The National Household Survey project was active in three areas. The research team first focused on designing, conducting, and analyzing a national household survey to describe, explain, and predict actual public preparedness, mitigation, and avoidance actions, intended actions, and relevant perceptions for all major hazards with an emphasis on the hazards of terrorism. Second, the team re-analyzed existing survey data on public response to extreme events. Finally, the team prepared a questionnaire that could be administered rapidly in the event of either a terrorist attack or a highly elevated warning.
Respondents were asked whether they had (since September 11, 2001) developed emergency plans; stockpiled supplies; purchased things to be safer; learned more about where to get information about terrorism; duplicated important documents; become more vigilant; reduced travel by airplane; reduced travel by train; reduced use of public transportation; changed mail handling procedures; avoided travel to certain cities; avoided tall buildings; avoided national landmarks. For each activity done, respondents were asked whether they did it because of terrorism, natural disasters, reasons other than terrorism or natural disasters, or any combination of the three reasons. With the exception of learning more about getting information and becoming more vigilant, less than 5% of respondents reported doing any of the actions exclusively because of terrorism. In contrast, between 6% and 20% of respondents reported doing each of the 11 activities (excluding vigilance and learning) because of natural hazards, other reasons or some combination of reasons. Reports of preparedness and mitigation activities did not differ between high risk areas (New York City, Washington, D.C., Los Angeles County) and the rest of the United States, or between race/ethnic groups. The major predictors of preparedness behaviors are information received and information observed. The more types of information received, the more sources from and channels over which it is heard, and the more respondents have seen others engage in preparedness behaviors, the more preparedness behavior was reported by respondents. Knowledge about terrorism, the perceived effectiveness of preparedness activities, and the amount of information respondents have actively sought are secondary predictors of preparedness. Other potential predictors, such as standard demographic variables, past experience in disasters, perceived future risk of terrorism and natural disasters, and perceived resilience of the respondent, and federal, state, and local governments to protect and respond to terrorism and natural disasters do not add to the model.
A national household survey was conducted. The questionnaire was based on a comprehensive review of the literature on disaster preparedness and mitigation, public education, risk communication and warnings (Mileti DS, Bandy R, Bourque LB, Johnson A, Kano M, Peek L, Wood MM, et al. (2006, September). Annotated bibliography for public risk communication on warnings for public protective actions response and public education (revision 5). Boulder, CO: University of Colorado, Boulder. Telephone interviews were conducted by California Survey Reseach Services, Inc. (www.calsurvey.com) using computer-assisted telephone interviewing (CATI) procedures between April 13, 2007 and February 13, 2008. Interviews were done in Spanish or English at the request of the respondent, and a $20 incentive was offered to encourage participation. The national sample was stratified into two levels of visibility, or risk. High visibility areas are high-profile areas with potential terrorist targets. The high-risk stratum included Washington, D.C. (including the District of Columbia, Arlington, Fairfax, Prince William, Loudoun, Montgomery, and Prince George's counties), Los Angeles County, and New York City (including Bronx, Brooklyn, Manhattan, Queens, and Staten Island). The low-risk stratum was the rest of the continental United States. The high-risk stratum was oversampled to enable comparisons both with the low-risk stratum and within the high-risk stratum. The sample was selected using random-digit-dialing (rdd), supplemented with random sampling from Hispanic and Asian/Pacific Islander surname lists in an attempt to allow separate analyses by race/ethnicity group. Interviews were completed with an adult respondent over age 18 from 3,300 households for a response rate of 35%.