Measuring Individual Disaster Resilience in Louisiana: A Multilevel Trend Study
(1) The gap was addressed by addressing the following hypothesis: Is there a significant relationship between objective community disaster resilience factors, subjective individual vulnerability factors and individual anxiety and depression?
Methods: The study was a multilevel, repeated cross-sectional design with a three-level, nested structure (individuals nested within Louisiana parish cohorts and Louisiana parish cohorts nested within parishes). Secondary data was gathered from the Behavioral Risk Factor Surveillance System for 3 cohorts in 2006, 2008 and 2010. Parish-level data came from 12 additional sources. A representative sample of 21,252 individuals representing Louisiana was used to test the model. The software package MLwiN was used to conduct the multilevel analysis using empirical Bayes Markov chain Monte Carlo (MCMC) estimation. Steps in model building included (a) fitting the unconditional model, (b) fitting the unconditional growth model, (c) fitting the model with the main effects; and (d) fitting the model with the interaction effects.
Results: In each step, the model fit improved significantly using the DIC statistic. The sample experienced a mean amount of 9.45 (SD=3.49) days over a 14 day period feeling depressed. The mean amount of days increased between 2006 and 2010 from 8.75 (SD=3.62) to 10.23 (SD=3.22). Twenty seven percent (27%) of variability was from between-parish differences, 15% from parish cohort differences over time, and 58% from individuals. Individual predictors that explained most of the differences between individuals were income, employment status, age and health. Controlling for these differences, the parish characteristics explaining most of the differences between the depression levels of parish residents were the amount of disasters occurring in a given year, infrastructure resilience indicators (sheltering needs), institutional resilience indicators (social connectivity), economic resilience indicators (poverty, health access) and community capital indicators (civic involvement).
Conclusion & Implications for policy: The outcome of the study allowed for identification of predictors that makes the individual within a community resilient towards disasters. Multilevel modeling was the desired analysis, since it allowed for the investigation of the complex relationships between community resilience and individual social vulnerability. Findings from the study can contribute to policy and ensure that the necessary tangible and intangible resources are allocated for individuals in need of assistance post-disaster. This will ensure that symptoms such as anxiety and depression are managed by individuals.