The proposed work will undertake a multistage cluster sample of household addresses in New Orleans, supplemented by a phone banks survey. The goal is to collect 500 respondents by May 2006. The survey questionnaire will ask about the decision to evacuate, survival experiences, demographic data, the kinds of needs that people had a different timepoints before, during, and after the hurricane. The survey has been designed by Edd Hauser, and will be piloted on refugees at UNC-Charlotte and Duke before used in the field. The first draft is under preliminary review by Duke Institutional Review Board, and will be modified according to their feedback. The first part of the analysis will summarize the survey data, and provide cross-tabulations that show how different survival experiences are associated with age, pet ownership, income, and so forth. This is standard statistical analysis, but will help identify which aspects played dominant roles at different phases of the emergency. The second part of the analysis will fit a social network model to the data, extending previous work on mathematical models for the kinds of help people need and from whom they get help. The nodes in the social network model will be people or agencies, and the edges in the network describe the kind of help that is required. The research will be conducted jointly by David Banks, Edd Hauser, and John Lefante (working with Maria Sirois). David Banks and John Lefante are in charge of the survey design and data analysis. Edd Hauser, head of the Regional Center for Homeland Security and Major Disaster Management, is in charge of questionnaire design. Maria Sirois is a graduate student in Biostatistics in the School of Public Health and Tropical Medicine at Tulane; she and John Lefante are in charge of data collection. Maria Sirois is a member of the National Guard, the Louisiana Search and Rescue Team, and a certified first-responder.
The intellectual merit of the proposed research is twofold. First it will provide a statistical analysis of the factors affecting the survival experience of people in New Orleans before, during and after Katrina. Second, it will extend the social networks models applied in previous studies of disaster response by using the new class of latent space models and the data obtained from the Katrina event. The broader impact of the proposed research is that it will provide guidance and insight for future disaster relief efforts, and highlight the interactions between people and agencies that were most/least effective at various stages of the response.
