This dataset uses Census Data following published social vulnerability index literature to provide an index at the Place level.
The Corps of Engineers has chosen SoVI as the “foundational SVA (Social Vulnerability Analysis) method for characterizing social vulnerability….” (Dunning and Durden 2013) The University of South Carolina has provided extensive and historic data for this methodology. Susan Cutter and her team have published their methodology and continue to maintain their database. Thus it was chosen as the “primary tool for [Army] Corps SVA applications.” (ibid) The downside is that this method is complex and hard to communicate and understand at times. (S. Cutter, Boruff, and Shirley 2003) The Social Vulnerability Index (SoVI) for this study was constructed at the U.S. Census Place level for the state of Utah. We utilized the conventions put forth by Cutter (2011) as closely as possible using the five-year American Community Survey (ACS) data from 2008 to 2012. The ACS collects a different, more expansive set of variables than the Census Long Form utilized in Cutter et al. (2003), which required some deviation in variable selection from the original method. However, Holand and Lujala (2013) demonstrated that the SoVI could be constructed using regional contextually appropriate variables rather than the specific variables presented by Cutter et al. (2003). Where possible, variables were selected which matched with the Cutter et al. (2003) work. The Principle Components Analysis was conducted using the statistical software R version 3.2.3 (R 2015) and the prcomp function. Using the Cutter (2011) conventions for component selection, we chose to use the first ten principle components which explained 76% of the variance in the data. Once the components were selected, we assessed the correlation coefficients for each component and determined the tendency (how it increases or decreases) of each component for calculating the final index values. With the component tendencies assessed, we created an arithmetic function to calculate the final index scores in ESRI’s ArcGIS software (ESRI 2014). The scores were then classified using an equal interval classification in ArcGIS to produce five classes of vulnerability, ranging from very low to very high. The SoVI constructed for our study is largely consistent with previous indices published by Susan Cutter at a macro scale, which were used as a crude validation for the analysis. The pattern of vulnerability in the state is clustered, with the lowest vulnerability in the most densely populated area of the state, centered on Salt Lake City (see Figure [UT_SoVI.png]). Most of the state falls in the moderate vulnerability class, which is to be expected.