15 Flanagan et al.: A Social Vulnerability Index for Disaster Management
A unique toolkit consisting of SVI data along with a simple mapping
application was initially distributed to 24 state and local public health departments
for review and feedback. The toolkit, which is flexible and easy to understand,
provides readily accessible data, including the following for each tract, or
“community,” in the United States: 1) an SVI value for each of the 15 census
variables, 2) an SVI value for each of the four overarching domains, 3) an overall
SVI, and 4) flags representing a percentile rank of 90 or higher for each of the 15
variables, for each of the four overarching domains, and for the total number of
flags for each tract. Toolkit user feedback and the identification of useful data and
applications are enabling the evolution of the SVI. For example, users suggested
the calculation of state-based indices, in addition to national-level indices, for
more meaningful within-state comparisons. State-based indices have been
calculated and will be included with the next version of the SVI package. Users
also requested raw census data for each of the variables, e.g., the total number of
persons in poverty in each tract, for targeted interventions. Raw census data will
also be added to the SVI toolkit. Additional socioeconomic variables
recommended for use in various vulnerability indexes, such as living alone, may
be added to the SVI variable set. We may also include in the toolkit map layers
such as nursing homes, hospitals, schools, and other facilities that house socially
vulnerable populations. In addition, custom mapping tools will be included, such
as a tool to estimate population numbers within study areas that cut across census
tracts or tools that return facility counts for neighborhoods.
The SVI is flexible, for use in different phases of the disaster cycle and for
different event types, depending on how best the readily accessible components fit
the user’s needs. Although the researchers in a recent study on heat vulnerability
did not use the SVI, they did incorporate variables of social vulnerability, such as
age, poverty, income, education, race and ethnicity, and living alone, with health
data, vegetation cover, household air conditioning data, and climate data (a
combination of vulnerability, resource, and hazard data) to identify areas for
intervention and further investigation (Reid et al. 2009). Future studies could
employ components of the SVI to do similar modeling, combining SVI data with
other data to more completely specify explanatory variables in risk models for
understanding and predicting disaster event outcomes.
Nevertheless, using the SVI has some limitations. One limitation is the
rapidly changing composition of some small-area populations in the intercensal
years. For instance, the present index uses year 2000 census data. Between that
census and 2005, when Katrina struck, much public housing in New Orleans
underwent major renovation, including demolition of many older multi-unit
structures and their replacement with one- or two-household structures. Many
people in these developments were at least temporarily relocated, so that the 2000
census data for those tracts were inaccurate by 2005. Similarly, the addition of
new subdivisions in suburban counties can quickly produce a significant
Published by Berkeley Electronic Press, 2011