Selecting Longitudinal Community Outcomes for Resilience Indicator Validation

Abstract

Having a scientifically grounded set of community resilience indicators has been a long-sought goal in multiple fields of study and practice. Community resilience is a complex system consisting of emergent properties. Accordingly, community resilience frameworks have been designed to estimate the level of resilience through varying sets of cross-sectional indicators. However, current framework designs are not well suited for tracking longitudinal outcomes of resilience, such as whether a community has mitigated the initial impact and recovered from the decline. While many such frameworks have been suggested as a solution to quantify the resilience levels of communities, research has not yet examined outcomes of community resilience to establish a dependent variable or a benchmark that can be used to validate individual baseline community resilience indicators and composite indicators in the existing frameworks. This study examines community resilience outcomes to reflect key elements of resilience definitions, such as robustness and recovery. Among the 2,140 counties that experienced one or more presidential disaster declaration events between 2002 and 2020, this study reports on the analysis of 995 counties in the US that experienced a single event during the observation period. The outcome indicators were designed to present archetypes of community changes based on pre- and postevent values. For each event, the outcome indicators were analyzed within a 7-year window, from 2 years prior to the event to 5 years after. A county can be categorized as resisted, recovered, or not recovered, in terms of the county’s ability to mitigate the initial impact and bounce back. This study presented 14 outcome indicators representing four domains of community resilience (population, economy, social vulnerability, and health). Then, four outcome indicators (population, employment, eviction rate, and life expectancy) were selected based on criteria including the distribution and expected connection to damage and preevent community characteristics. Generalized structural equation model (SEM) and hierarchical cluster analysis methods were used to quantify a hypothetical construct of resilience outcomes. The estimated outcomes provide an intuitive and practical summary of observed longitudinal community responses to disruption, and support researchers and practitioners to test the quality of resilience indicators through a validation method rooted in empirical evidence.