The Public’s Preparedness: Self-Reliance, Flashbulb Memories, and Conservative Values

Abstract

Objectives. We surveyed how many US residents engaged in 6 preparedness activities and measured the relationship between engagement and personal experience in hazard events, flashbulb memories of major events, self-reliance, and other indicators of a conservative philosophy.

Methods. We used random digit dialing for national landline (75%) and cell phone (25%) surveys of 1930 US residents from July 6, 2011, to September 9, 2011; 1080 of the sample lived near 6 US Department of Energy nuclear waste management facilities and 850 were a national random sample.

Results. The median respondent engaged in 3 of the 6 activities; those who disproportionately engaged in 4 or more had experienced a hazard event, had distressing and strong flashbulb memories of major hazard events, and had strong feelings about the need for greater self-reliance. The results for the national and US Department of Energy site–specific surveys were almost identical.

Conclusions. A cadre of US residents are disproportionately engaged in disaster preparedness, and they typically have stronger negative memories of past disasters and tend to be self-reliant. How their efforts can or should be integrated into local preparedness efforts is unclear.

Preparedness has been an important subject in the American Journal of Public Health, including in an electronic journal supplement in 2007.^1–5 Much of the literature, with some important exceptions,^1,3 has been about defining preparedness and the public health system’s capacity. Much less attention has been devoted to the public’s preparedness. The message from the literature is that much of the public is not ready to respond to serious events.^6–10 Why do so many not have a fire extinguisher, not have a communication plan with loved ones in the case of an event, not have a food supply set aside, and not have taken other steps that would increase their probability of survival in a serious event?

Our focus in this article was on identifying those individuals who are most prepared and comparing their attributes with those who are less prepared. Using survey responses collected in 2011, our purpose was to answer 2 questions:

1. What proportion of US respondents have engaged in preparedness actions ranging from knowing how to use a fire extinguisher to predetermining a place to meet if a hazardous event occurs?

2. What factors are most strongly associated with taking multiple preparedness actions?

Several key elements of the preparedness literature guided this research. Defining “preparedness” was one. Bourque et al.¹⁰ used National Survey of Disaster Experiences and Preparedness data that included developing emergency plans, stockpiling supplies, purchasing things to be safer, learning how to get information, duplicating documents, and becoming more vigilant. They also examined avoidance activities, such as reducing airplane, train, and public transportation use; avoiding some cities, tall buildings, and national landmarks, and changing mail-handling practices.

They found that 94% of respondents engaged in 1 or more of the 6 proactive activities, 84% reported that they became more vigilant, 32% had an emergency plan, 37% had stockpiled supplies, and 60% had learned how to get information. The median response was to have engaged in 3 of 6 activities. Residents of New York City and Washington, DC; men; and high-income respondents reported more preparedness activities. The authors concluded that US households are “remarkably unprepared for disasters and emergencies, including terrorism,”^10(p402) a common conclusion in the literature.^6–15

Why are so few people prepared? One explanation is people are bombarded with so much information, including information about risks that seem more likely and threatening than a tornado or terrorist attack.¹⁶ Second, preparedness is not clearly defined in ways that lead them to take action.^2,5,17 For some, preparedness might be learning how to access the latest information, but for others preparedness might mean leaving and knowing where to go before any warnings. Third, people may become numb to the threat if they have been warned and asked to move too many times and no threat materialized. Some may have already taken action and become less worried about the threat.¹⁸

Given many people’s lack of engagement, researchers have been developing conceptual models to predict preparedness,^19–22 and these models were the basis for our research design, including such factors as personal experience in a disaster, strong emotions about disasters, trust in authority, self-efficacy, optimistic bias, psychological distance from the issue, and demographic attributes.^23–38

This literature suggests that preparedness is a deliberative action requiring strong reasons to participate because the events are low probability and most people do not believe they will be affected. We expected, first, that personal experience with a hazard event that left strong distressing memories, such as horror and fear, is strong motivation for preparedness. Our second expectation was that deeply embedded memories of major national or international disasters drive preparedness because these events were so shocking, salient, and consequential that they left so-called “flashbulb” memories, which are detailed recollections about the event, including where the respondent was and what he or she was doing during the event.^39–47 Indeed, deeply embedded memories can be used as one diagnostic to predict posttraumatic stress after wars, the 9/11 terrorist attacks, earthquakes, and hurricanes, even among those with Alzheimer’s disease.^42–47 Hence, we expected personal experience in events, emotional reactions to those events, and flashbulb memories of some of the worst events in recent history to be the strongest correlates of preparedness.

We expected that although strong negative memories and personal experiences would be signal events, they would not lead everyone to prepare for low-probability hazard events. Those who took action would tend to assert conservative values that emphasize self-help, lack of trust in authority to protect them, and a resistance to immediately agreeing with government-initiated policy proposals. In addition, we expected demographic attributes, as well as respondents’ location, to influence the results. Overall, our key expectation was focused on strong negative memories and personal experiences; secondarily, we focused on the role of self-reliance, demographic attributes, and location.

METHODS

The data were gathered as part of a US Department of Energy survey that focused on the US public’s preferences for and perceptions of different electrical energy sources, as well as on nuclear waste management after the March 2011 earthquake, tsunami, and subsequent nuclear power hazard events in Japan. This survey’s larger purpose connected with that of this study because the Department of Energy must maintain a high level of preparedness and is motivated to understand the preparedness of those who live near its facilities and the correlates of preparedness.⁴⁸ The Department of Energy survey included questions about respondent demographic attributes, trust, values, and preferences. We used some of these questions in this analysis. The first author also added a block of questions assessing preparedness to investigate the 2 research questions posed in this article.

Many activities may be classified as preparedness. The first author prepared a list of approximately 2 dozen activities from the Federal Emergency Management Agency’s Disaster Planning Guide,⁴⁹ local published guides,⁵⁰ the published literature,¹⁰ and Web links to organizations that promote emergency preparedness, such as the Boy Scouts of America.⁵¹ Greenberg discussed these ideas with members of a center for disaster preparedness and emergency response group of which he is a part. After debate, this process ultimately yielded 6 questions.

We said, “I’d like to ask you about your preparedness for a major disaster in your area.” Respondents were asked to answer “yes” (1) or “no” (0) regarding their engagement in 6 activities (Table 1). We expected many people to answer yes to the 2 questions about fire extinguishers, “Do you know how to use a fire extinguisher?” and “Do you have a fire extinguisher at home?” We expected far fewer respondents to answer “yes” to other questions, for example, “Do you have a disaster supply kit for your home with emergency supplies such as water, food, and medicine that is kept apart from everyday use?” The other questions asked respondents whether they had a plan for an extended stay at home, a family communication plan, and an agreed-on place to meet the case of an emergency (Table 1).

TABLE 1—

Proportion of Respondents Engaged in Proactive Preparedness Activities: Phone Survey of US Residents, July 6, 2011–September 9, 2011

Summary Data	National (n = 803), %	Site Specific (n = 1038), %
Aggregate activities of 6
0	2.6	2.3
1	9.3	9.4
2	23.2	19.4
3	19.8	20.5
4	17.6	17.8
5	14.6	15.5
6	13.0	15.0
Central tendency, average	3.5	3.4
Individual activitiesa
Know how to use a fire extinguisher	89.6	91.3
Have a fire extinguisher at home	74.6	74.4
Have a family communication plan so that you will be able to contact family members or loved ones if you get separated during an emergency	56.8	54.9
Have developed a plan for an extended stay at home in case of a disaster	40.8	42.8
Have a disaster supply kit for your home with emergency supplies such as water, food, and medicine that is kept apart from everyday life	39.1	41.2
You and your family have agreed on a place to meet in case an emergency prevents you from being home	36.7	39.3

^aNone of the national and site-specific differences are different as measured by statistical significance tests at P < .05.

With regard to personal experiences, we asked respondents whether they had ever been in a hurricane, flood, earthquake, tornado, major fire, train derailment, mudslide, or explosion. Then we asked whether they or a close family member had ever been evacuated from a hazard event. We expected that a positive response would be a stronger predictor than evacuation because evacuation without a serious event leads some people to not respond to future calls for evacuation.

A unique contribution of this survey was to test for a relationship between flashbulb memories and preparedness. In the preamble to the questions, respondents were asked how much they remembered about 6 major hazard events. Response options were “not much,” “a few details,” “many details,” and “many details and clearly remember where I was when I learned about it.” The 6 events began with the World Trade Center attack in 2001 and ended with the earthquake, tsunami, and nuclear power plant events in Japan in 2011 (Table 2). This list was presented randomly to avoid order bias. We coded responses on a scale ranging from 1 (“not much”) to 4 (“remember many details and where I was when I heard about the event”). A response of 4 was coded as a flashbulb memory, and we then used the total number of flashbulb memories as a predictor. We expected those who had many flashbulb memories of these painful events to be more concerned about major hazard events and likely to take proactive actions. We followed these questions with an open-ended probe that asked respondents to indicate the first 2 emotions associated with the flashbulb memory of greatest strength.

TABLE 2—

Flashbulb Memories and Number of Preparedness Actions: Phone Survey of US Residents, July 6, 2011–September 9, 2011

Event	Don’t Remember Anything	Remember a Few Details	Remember Many Details	Remember Many Details and Where I Was When Learned About Event
World Trade Center, 2001
NTL	1.9	6.2	19.3	72.6
SS	1.9	9.1	20.9	68.1
Hurricane Katrina in New Orleans, 2005
NTL	2.4	16.5	38.5	42.6
SS	2.1	18.5	42.0	37.3
Offshore oil drilling platform blowout and oil spill in the Gulf of Mexico, 2010
NTL	4.3	23.0	43.2	29.5
SS	5.1	26.1	41.1	27.7
Coal ash spill in eastern Tennessee, 2008a
NTL	60.2	30.4	5.4	4.0
SS	51.0	31.3	10.3	7.5
Tsunami in the Indian Ocean that affected Thailand, Sri Lanka, Indonesia, India, 2004
NTL	11.5	37.8	30.3	20.4
SS	12.9	43.4	27.2	16.4
Earthquake, tsunami, nuclear power plants failure in Japan, 2011
NTL	3.4	18.5	39.1	39.0
SS	3.7	20.6	38.2	37.5

Note. NTL = national; SS = site specific. For 6 events, Cronbach’s α = .734. Average number of respondents in the national sample was n = 847; for the site-specific survey, it was n = 1075.

^aSite-specific sample significantly higher than national sample at P < .05. (See Table A2, available as a supplement to the online version of this article at http://www.ajph.org, for Oak Ridge data.)

We created a second set of variables to identify respondents who manifested conservative values. Key among these values are self-reliance and resistance to changing some existing policies, such as those regarding energy. We asked people to tell us whether they agreed with the statement that “too many people expect society to do things for them that they should be doing for themselves.” We expected those who agreed to be more engaged in preparedness. Public opinion polls have indicated that respondents strongly favor greater reliance on solar energy and much less reliance on oil and coal to generate electricity.^52,53 Respondents were asked whether the United States should rely on these forms of energy “more” or “less” or whether reliance should “stay the same.” We expected to find a disproportionate number of people in our proactive preparedness group who favored coal and oil and were less enamored of solar energy.

Public opinion polls have shown that the majority of Americans state that they “strongly” support environmental protection programs or “somewhat” support them.^52,53 Relatively few respondents state that they are either neutral about environmental issues or are not even concerned about them. We expected our proactive preparedness group to include a disproportionate number of those who were less concerned and neutral. Furthermore, we expected many of these respondents to be pessimistic about the future of the environment in their own region. Another question was about trust. We expected respondents who did not trust responsible parties to manage a potential problem to be more likely to take proactive steps. As part of the larger survey, we asked respondents to tell us whether they trusted the US Department of Energy to manage new nuclear facilities on their sites. We expected those who disagreed to disproportionately be part of our proactive planning group. The Republican Party is thought to be associated with smaller government and more independent action. Accordingly, we assumed that people who self-identify as Republican would be disproportionately more likely to be part of our self-reliant, conservative-values group.^54,55

We built the last set of questions around the idea of individuals who would not only want to take proactive steps but also feel as though they had the capacity to do so. We expected White men aged 35 to 64 years with incomes higher than the national norm to be part of the proactive group because as a whole they have more income and more political access. We tested these expectations by asking respondents about their income, educational achievement, race/ethnicity, gender, and age. Our last demographic questions concerned the locations of the national sample and the 6 Department of Energy sites.^48,52,53

Sample Administration

The survey protocol included a combination of landline (75%) and cell phones (25%) because of the rapid growth in the number of cell phone–only users.^56,57 Of the sample, 850 respondents were national and 1080 were site specific (6 locations × 180 respondents per location). The site-specific samples included people who lived within a 50-mile radius of 6 designated nuclear sites that we had previously sampled: (1) Idaho National Laboratory (Idaho Falls, ID), (2) Oak Ridge National Laboratory (Oak Ridge, TN), (3) Waste Isolation Pilot Plant (Carlsbad, NM), (4) Savannah River National Laboratory (Aiken, SC), (5) Los Alamos National Laboratory (Los Alamos, NM), and (6) Hanford Site (Richland, WA). These sites include the major defense waste management sites and several sites that also have substantial science and weapons testing functions.^48,52,53

We observed the American Association for Public Opinion Research’s protocols. These protocols eliminate phones not in service and exclude businesses and other inappropriate landline numbers. In the case of cell phones, we had to eliminate more numbers for the site-specific samples than for the national sample because many people take their cell numbers with them when they move. Our target response rate and cooperation rate were 20% and 30%, respectively. On the basis of recent trends in phone sampling, we assumed that 10 calls to each number would be required.^58,59 The survey was administered in both English and Spanish, with 2.4% administered in Spanish.

Analysis

Given that this survey included both a national sample and site-specific samples, the results for question 1 have been separated into national and site-specific results. We compared the national sample results with an aggregate of those for the 6 site-specific samples using the t-test of the difference of means. Appendix 1 (available as a supplement to the online version of this article at http://www.ajph.org) presents separate data for each of the 6 sites. We compared the averages for each site with the national sample and each other using 1-way analysis of variance (the F test for the entire distribution) and post hoc tests (the Tukey B and Scheffé) for the national sample and each site sample.

We used ordinal regression analysis for question 2 because the dependent variable is number of preparedness activities (response options were 0, 1, 2, 3, 4, 5, and 6). We used the Nagelkerke pseudo-r² and the Wald statistic to evaluate each variable and the overall results. The regression results are for 2 models that pooled all the data. To reiterate, of the 1930 respondents, 850 were from a national random sample and 1080 (180 for each of 6 sites) were from the site-specific samples. Each site sample shares some similar attributes with the national and other site-specific samples, and yet something unique about each site may separate it from the other sites. Indeed, the first author has previously studied these 6 sites, and we expected differences for at least 1 of the 6.^48,52,53 We captured these site-related differences by using 6 dummy variables, 1 for each site.

We constructed 2 regression models to test the predictors. The first focused only on personal experiences, reactions to those experiences, and negative recollections of recent major US hazard events. We added the role of self-reliance, other conservative values, demographic attributes, and location in a second regression model. Also, the first author used Cronbach’s α to determine whether a single preparedness scale could be derived from the individual 6 preparedness behaviors and whether a single flashbulb memory scale could be built from the 6 major hazard events we queried about.

RESULTS

The surveys were pretested, which led to minor changes. The data gathering began on July 6, 2011, and ended on September 9, 2011. The first author wrote the questions with the assistance of the Edward J. Bloustein School Survey Research Center, and the responses were collected by Abt-SRBI (New York City). We compared the national sample and the site-specific samples with the national and site-specific regional population baselines for differences by age, race, and gender. We adjusted for sample bias by weighting the results by race and age (i.e., White and non-White and age ratios 18–44, 45–64, and ≥ 65 years). Note, however, that weighting does not eliminate all sample survey bias. (See appendix 2, available as a supplement to the online version of this article at http://www.ajph.org, for more detail about the survey.)

Both cell and landline surveys are essential because so many people now rely on cell phones. We expected age differences because cell users are disproportionately younger. Age differences per se were not a concern because the data can be merged. The concern was that cell and landline users, after controlling for age, might provide different answers to the key questions and thus bias the sample. We found no significant differences by age group that were not associated with age and therefore merged the cell and landline data (see appendix 1 for additional detail about the sample).

Date of sample was another plausible source of bias in this analysis. Of the sample, data for 48% were obtained in July and for another 48% in August. The data for the remaining 4% were obtained in September, and these results could be significantly different because the 10-year anniversary of the terrorist attacks of September 11, 2001, could have led to more awareness of the 9/11 events and more preparedness. However, sample date did not have an impact on the results. The proportions of respondents who had flashbulb memories of the World Trade Center events were 72%, 72%, and 73% for July, August, and September, respectively. The average number of preparedness events noted by respondents in July, August, and September were 3.47, 3.48, and 3.47, respectively. In short, we did not need to include date of survey as a factor in regression analyses.

Engagement in Preparedness

Table 1 shows that about 2% of the sample had not engaged in at least 1 of the 6 preparedness activities and that approximately 14% said that they had engaged in all 6. (The percentages in the text are average for the aggregate of national and site-specific samples.) The median was 3 of 6. Knowing how to use a fire extinguisher (about 90%) and having 1 at home (74%) were the most prevalent preparedness activities for both the national and site-specific samples. The 4 actions that required deliberative planning were less prevalent, for example, from about 41% for preparing for an extended stay at home to approximately 37% for choosing an agreed-on place to meet in case an emergency prevents one from being at home. These results are similar to those reported in the literature, and the Cronbach’s α among the 6 was .744, which means that they could be used as a single preparedness scale.

Association With Predictor Variables

Table 2 presents the flashbulb memory data (percentages in the text are average for the aggregate of national and site-specific samples). Of the respondents, 70% had flashbulb memories of the World Trade Center attacks. A large proportion indicated that they had very detailed memories of Hurricane Katrina (about 40%) and the disastrous 2011 earthquake, tsunami, and nuclear failure in Japan (about 38%). The offshore oil rig blowout in the Gulf of Mexico produced flashbulb memories among approximately 29%. The tsunamis that hit the Indian Ocean and killed hundreds of thousands of people in Southeast Asia produced far fewer detailed recollections (about 18%), and a coal impoundment break in Kingston, Tennessee, produced even fewer (approximately 6%). Notably, however, 1 of our 6 site-specific areas is centered on Oak Ridge, Tennessee. Of those respondents, 35% had considerable recall of the coal impoundment events, underscoring the salience of physical and psychological distance. We found strong statistical relationships between memories of all of these events. That is, some respondents had detailed recollections about 5 or 6 of the 6 events, and some had little recollection. Cronbach’s α among the 6 was .734. We added the 6 flashbulb memory measures to produce a flashbulb memory scale ranging from 0 to 6 (0 = “no memories of any of the 6,” 6 = “flashbulb memories of all 6”).

Table 3 presents the results of 2 ordinal regressions. Model 1 measured the relationship between preparedness activities, personal experience, and flashbulb recollections. Having experienced a hazardous event and more flashbulb memories were significant predictors of the number of preparedness activities. The parameter estimates for the flashbulb memories gradually increased. Having been evacuated and recalling feelings of horror or fear from the 6 events were not significant predictors.

TABLE 3—

Experiences, Flashbulb Memories, and Proactive Actions: Phone Survey of US Residents, July 6, 2011–September 9, 2011

Variable	Model 1, B (SE)	Model 2, B (SE)
Images
Recall horror	0.420 (0.215)	0.366 (0.219)
Recall fear	0.056 (0.219)	0.108 (0.222)
Flashbulb memories
0 (Ref)	0	0
1	0.195 (0.300)	0.217 (0.306)
2	0.811* (0.285)	0.723* (0.293)
3	0.949* (0.284)	0.805* (0.291)
4	0.996* (0.279)	0.926* (0.286)
5	1.125* (0.273)	1.026* (0.280)
6	1.262* (0.274)	1.185* (0.283)
Experiences
Experienced hazard event	0.639* (0.089)	0.559* (0.091)
Evacuated	0.127 (0.107)	0.162 (0.108)
Values and preferences
Agree that people rely on government to do things that they should do for themselves		0.385* (0.091)
Self-identifies with Republican Party		0.233* (0.103)
US should increase reliance on coal		0.111 (0.100)
US should increase reliance on solar energy		−0.075 (0.140)
US should increase dependence on oil		0.206* (0.104)
Am not concerned about environmental problems in the United States		0.486* (0.234)
Environment in my area will be better in 25 y		−0.033 (0.091)
Trust DOE to manage new nuclear facilities safely (disagree and strongly disagree)		0.249* (0.093)
Demographic attributes
Aged 35–64 y		0.108 (0.086)
Male		0.345* (0.085)
Annual family income > $75 000		0.012 (0.103)
College graduate		−0.012 (0.095)
Self-identifies as White		0.108 (0.148)
Self-identifies as Black		−0.249 (0.209)
Self identifies as Latino		−0.003 (0.136)
Lives in Hanford site area, WA		−0.020 (0.154)
Lives in Idaho site area		0.637* (0.155)
Lives in Los Alamos area, NM		−0.017 (0.156)
Lives in Oak Ridge area, TN		0.160 (0.154)
Lives in Savannah River area, SC		−0.028 (0.161)
Lives in WIPP area, NM		0.104 (0.156)
Nagelkerke pseudo-r2	0.174	0.241
Cox and Snell pseudo-r2	0.172	0.241

Note. DOE = US Department of Energy; WIPP = Waste Isolation Pilot Plant.

*P < .05. P values determined by the Wald test.

Model 2 included the demographic characteristics, preference for, and indicators of conservative values, which increased the predictive capacity of the model from Nagelkerke pseudo-r² = .174 to Nagelkerke pseudo-r² = .241. The significant predictors added in model 2 were agreement that people rely too much on government to do things that they should do for themselves, lack of concern about environmental problems, and preference that the country increase reliance on oil to produce electrical energy. Furthermore, these respondents were disproportionately men and lived in the Idaho nuclear defense region.

DISCUSSION

Those who had engaged in 4 or more of the 6 preparedness activities disproportionately had personal experience with hazard events and had many flashbulb memories of some of the most serious recent disasters. Individually, they disproportionately believed that people should rely on themselves and less on government to provide services and support. As an illustration of what seems to be a conservative personal philosophy, and contrary to most US residents, they are not as eager as their counterparts to give up coal and oil as fuel sources or adopt solar energy. Men were more likely to be in this preparedness group, but the demographic differences were overshadowed by other attributes.

We point to several important limitations. Given these findings, having indicators that would allow us to obtain a more nuanced understanding of the respondents would have been valuable. For example, were they also engaged in proactive activities in the community or at work? We would like to know more about their family history of engaging in preparedness. Given their positions on self-reliance, adding a personal mastery scale in future surveys would be helpful. The last limitation is that 850 of the respondents were a random US sample and 1080 were in specific US regions. As the interesting work of Eisenman et al.^13,14 has shown, oversampling is needed in urban areas in which the preparedness challenge may be the greatest. Urban oversampling was not part of this design. Hence, a next step is to repeat this survey oversampling from large cities and dense suburbs because the large multiethnic population in large cities feels an increased vulnerability and lack of control.^13,14,60

Notwithstanding the limitations of the study, the results raise an interesting policy dilemma for local preparedness managers. Preparedness is associated with local social capital,^28,37and certainly historical examples exist of successful efforts to galvanize community groups around a common preparedness and response motive.^61,62 Presumably, some members of this high-preparedness group could work with local officials to improve local preparedness or could at least help others with self-help skills. Yet, would they welcome efforts to involve them, or would they distance themselves from such efforts? Do they even support government investments in preparedness?

Would local first responders want them to be involved and, if so, to what extent? Enarson⁶³ argued that too much of preparedness is being left to individuals and households rather than to communities, yet their involvement in community-based efforts has potential costs that merit careful discussion. Researchers have found a critical need to build community- and neighborhood-level networks.^64,65 These networks are grounded in personal relationships and local organizations. Considerable effort is required to create and maintain the expertise and commitment of volunteers who respond before, during, and after a hazard event. These groups have an esprit de corps that is typically built by working on food security, local education, and clothing distribution. Some who are personally prepared for a hazard event may already be part of a community group and be comfortable with working with their neighbors, local not-for-profits, and local and state governments. Others may be hesitant to engage and, if they did engage, could disrupt the group’s capacity to respond. On the basis of this survey, we cannot say what role, if any, this group of prepared US residents should play. We believe that each local government preparedness community, using its contacts and knowledge of local circumstances and people, is best situated to consider the pros and cons for their area.