THIS LACK OF PRE-EVENT DATA IS A PROBLEM BECAUSE HUMAN RESPONSES TO DISASTERS OFTEN DETERMINE RESILIENCE AND RECOVERY

The public is a critical partner for local, state, and federal agencies in disaster response. ^4,5 Public perceptions drive public reaction, both immediately (protective behavior, evacuation decisions) and over the long term (migration, market demand, and pressure on policy-makers). Anticipating, responding to, and managing public reactions require a means of tracking and understanding those reactions. Self-reports from the public can be the best or only source of information, such as perceived risk, behavioral intentions, and social learning. To best inform intervention and policy, that information needs to be collected using rigorous scientific methods, including well-designed surveys with large probability-based samples and longitudinal assessment of individuals across the life-cycle of a disaster and across multiple disasters. ^{Reference Lurie, Manolio and Patterson6}

OUR EXPERIENCE ILLUSTRATES THE NEED FOR DISASTER SURVEYS THAT ARE PROSPECTIVE, RATHER THAN REACTIVE

We designed our 2017 Hurricane Harvey survey to leverage the 2016 Study of Trauma, Resilience, and Opportunity among Neighborhoods in the Gulf (STRONG), ⁷ conducted by the Consortium for Resilient Gulf Coast Communities. STRONG is a regionally representative survey of 2520 Gulf Coast residents that assessed the long-term health, economic, and social consequences of the Deepwater Horizon oil spill. The STRONG survey included many public response measures – mental and physical health, financial well-being, social support, perceived exposure, perceived risk – that were relevant to Hurricane Harvey but were fundamentally designed for a different purpose.

In response to Harvey, we re-contacted the 623 STRONG respondents in Houston and coastal Texas and obtained longitudinal data from 295 (a 47.4 percent retention rate). These data provide valuable insights into the human impacts of a hurricane the size and scale of Harvey, helping us understand how people’s prior exposure to a disaster affects their response to a new one, and how new disasters impact recovery from prior disasters. What was missing, and is missing from most studies, was a prospective research infrastructure designed to capture exactly this sort of representative, longitudinal data. Researchers need longitudinal data that include pre-event (baseline) data collected for this purpose, assessment throughout acute disaster phases, and long-term public reactions. Researchers also need a use-inspired ^{Reference Stokes8,Reference Wall, McNie and Garfin9} approach that supplies data for a broad set of research and practice needs, and allows the easy integration of other data sources.

WE CONDUCTED A MAPPING REVIEW OF THE CURRENT STATE OF ACADEMIC DISASTER SURVEY RESEARCH OVER THE PAST 20 YEARS

We searched 2 large, general databases (SCOPUS and Web of Science) for journal articles between 1998 and 2018 about large-scale, probability-sampled surveys of public reactions to US disasters. ¹⁰ Our initial search identified a substantial literature of 3103 documents, but just 79 matched our criteria because most studies lacked strong probability sampling methods, which are crucial to ensure that the surveyed sample reflects the overall population.

The studies covered diverse events: 31 articles on terrorist attacks (primarily about September 11), 21 on disease outbreaks (primarily H1N1 and Ebola), 10 on hurricanes (primarily Katrina and Ike), and 4 on the Deepwater Horizon oil spill (multiple forthcoming papers from our own team will add to this list for Deepwater Horizon and Hurricane Harvey). Other studies covered drought, earthquake, and US reactions to the combined earthquake, tsunami, and nuclear accident affecting Japan’s Fukushima Daiichi power plant. Just 3 articles addressed multiple disasters.

Most articles focused on narrow outcomes (eg, risk perception, mental health, physical health), rather than capturing a breadth of psychological, health, economic, and social effects. What is remarkable about this literature is how often it appeared to be reactive: just 19 (24%) articles involved predisaster data. Most built off of existing data and samples originally collected for other purposes, as we did with STRONG. Five were ongoing studies that re-oriented or added capacity because of the disaster, as was done with the Chernobyl study. None were planned pre-post designs. In addition, relatively few (28 of 79; 35%) articles reported data that were collected within 1 month of the disaster event. Most articles (56 of 79; 71%) reported data collected during the extended recovery period (Figure 1).

FIGURE 1 The Proportion of 1998-2018 Journal Articles with Probability-Sampled Surveys at Different Disaster Stages, Demonstrating How Disaster Survey Research Tends Not to be Prospective.

MOVING FORWARD, THE RESEARCH COMMUNITY NEEDS PLANNED ASSESSMENTS OF PUBLIC REACTION TO DISASTERS TO INFORM SCIENCE AND COMMUNICATION, AND MAKE THE PUBLIC A MORE EFFECTIVE RESPONSE PARTNER

This requires a research infrastructure for tracking public fears, understanding, and behavior – preferably in near-real time. The National Science Foundation (NSF) recently took an important step in this direction, taking a powerful model for quickly and nimbly fielding geoscientists and engineers to capture perishable post-disaster data, through its Geotechnical Extreme Events Reconnaissance (GEER) program, ¹¹ and extending the model to the social sciences (SSEER) and interdisciplinary efforts (ISEER). ¹² These programs place scientists on the ground in disaster zones and are critical to our understanding of human dynamics immediately after a disaster. Federal granting agencies (notably NSF and the National Institutes of Health) also employ rapid grant programs for quickly fielding research after disasters, but these efforts are by their very nature delayed because they are reactive (our Hurricane Harvey study was funded through an NSF RAPID grant).

What we do not have is the pre-positioned capacity to nimbly and longitudinally capture public response, at a population level, from pre-event baseline, through acute disaster phases, and into what is now known to be an extensive recovery period.^{Reference Norris, Stevens and Pfefferbaum13,Reference Gill and Picou14} Laying the groundwork for this necessary data collection will take planning and investment. Operationally, this will likely involve a large-scale panel study. The United States invests in many such nationwide studies, such as the Health and Retirement Study and the Panel Study on Income Dynamics, but a disaster panel would need to be more nimble and adaptive than such studies. Developing this panel study will require substantial innovation and attention to not only theoretical and methodological detail, but also to the practical needs of the communities, policy-makers, and disaster responders.