Household preparedness motivation in lahar hazard zones: assessing the adoption of preparedness behaviors among laypeople and response professionals in communities downstream from Mount Baker and Glacier Peak (USA) volcanoes

Theoretical framework and research questions

In PMT, an individual’s threat appraisal (i.e., risk perception; Grothmann and Reusswig 2006) and coping appraisal determine the motivation and direction of protective actions. Threat appraisal incorporates perceived probability of exposure, perceived severity of damage, and fear of the hazard. Coping appraisal incorporates perceived protective response efficacy (i.e., judgments of the effectiveness of preparedness actions for addressing the threat), perceived self-efficacy (i.e., belief in one’s ability to act), and perceived protective response costs (i.e., perceptions of the cost associated with protective actions in terms of money, time, knowledge, and effort). Grothmann and Reusswig (2006) demonstrate that one’s coping appraisal influences preparedness behaviors to a far greater degree than one’s threat appraisal. However, Grothmann and Reusswig do not distinguish between the role of each of the three components that make up the coping appraisal.

More detailed investigations of some of these individual components are found in studies applying PADM (e.g., Houts et al. 1984; Lindell and Perry 1992, 2012; Lindell and Whitney 2000; Lindell and Prater 2002; Terpstra and Lindell 2012). PADM explains that preparedness intentions and behaviors depend on hazard-related and resource-related attributes. Hazard-related attributes represent a more detailed measure of perceived response-efficacy that includes efficacy for protecting people, efficacy for protecting property, and utility of the adjustment for other purposes. Response-related attributes measure characteristics of protective actions rather than characteristics of the people responding. PADM’s response-related attributes focus on the perceived amount of money, time, effort, knowledge, and skills required for implementing a protective action. The perceptions of money and time expenditures are similar to PMT’s concept of perceived protective response costs. Yet, differences exist in how PADM and PMT are conceptualized and operationalized in different studies, and PADM lacks the perceived self-efficacy concept found in PMT.

In VBN theory, Stern and colleagues (Stern et al.1999; Stern 2000) propose that support for environmental movements emanates from one’s personal values, a worldview consistent with the New Ecological Paradigm (i.e., an ecological worldview centered on the relationship between humans and the natural world), an awareness of consequences, an ascription of responsibility to self, and personal norms (Stern et al. 1999; Stern 2000; Slimak and Dietz 2006). In short, individuals hold certain values, beliefs, and worldviews and recognize when they are threatened. Aware of negative consequences and believing themselves personally responsible for protecting their values, individuals feel obliged to take action.

The relationship between ascription of responsibility and behavior motivation was also investigated in the context of natural hazard preparedness by Lindell and Whitney (2000) and Arlikatti et al. (2007). Both studies examine the level of protective responsibility that respondents attribute to various stakeholders (e.g., federal, state, and local government officials, the media, employers, university officials, friends/peers, family, and self) with regard to seismic hazards. In both studies, respondents rate themselves as most responsible for their personal safety, with government officials considered the next most responsible, and the media, one’s peers, and one’s friends rated as least responsible. Lindell and Whitney further note that positive correlations exist between personal responsibility and preparedness intentions and actions. Taking into account these findings, Lindell and Perry’s (2012) theory modification incorporates attribution of protective responsibility (i.e., ascription of responsibility) as part of PADM’s perceptions of social stakeholders. PMT, however, currently lacks a concept equivalent to ascription of responsibility.

To improve the adoption of preparedness behaviors, previous studies (Barberi et al. 2008; Paton et al. 2008; Wachinger et al. 2013) have advocated for increasing public participation in the hazard management process. By more frequently interacting with emergency officials, the public improves their knowledge of local hazards and how to prepare. They gain an appreciation for the role of emergency agencies during hazard responses, learning what external support to reasonably expect and when to rely on their own agency. People reclaim responsibility for their personal safety rather than placing this responsibility in the hands of emergency services. This recognition of personal responsibility, coupled with elevated self-efficacy, helps motivate preparedness actions (Paton 2003; Wachinger et al. 2013). Interactions with officials also strengthen individual and community trust in officials, which, combined with an understanding of the role of emergency agencies, fosters a setting in which individuals heed emergency information and warnings (Wachinger et al. 2013). As the public feels increasingly knowledgeable, empowered, and trusting, they become more motivated to adopt preparedness actions.

Past studies (Barberi et al. 2008; Paton et al. 2008; Wachinger et al. 2013) focus almost exclusively on the impact of increased participation by members of the general public. However, participation takes many forms, from public responder programs (e.g., CERT) to public engagement in meetings and discussions with emergency officials. Additionally, the public are not the only community members that participate in hazard management and influence whole community preparedness. Response professionals (i.e., individuals who work as first responders or in a leadership role within the local city government, hospitals, school districts, Red Cross, or utilities, transportation, or water companies) actively participate in hazard response planning and implementation. Thus, one might expect that the demonstrated positive effects experienced by members of the public who participate in hazard management would be equally or more considerably felt by response professionals. However, few studies have examined the influence of professional participation on preparedness behaviors in a similar fashion to studies of public participation.

Recent research on response professionals deals largely with organizational preparedness and professional competencies (i.e., whether or not an individual has the knowledge, skills, and abilities required to perform their professional response duties) with a focus on health care professionals (Parker et al. 2005; Slepski 2007). Those few studies that examine household preparedness levels among public health employees (Blessman et al. 2007; Rebmann et al. 2013) and first responders (Federal Emergency Management Agency n.d.) consistently indicate that household preparedness among respondents remains low. Yet, these studies fail to examine public household preparedness levels for comparison. As such, the comparative influence of hazard management participation at a professional level on household preparedness, knowledge, self-efficacy, personal responsibility beliefs, and trust remains unclear.

Hazard and site description

This study focuses on volcanic lahar hazards from Mount Baker and Glacier Peak in the Skagit Valley of northwestern Washington state. Commonly referred to as volcanic mudflows (Waitt et al. 1995), lahars are mixtures of water and debris that originate on a volcano and flow downslope under the influence of gravity (Vallance 2000; Volcano Hazards Program U.S. Geological Survey 2016). Lahars can be triggered during volcanic eruptions when hot, freshly erupted material melts and mixes with glacial water or a portion of the volcanic edifice collapses. Lahars can also occur post-eruption or form independent of an eruption (Rodolfo 2000; Vallance 2000). These secondary lahars typically result from non-volcanic earthquakes that trigger edifice collapses or intense rainfall and lake breakouts that remobilize loose pyroclastic deposits.

After initiation, lahars move downslope, typically along river drainages, and bulk up by eroding and incorporating surrounding material (Rodolfo 2000; Vallance 2000). Dense, cohesive lahars with high carrying capacities entrain large boulders and debris (Pierson and Scott 1985; Vallance 2000). More dilute, non-cohesive lahars allow large debris to settle out while smaller, more buoyant particles remain entrained (Pierson and Scott 1985; Vallance 2000). Both cohesive and non-cohesive lahars can cause extensive damage to the built environment as boulders destroy structures and mud floods into and buries communities. Their speed, which ranges from a few meters per second to several tens of meters per second (Volcano Hazards Program U.S. Geological Survey 2016), exacerbates the potential for damage.

The Skagit Valley was selected for this study because populated communities intersect substantially with lahar hazard zones from two volcanoes (Fig. 1), both of which exhibit conditions and eruptive histories that favor lahar generation. Mount Baker is the second most glaciated volcano in the Cascades after Mount Rainier (Gardner et al. 1995), and Glacier Peak is the second most explosive after Mount St. Helens (Waitt et al. 1995). Extensive glaciation and available pyroclastic material leaves each prone to lahars. In fact, geologic assessments indicate multiple episodes of eruptive activity and lahar generation at both volcanoes during the past 14,000 years (Hyde and Crandell 1978; Beget 1982, 1983; Gardner et al. 1995; Waitt et al. 1995; Diefenbach et al. 2015). The largest of these episodes included lahars that traveled over 100 km downstream to Puget Sound and into now populated areas (Hyde and Crandell 1978; Beget 1982; Gardner et al. 1995; Dragovich and McKay 2000; Kovanen et al. 2001).

Seven towns (Mount Vernon, Sedro-Woolley, Burlington, La Conner, Concrete, Hamilton, and Lyman) lie either partially or fully within the lahar zones for both Mount Baker and Glacier Peak (Fig. 1). Risk mapping using geographic information systems (GIS) reveals that nearly 40,000 lives and 15,000 homes are at risk in these lahar zones (Corwin 2016). If the entire lahar zone were inundated, hundreds of square kilometers of agricultural land could be rendered useless. Monetary losses from property alone could escalate to over $5 billion based on the value of individual parcels (Skagit County Digital Data Warehouse 2014; Corwin 2016). Along with immediate financial impacts, the subsequent loss of nearly $62 million dollars in tax revenue to the county could present further challenges for a recovering community. Major transportation routes, such as Interstate 5 (north-south) and the North Cascades Highway (east-west), intersect the lahar zone and damage to these networks could negatively impact response capacity. This potential for loss and the prospect of improving hazard management and community preparedness represent the primary motivation behind this study.

Measuring household preparedness

Survey participants were shown a list of commonly recommended household preparedness measures and asked to indicate which of the six activities they had undertaken and which of the fourteen items they had prepared. These activities and items are consistent with those recommended in the widely used Kit-Plan-Inform messaging framework and can be divided into three categories: (1) gathering supplies, (2) making a plan, and (3) seeking information. National, local, and non-governmental organizations promoting this framework include the Department of Homeland Security’s Ready.gov, Washington State Emergency Management Division, Skagit County Department of Emergency Management, American Red Cross, and Cascade Volcano Observatory.

The checklist of preparedness measures on the questionnaire included fourteen supply items (e.g., food and water for 3 days, flashlight and radio with extra batteries, blankets), two planning activities (e.g., establishing a plan for contacting family members, designating an out-of-area emergency contact), and four information seeking actions (e.g., seeking information on local volcanic hazards, learning first aid, learning who in the community may need additional help). These measures were grouped into a typology of hazard adjustments based on the Kit-Plan-Inform categories, which we refer to as the supplies, planning, and action indicators hereafter.

The number of activities and items participants checked off on the provided list was tallied to generate a raw score for each indicator. These raw scores were standardized to their corresponding z-scores by subtracting the indicator’s mean from each individual score and dividing by the indicator’s standard deviation. The resulting scores exhibit a mean of zero and a standard deviation of one. Finally, the three standardized indicator scores were combined and averaged into an equally weighted linear composite score for preparedness (hereafter CP score). The average CP score is zero, positive scores indicate above average preparedness, and negative scores indicate below average preparedness.

Measuring preparedness with an equally weighted linear composite score rather than a raw count of activities and items substantially increases the importance of the planning and action indicators. Although these two indicators consist of far fewer measures than the supplies indicator, with the CP score, each indicator accounts for a third of the participant’s total preparedness. We assume that planning and information seeking actions are equally as important as gathering supplies and designed the CP score to reflect this assumption. Overall, the CP score provides (1) a continuous variable for measuring household preparedness, (2) a means for comparing household preparedness across individuals and groups, and (3) a measure that places less emphasis on individual supplies and more on planning and information seeking actions.

Data analysis

A combination of data types were collected including nominal, ordinal, and continuous. We relied extensively on 5-point Likert-type questions that, with some variations, asked participants to rate their agreement with statements on a scale of 1 (strongly disagree) to 5 (strongly agree). Where necessary, an “I don’t know” category was included.

Data analysis was conducted using R version 3.3.2 (R Core Team 2016) and standard statistical tests. To test for independence between two variables based on observed response frequencies, we use chi-square tests. For Likert-type responses, the 1 and 2 rankings as well as the 4 and 5 rankings were combined to ensure minimum expected values greater than five in chi-square tests. T-tests were used to test for differences between the means of two groups. The Spearman’s rho (ρ) correlation coefficient was used to measure association between two variables because at least one variable was ordinal in each comparison.

Barriers to preparedness behaviors

Ratings for all perceived self-efficacy statements correlate positively with CP scores (Table 4), meaning those who rate their knowledge, skills, and abilities higher also tend to be better prepared. This is true across all indicator variables. Additionally, correlations between preparedness and self-efficacy are stronger when the self-efficacy statement refers to preparedness for natural hazards in general (first statement in Table 4) rather than response and recovery activities for lahars specifically. In fact, the correlation between preparedness and the first self-efficacy statement represents the strongest correlation recorded in the present study. No significant difference is found between preparedness and perceived self-efficacy when considering a frequently occurring hazard (flooding) as opposed to a rarer hazard (lahars) given the overlap in the 95^th percentile confidence intervals.

Professional participation’s influence on household preparedness & personal beliefs

Previous studies indicate that increased public participation in hazard management should positively influence preparedness levels, knowledge, trust in officials, ascription of responsibility, and perceived self-efficacy (Paton et al. 2008; Wachinger et al. 2013). To test whether or not these specific advantages also apply in the case of professional participation in hazard management, we divide the survey participants into two populations, response professionals and laypeople, and compare their responses on these five topics. Response professionals are 73 individuals who self-identify as first responders or leaders in local city government, hospitals, school districts, Red Cross, or utilities, transportation, or water companies. These individuals participate in hazard management at a professional level. The term laypeople refers to the other 383 survey respondents from the public.

Confidence in officials

Response professionals express greater confidence in officials’ abilities to provide timely and effective instructions, response, and evacuation during a hazard than do laypeople (Fig. 2). Nearly half of response professionals are confident in the ability of officials to respond to a hazardous event successfully compared to only a quarter of laypeople. Additionally, less than a third of response professionals lack confidence in the abilities of officials compared to 44% of laypeople.

Ascription of responsibility

Statistically, both response professionals and laypeople assign responsibility for their personal safety and provision of resources in a similar manner (Fig. 3). Respondents in both groups state they, themselves, are most responsible for their personal safety during a natural hazard. They consider local emergency services as well as friends and family members the next most responsible followed by other community members and FEMA. Ascription of responsibility beliefs do not appear to change with increased professional participation.

Perceived self-efficacy

Significantly more response professionals agree with each self-efficacy statement than do laypeople (Fig. 4). This trend holds regardless of if the statement refers to a frequent hazard (flooding) or a rare hazard (lahars). However, the increase is more pronounced when considering flooding. On average, agreement with self-efficacy statements increases by 21% (range: 15–24%) when respondents identify as response professionals.

Both response professionals and laypeople indicate lower levels of perceived self-efficacy when considering lahars as opposed to floods (Fig. 4). Among response professionals, agreement with the statement “I have the ability to protect myself and/or others” drops by 34% when framed in terms of a lahar instead of a flood. Among laypeople, this decrease is lower at 25% but still substantial. Participants’ confidence that they will “know what to do during and after” a hazard decreases by approximately a third among both groups when considering a lahar instead of a flood.

Understanding preparedness behaviors in the context of previous models

We find that protective response costs (e.g., knowledge, money, time) and perceived response-efficacy fail to emerge as overwhelming barriers to preparedness behavior adoption. A third of respondents indicate that a lack of hazard knowledge prevents them from preparing and a quarter indicate that cost, time commitment, or a lack of preparedness knowledge influences their choices. Barely seven percent of respondents state that low response-efficacy beliefs (“items will not help me protect myself”) prevent them from preparing further.

Perceived self-efficacy and ascription of responsibility are less readily recognized as barriers by respondents but significantly affect actual preparedness levels. The vast majority of respondents believe that a reliance on others for assistance does not reduce their preparedness behaviors (Table 2). Overall, 68% of respondents disagree with the idea that a reliance on their neighbors or other community members for supplies and assistance during a hazard event reduces their preparedness. Similarly, 83% feel that a reliance on emergency services for supplies and assistance does not hinder their own preparedness. Yet, correlations show a significant decrease in the adoption of preparedness actions among respondents who ascribe greater responsibility for their safety to others and a significant increase in preparedness among respondents who express high self-efficacy and personal responsibility (Tables 3 and 4).

The following discussion opens by describing our findings regarding perceived response-efficacy and protective response costs in the context of previous studies. Although perceived response-efficacy and protective response costs are components of both PMT and PADM, PADM measures perceived response-efficacy in three ways–efficacy for protecting persons, efficacy for protecting property, and utility for other purposes. We specifically consider response-efficacy in terms of efficacy for protecting persons in the present study. As such, perceived response-efficacy in the following discussion refers exclusively to efficacy for protecting persons. This section closes with a discussion of findings related to perceived self-efficacy and ascription of responsibility beliefs.

Professional participation’s influence on household preparedness & personal beliefs

Professional participation appears to improve information seeking habits, confidence in officials, and self-efficacy. Yet, response professionals largely mirror laypeople in terms of their household preparedness levels, ascription of responsibility beliefs, and ability to read and interpret hazard maps. These results indicate that differences exist in how public and professional participation affect an individual’s preparedness behaviors and personal beliefs. This raises the question: why do both types of participation positively affect information seeking behavior, confidence in officials, and self-efficacy, while only public participation positively influences household preparedness, knowledge, and ascription of responsibility?

Self-efficacy and confidence in officials appear to improve regardless of the type of participation (e.g., public or professional) in which an individual engages. Wachinger et al. (2013) posit that an individual’s self-efficacy and confidence in officials improve as they interact more with emergency officials. Both public and professional participation facilitates such interaction. The former increases interactions between the public and officials, while the latter increases interactions among officials. Additionally, it seems logical that response professionals foster higher self-efficacy–the belief in their ability to prepare and respond to hazards effectively–since they elected to pursue careers where their abilities are constantly tested.

Regarding ascription of responsibility, Wachinger et al. (2013) highlight the role that participation in hazard management could play in helping people take greater responsibility for their own safety. Wachinger et al. note that interactions with officials help the public gain a more realistic understanding of their own abilities and the abilities of officials. Members of the public become better acquainted with the measures they can take to prepare, as well as what officials will expect them to bear personal responsibility for during an event. Similarly, Paton et al. (2008) emphasize the need for officials to “empower” the public to take personal responsibility for their safety. Given this emphasis on public participation’s positive influence on personal responsibility, the similarity between the ascription of responsibility beliefs of response professionals and laypeople in our study seems to contradict expectations. However, it is important to note that laypeople in the Skagit Valley already feel primarily responsible for their own safety. With 95% of laypeople already claiming that they are responsible for their own safety, there is not much room for improvement among the response professional community.

In terms of household preparedness, response professionals appear better prepared than laypeople based on their average CP score, but a closer analysis of indicator scores reveals that response professionals are only significantly more prepared in the action indicator category. This difference results because more response professionals have someone in their family who knows first aid and are aware of vulnerable people living in their community. Both of these recommended preparedness actions are strongly tied to professional responsibilities, particularly for first responders and hospital administrators. Thus, it may be more reasonable to attribute increases in these two measures to occupational requirements rather than voluntary preparedness behaviors induced by participation in response planning. All other variations in preparedness of individual measures are minor or not significant. This fact is emphasized by the lack of statistically significant differences in the average planning and supplies indicator scores.

The lack of improvement in household preparedness among response professional respondents may originate because public and professional participation in hazard management represent fundamentally different types of participation. While both aim to improve overall community preparedness and hazard response capabilities, each takes a different approach with separate objectives. Public participation programs tend to be geared toward improving household preparedness or ensuring that hazard plans align with community values. In contrast, trainings for response professionals might only discuss household preparedness as a minor component of a program largely focused on occupational responsibilities for whole community preparedness and response.

For example, one way the public participates in hazard management in the Skagit Valley is through the Community Emergency Response Team (CERT). CERT training teaches individuals about relevant local hazards, preparedness options, and basic disaster response skills (e.g., fire safety, light search and rescue, team organization, and disaster medical operations; Federal Emergency Management Agency 2016). Participating individuals are encouraged to get involved in community preparedness projects. Professional participation activities, on the other hand, focus more on developing an individual’s professional competencies-knowledge and skills that allow their organization to respond effectively within the broader emergency management framework. Household preparedness may increase among the public because participation programs specifically and strongly emphasize how an individual can protect their home and family.

Although response professionals may be acquainted with recommended household preparedness measures, they may still fail to adopt these measures at home. For many response professionals, household preparedness measures do not directly benefit them because they are actively responding to a hazard. However, preparedness measures can help their families, and public health professionals admit that one of their primary concerns during a hazard event is the protection of their family (Slepski 2007). Such concerns can cause distraction or even prevent response professionals from reporting for work (Blessman et al. 2007). Thus, rather than focusing training programs on what to prepare and why, training should focus on how household preparedness can specifically benefit response professionals. Training programs should take a ‘whole community’ approach–emphasizing how household preparedness protects family members, helps response professionals better perform their job duties, and strengthens the whole community. Additionally, we agree with Blessman et al. (2007) recommendation to focus on providing response professionals with small, easily accomplishable steps.

The fact that response professionals and laypeople foster similar household preparedness levels has implications for previous studies of response professionals. The low levels of preparedness previously found among public health employees (Blessman et al. 2007; Rebmann et al. 2013) and first responders (Federal Emergency Management Agency 2016) may be indicative of low levels of preparedness among the public in general. A more representative survey examining a random sample of response professionals and the general public would be necessary to confirm this argument.

Additionally, questions remain regarding the generalizability of these findings, especially given the discrepancy between the expected and demonstrated influence of professional participation on household preparedness. We combined a variety of professions into the group “response professionals,” but the type of participation performed by a first responder may differ substantially from that of a utilities, school, or hospital administrator. By more narrowly defining the “response professionals” category, future studies could determine how well these findings characterize response professionals in general versus subgroups based on specific characteristics of participation (e.g., occupation type, length of involvement, or types of trainings attended). We also lack demographic data on response professionals in the Skagit Valley with which to compare the demographics of survey participants. However, given the tendency to sample more engaged participants with convenience samples, we might expect responses from response professionals to actually overestimate the effects of participation on preparedness.

Advantages and disadvantages of survey methodology

Several limitations stem from the sampling method employed. First, a convenience sample lacks randomness; thus, selection bias may affect the sample, reducing the generalizability of the results. Since a response rate cannot be determined with this method, we cannot account for an individual’s inherent interest or willingness to participate. Second, the questionnaire was only available in English, which limited the participation of non-native English speakers, particularly among the Spanish-speaking population. In Skagit County, 5.1% of the adult (18+) population speaks Spanish or Spanish Creole at home (U. S. Census Bureau 2015). Thus, the survey responses likely underrepresent the views of Spanish-speaking residents. Third, using an online platform limited the number of responses from those without access to a computer or sufficient computer literacy to navigate the questionnaire. The accessibility of the survey tool limited responses from the elderly and those from lower socio-economic backgrounds.

The influence of socio-economic status on response rate is of particular concern in the Skagit Valley, especially among the smaller upstream communities that are most at-risk from lahars. Except Lyman, all of the towns we examine are characterized by lower median household incomes and higher percentages of people living below the poverty level than Washington in general (U.S. Census Bureau 2015). Washington’s median household income is $60,294, but in La Conner, Mount Vernon, Burlington, Sedro-Woolley, Hamilton, and Concrete, median household incomes ranges from $33,977 to $48,399. The percentage of people 18 or older living below the poverty level in Washington is 12.2% but ranges from 14.7% to 33.8% in these six towns. Concrete and Hamilton are closest to the volcanoes and the most threatened by lahars. In these two towns, 28.2% and 33.8% of adult residents live below the poverty level, respectively. The unique socio-economic status of Skagit Valley residents presents difficulties for conducting an online survey given the possibility of reduced computer access in lower income households. Respondent demographics (Table 1) also confirm reduced response rates (-7%) among people with household incomes below $50,000. Such biases could be avoided in the future by using or offering an option for requesting a physical questionnaire.

Although limitations exist, using an online questionnaire with a convenience sampling survey design provided an inexpensive, straight-forward, and relatively rapid means of collecting responses. This method was consistent with previous risk perception and preparedness studies (e.g., Siegrist and Cvetkovich 2000; Bird et al. 2010). While nonrandom sampling limits the ability to extrapolate trends to the broader population, such surveys still provide valuable information on perception and preparedness among the surveyed population. Identified trends demonstrate risk perception and preparedness levels among a portion of the community and may be indicative of broader trends that a future randomized sample survey could investigate.

Additional theoretical and practical implications

From this work, a number of additional theoretical implications arise for future research into preparedness barriers and the benefits of hazard management participation. Our use of respondent opinions in evaluating preparedness barriers reveals the need to refine these questions in terms of framing and format. When asking respondents to indicate the degree to which different factors prevented them from preparing, we framed each option as a potential barrier. In contrast, Terpstra and Lindell (2012) asked respondents which factors were most important in their preparedness decision-making. Their framing did not assume these factors were barriers or promoters of preparedness, but simply factors influencing decisions.

In terms of format, our study expanded upon Terpstra and Lindell’s (2012) use of a dichotomous variable to assess importance. We allowed respondents to express a range of support for different barriers using an ordinal scale. Based on these ratings, we were able to construct a relative ranking of barrier importance. In addition to using an ordinal design, we recommend that future studies allow respondents to rank the relative importance of each barrier or hazard adjustment attribute to provide even greater insight into respondents’ thoughts.

This research highlights the need for incorporating independent or indirect measures of barrier variables for comparison with preparedness behaviors. By including additional questions assessing perceived self-efficacy and ascription of responsibility, we were able to note contradictions between correlations and expressed opinions. Lindell and Whitney (2000), Lindell and Prater (2002), and Terpstra and Lindell (2012) also apply indirect measures of the importance of hazard adjustment attributes. They calculate correlations of attribute ratings for different adjustments with adoption intention and actual adoption. Future studies should similarly allow respondents to express their opinions regarding what motivates their preparedness choices and include corresponding independent measures to compare with preparedness levels. Including both measures provides insight into respondents’ perceptions of barriers as well as actual correlations with behavior.

In terms of knowledge assessment, a third of the respondents who are aware that volcanic hazards exist in the Skagit Valley still indicate that a lack of hazard knowledge prevents them from preparing. This highlights the need to identify what specific knowledge respondents feel they are missing. A general awareness that volcanic hazards exist may feel insufficient. Respondents need to understand what a hazardous event will mean for them personally because understanding the personal impacts of a hazard influences preparedness motivation (Lindell and Perry 2012). People also need to know where to access hazard information. Nearly 23% of the survey population found that current information was difficult to find or understand and 29% felt information was easy to find but unclear. Paton et al. (2008) emphasize that providing information consistent with population needs, values, and beliefs helps emergency managers strengthen trust, reduce uncertainty, and improve the acceptance of information. By determining what specific information the public lacks and desires, as well as how best to present this information, emergency managers can better tailor educational efforts to ensure that the messages and information disseminated are appropriate for their community.

Finally, our results underscore the need for more detailed studies of hazard knowledge, risk perception, and preparedness among the response professional community. Studying response professionals is important because they play a significant role in the success of hazard response efforts and can act as role-models for the broader community. Training programs often introduce response professionals to the concept of household preparedness, yet to date, the household preparedness behaviors and personal beliefs of response professionals remain largely unstudied. Increased program evaluation would provide a clearer understanding of whether or not professional training translates into household readiness. Additionally, comparative studies of response professionals and the general public could offer a means of measuring the success of training programs and provide a more extensive understanding of whole community preparedness.

Furthermore, analyses based on occupation could identify different types of professional participation and how each influences household preparedness and personal beliefs. Such studies could isolate elements shared between the most effective training programs within and across professional boundaries. The goal of these efforts being to increase household preparedness and reduce possible distractions facing response professionals. If response professionals feel confident in the safety of their families, they can feel comfortable responding, which ultimately benefits the whole community. Overall, the results presented here reveal the important role that participation type plays in determining household preparedness actions.

In terms of practical implications, the findings presented herein will be provided to local and state emergency managers to assist in the development of improved public education programs, professional training programs, and response plans. We support the recommendation of Paton et al. (2008) that emergency managers should strive to empower the public. Managers should help individuals recognize their own agency during hazard events and improve their self-efficacy, both of which clearly and positively influence preparedness behaviors. Hazard management participation efforts should also be expanded given the positive impact that participation appears to have on self-efficacy and feelings of responsibility, impacts which do not appear to be tied to specific types of participation. For response professionals, household preparedness measures should be presented as small, easily achievable steps that will benefit their family and help them better perform their response duties. Further research could also explore the opportunities that enhanced school-based hazard education and its links to community-based public education offer as a way to increase community empowerment and participation in preparedness activities (Johnson et al. 2016; Ronan et al. 2016).

This research will be shared with the Cascade Volcano Observatory (CVO) as well. The CVO’s input in the design of the survey questionnaire ensured the collection of information relevant to their design of volcanic hazard maps. The current hazard maps successfully communicate the main details of the hazard, but more nuanced elements are not as easily conveyed.