are not expected to take precautionary actions (Paton et al. 2008a(Paton et al. , b, 2009); Paton and Tedim 2014).
In contrast, those who believe that preparing for a disaster can have beneficial consequences are expected to be more likely to prepare (Paton et al. 2008a, b). In addition, some studies have supported the positive role of POE on preparation (Paton et al. 2008a, b).

Psychological barriers: the dragons of inaction
A theory that has been employed very frequently to explain individuals' behavior choices, the theory of planned behavior (Ajzen 1985) simply posits that intentions lead to actions. However, many studies report that the correlation between intention and action is weak. In the present context, preparation for hazards often does not occur, or is inadequate, even with good intentions to do so.
In order to account for the discrepancy between intentions and action, Gifford (2011) proposed several dozen "dragons of inaction", that is, psychological obstacles such as rationalizations and justifications that individuals often use to justify one's gap between intentions and actions. These "dragons" have successfully helped to explain the gap between intentions and actions for climate-positive food choices (Gifford and Chen 2017) and for energy conservation actions (Lacroix and Gifford 2018). For that reason, these psychological barriers to good intentions were considered to be potentially useful for explaining hindrances to residents' intentions to prepare for a possible tsunami.
Evidence from research on other disasters supports the possibility that residents use justifications for not turning their intentions into action. For example, the relation between negative outcome expectancy beliefs and bushfire preparation was mediated by "preparation inhibitors" (e.g., not prepared to work with others, time, financial; Paton et al. 2008a). The notion of investigating barriers to action has been discussed by some authors (e.g., Jackson and Mukerjee 1974;Mulilis et al. 1990;Paton 2019). However, to the best of our knowledge, the role played by psychological barriers in explaining the gap between intentions to prepare for disaster and preparedness actions has not yet been actually investigated.

The present study
One objective of this study was to enhance knowledge about at-risk residents' tsunami preparation by investigating the influence of community-based antecedent factors on resident preparedness. A second objective was to examine the usefulness of the dragons of inaction, which are hypothesized to mediate the effects of antecedent influences on preparation and residents' levels of preparedness. This is a proposed enhancement to Paton et al.'s (2008a) model (Fig. 2). The findings should be useful to civic agencies that are charged with encouraging threatened residents to prepare for natural hazards and assist understanding of disaster preparation at the theory level.
The present model includes two community-level factors: sense of community and community participation. Sense of community is hypothesized to have a direct and positive influence on community participation. In addition, in accordance with Paton et al.'s 2008a) Dragons of inaction Sense of community Community participation Preparation Positive outcome expectancy Fig. 2 The hypothesized model, as adapted from Paton et al. (2008a) model, residents with a greater sense of community were expected to have greater levels of preparedness. Community participation was also expected to lead to more preparation.
In addition, the model proposes that preparedness is the outcome of a process that commences with a personal-level factor, positive outcome expectancy. Individuals who hold positive outcome expectancy beliefs are predicted to engage in more preparation.
Finally, the dragons of inaction are hypothesized to negatively predict reported preparation (more rationalizations lead to fewer preparations), but also to mediate the relation between positive outcome expectancy beliefs, sense of community, community participation and preparation. Although we are unaware of any evidence that dragons of inaction mediate their relation, a similar construct, preparation inhibitors, mediated the relation between positive outcome expectancy and bushfire preparation (Paton et al. 2008a). Reported preparedness is the ultimate element in the model, that is, as the endogenous (dependent) latent variable. Apart from the model, the study also considered the role of ten demographic variables (age, gender, education, religion, ethnic origin, marital status, having children or not, household type, car and home ownership) on preparation and its predictors. Several earlier studies in very large samples have concluded that demographic characteristics have small but statistically significant correlations with preparedness levels; that is, that demographics have small effect sizes (Baker 1991;Jackson 1977Jackson , 1981Lindell 2013). Therefore, we expected that most of the demographic characteristics would be unrelated or only weakly correlated with preparation and its antecedents. These demographic variables and their codings are shown in Table 1.  (1), High school (2), Diploma/college (3), Bachelor (4), Postgraduate or professional (5) Ethnic origin European (1), Hispanic or Latino (2), Black or African origin (3), First nations (4), Asian (South) (5), Middle-eastern (6), Other (7)  Religion Have a religion (1), No religion (2) Marital status Not married (1), married or domestic partnership (2) Household type Single-person household (1), group household (i.e., shared), (2), couple with no children (3), couple with children (4), single-parent family (5) House ownership Owner (1), Tenant (2) Age of building -Does someone in your home own a car?

Participants
Adult residents who lived in high-and moderate tsunami risk areas (i.e., a residence within four vertical meters of sea level) of southern Vancouver Island were invited to participate. About 40% (n = 101) of those approached agreed to complete a survey online, of which 55 were female and 46 were male.  were First Nations, 4% were Asian, 7.9% stated "Other," and 10% did not state a heritage. Almost none reported having experienced a tsunami.

Procedure and materials
Volunteers went door-to-door to residences that were within 4 m of mean high tide and asked residents over 18 years old to complete the survey online by giving them a card with the survey link printed on it. The survey included questions that were derived from Paton et al.'s (2008a) study and the dragons of inaction (Gifford 2011). The scales and their scoring ranges are seen in Tables 2, 3 and 4. The survey had five sections. The first included questions about positive outcome expectancy beliefs (e.g., "Preparing for tsunamis will significantly reduce damage to my home, should a tsunami occur" on a 7-point agreement scale). The second contained questions about sense of community (e.g., "I feel like I belong in this community" on a 7-point agreement scale). The third queried community participation (e.g., "I participate in local activities or events (e.g., festivals, fetes, fairs)," on a 7-point agreement scale. The fourth included the dragons of inaction items (e.g., "It's too difficult for me to prepare for tsunamis" on a 7-point agreement scale. The fifth was about preparations (e.g., "Does your household have an emergency plan for a tsunami?" (no or yes). A sixth section was about the resident's intention to prepare, which was later discarded. 1 The final section contained the demographic questions. The analyses were conducted using the structural equation modeling (SEM, using AMOS 24). It was chosen because it allows the simultaneous estimation of multiple and interrelated dependence relations (Byrne 2001;Cheng 2001;Kline 1998)  1 Intentions to take future precautionary actions were excluded at the analysis stage because they were asked after the preparation questions in the survey. Thus, the responses to the intention questions presumably were differentially dependent on what the householder had already done-that is, no preparations, or some, or many-which rendered future intentions, as a variable, too muddied to be useful as a clear measure.

3
SEM can test the model as a whole to determine how accurately a given model reflects the data, using several goodness-of-fit indices.

Means, standard deviations, and scale reliabilities
The respondents reported middling levels of positive outcome expectancy (m = 4.15, sd = 1.13) and slightly higher levels of community participation (m = 4.84, sd = 1.20) and sense of community (m = 5.25, sd = 0.96). Their mean responses were relatively low for the dragons of inaction (m = 2.89, sd = 0.82). They reported having undertaken slightly less than half of the possible preparations (m = 10.19 out of 24 items; sd = 5.08). The multi-item scales had Cronbach alpha reliabilities ranging from α = 0.69 to 0.88 (see Table 5 for details).

Correlations among key variables
The more individuals believed that preparing for tsunamis would significantly reduce damages and improve their levels of preparedness, the fewer obstacles they perceived (r = − 0.37, p < 0.001). Community participation and individual's level of preparation were significantly related (r = 0.34, p < 0.001). Residents who were less prepared for a possible tsunami perceived fewer preparedness impediments (r = − 0.22, p < 0.01). Those who were involved in community activities reported a greater sense of community (r = 0.41, p < 0.001) ( Table 6).

Modeling the predictors of preparation
A few changes were made to the original model because some of the hypothesized connections in it were not significant and therefore were excluded from the analyses. The revised model fit to the data well (Fig. 3). Its goodness-of-fit statistics were: χ 2 = 3.5, df = 5, p = 0.62, RMSEA = 0.00 < 0.05, NFI = 0.95 > 0.95, CFI = 1.00 > 0.95, GFI = 0.98 > 0.90, and AGFI = 0.95. The model accounted for 21% of the variance in reported tsunami preparation. In particular, the hypothesized direct relations between positive outcome expectancy and sense of community with reported preparation were not confirmed. However, the analyses supported the hypothesis that sense of community directly and significantly influences community participation (β = 0.41, p < 0.001) and that community participation then directly and positively predicts the level of preparation (β = 0.37, p < 0.001). Finally, the hypothesis that the relation between positive outcome expectancy and level of preparation would be mediated by the dragons of inaction was supported (β = − 0.38, p < 0.001), as was the hypothesis that the dragons of inaction would then (negatively) directly predict the level of preparation (β = − 27, p < 0.001).

Demographic influences
Residents' level of preparation and its predictors were examined in relation to the ten demographic variables. However, as predicted, not many were significant. Females did report more positive outcome expectancy beliefs than males (β = 0.36, p < 0.001) and older individuals reported participating in more community activities (β = 0.23, p < 0.04). Owning a car was negatively correlated with the level of preparedness (β = − 0.28, p < 0.01). Tenants were less likely to be involved in community activities than homeowners (β = − 0.43, p < 0.01) and reported a weaker sense of community (β = − 0.25, p < 0.01) and fewer preparations (β = − 0.44, p < 0.01). Those who Fig. 3 The structural equation model for the prediction of preparation adhered to (any) religion reported a greater sense of community than those who did not (β = − 0.22, p < 0.03). Finally, married respondents and those who had a domestic partner had completed more preparations (β = 0.24, p < 0.01). Because over three-quarters of the sample were of European origin, the relation of heritage to other variables was not pursued because too few residents came from other heritages.

Discussion
In a tsunami-threatened region, an enhanced version of an existing model for predicting disaster preparation that includes psychological barriers was proposed, tested, and supported. Overall, the fit indices for the model were very good. The present model demonstrates that community-level factors play a vital role, both directly and indirectly, in residents' reported level of preparedness. However, the individual-level variable, positive outcome expectancy, was not directly related to preparation, which differs from the results of some earlier studies (Jang et al. 2016;Paton et al. 2008aPaton et al. , b, c, 2010Sagala et al. 2009;Ejeta et al. 2018) although it was indirectly related to preparation via the dragons of inaction. Stronger positive outcome expectancy beliefs led to perceiving fewer dragons of inaction, and the perception of fewer dragons led to undertaking more preparations. In sum, individual-level factors operated through a mediator but not direct influences on preparation.
The present results provide support for the hypothesis that residents with a stronger sense of community and who are more involved in community activities are more likely to prepare for a possible natural disaster. In accordance with the results of Paton et al.'s (2008a) model, this relation was relatively strong. Sense of community also appears to lead to more community participation, which in turn was the most important predictor of preparation in this study. This confirms the idea that attitudes toward environmental risk mitigation are heavily influenced by social factors (Lion et al. 2002;McGee and Russell 2003;Earle 2004). These findings aligned with the results of studies (e.g., Said et al. 2011) suggest that making plans based on a community-based approach in order to mitigate tsunamis is viable.
The role of ten demographic variables was examined, and relatively few significant correlations were found, which (a) confirms the results of earlier studies that the influence of demographic characteristics is moderately or weakly correlated with reported preparation Lindell 2013) and (b) suggests that the present model has good generalizability.
That being said, some demographic findings are worth noting for theoretical and practical policy reasons. Males reported weaker positive outcome expectancy beliefs and reported fewer preparations. Younger individuals participate less in their community; this might lead younger residents toward lower levels of preparedness.
House and car ownership play a role. Not having a car was associated with lower levels of preparation; presumably this is a marker of some other more direct influence such as lower income, rather than as a cause. Tenants, perhaps not surprisingly, were less likely to be involved in community activities and had a weaker sense of community, but more importantly they had completed fewer preparations. Individuals who were married or were living with their partners reported greater levels of preparedness than unmarried ones. Being part of a religion was associated with greater preparation. In sum, efforts to increase risk-reduction behaviors might well focus on males, younger residents, tenants, single persons, and residents with no religious affiliation.
In order to facilitate at-risk residents' preparedness, strategies that emphasize strengthening community ties and removing psychological barriers to residents' preparations should be considered by emergency management organizations. First, sense of community might be enhanced by encouraging at-risk residents to volunteer in community activities and events. This should increase their level of preparedness indirectly.
Second, public education strategies that encourage at-risk community members to discuss their risk from tsunamis, share their emergency plans, and talk about how to reduce tsunami impacts could be adopted by local emergency planners.
Third, planners in tsunami-threatened communities could gather community members together to educate them in skills such as tsunami evacuation drills, survival techniques in the face of a tsunami, and first aid. Lack of knowledge and required skills were two obstacles that hindered people from preparations. At the same time, these efforts should be wary of frightening residents.
Fourth, programs designed to facilitate preparedness by eliminating the most frequently used dragons of inaction (Gifford 2011) should be targeted by emergency management agencies. Although the respondents did not report very large levels of these justifications, managers might pay most attention to those most often employed: "I'm satisfied with my current level of preparedness" (m = 3.79/7), "I'm content with the extent to which my current preparation level reflects who I am as a person" (m = 3.65/7), and "I have already prepared enough for a tsunami" (m = 3.37/7). Thus, participants in the present study were, on average, relatively content with their current levels of preparedness. However, the correlations between these statements and their own reported preparation were not impressive: The correlation between an index of these three measures and reported preparation was 0.35, so about 12% of the variance in preparation is accounted for by "I'm fine, thanks." Contentedness with preparation is a serious challenge for emergency management officials to overcome.
Apart from the usual limitation that the study was conducted in a single region, albeit a threatened one, its inclusion of the dragons of inaction might usefully be investigated as part of models other than the one with which the present study began (Paton et al. 2008a). Second, again as with most other studies, discrepancies between reported and observed levels of preparedness Ejeta et al. 2018) must be considered. Nevertheless, the present study helps to highlight the importance for disaster preparation of community factors and the dragons of inaction. Third, the findings that having a weaker sense of community and being less inclined to participate in community activities by tenants are limited since the duration of residence in their home and in the community was not queried. This should be considered in future studies.
An underestimation of the effects of positive outcome expectancy and implementation barriers on preparedness could have been caused by only selecting respondents only from within the tsunami inundation zone as opposed to a sample comprised of those who reside inside as well as just outside the tsunami inundation zone (see Doyle et al. 2018).
The current study has confirmed the role of community-level variables for tsunami preparation and has demonstrated for the first time the importance of the dragons of inaction as a mediator in preparation models. The results should assist policy-makers as they grapple with engaging communities that are threatened with potential tsunamis.