To compare social vulnerability to climate change among areas within the state, we created an index that combines a number of individual vulnerability factors into a single, composite indicator. Our Social Vulnerability Index is useful for assessing overall vulnerability and comparing areas within the state. It should help policymakers to identify areas where efforts are especially needed to build community resilience to climate-related impacts.
We compiled the social vulnerability index at the Census Tract level. We calculated a vulnerability index value for each of the 7,049 census tracts in the state (which, in 2000, averaged about 5,000 people in each). We used the census tract boundaries from the 2000 Decennial Census, rather than the more recent 2010 Census boundaries. We used the older boundaries because much of the data we used was collected from 2005–2009 and was compiled according to the year-2000 census tract boundaries.
The Census Bureau publishes some variables, like population and race, at a smaller scale, such as Census Blocks or Block Groups. For much of the data we collected, Tracts were the smallest geographic boundary at which the data was aggregated.
One must be careful when interpreting the results of our analysis. Whenever a population’s characteristics are averaged across large areas, smoothing is inevitable and local variation within that area is covered up. A Census tract with a low social vulnerability index indicates a populace that is above average in terms of health and income. Within the tract, however, there may be families living in poverty, senior citizens, or disabled persons that are highly vulnerable. These residents are likely to require assistance in responding to and recovering from a natural disaster or other disturbances.
Our vulnerability index combines 19 individual vulnerability factors. In reality, there are many factors that may contribute to a region's vulnerability, but for which reliable data have not been collected or published. For example, the level of community organization in an area will affect the population’s ability to respond to and recover from climate change-related impacts.
Each of the 19 variables was measured and reported in its own units (for example, the number of low-income residents, or percent impervious cover). It does not make sense to add up variables that are expressed in different units. So, as a first step, we used a statistical method to transform each variable. We standardized the variables by calculating z-scores, which is a measurement of an observation's distance from the average. This method was used before us by researchers at the University of South Carolina in developing a similar index (Cutter et al. 2003).
With each of the variables reported here, a higher value means the region is more vulnerable. For example, a region with a greater percentage of low-income households will be more vulnerable. As another example, a high percentage of households without air conditioners means the region's populace is more vulnerable to heat stress.
Some variables were transformed from their original units so that higher values correspond to greater vulnerability. For example, we aquired data on the number of high school graduates in each census tract. We transformed this from “percentage of high-school graduates” to “percentage without a high-school diploma,” which corresponds with higher social vulnerability.
Once all of the variables were transformed, we averaged the z-scores for the 19 variables in each tract to create the social vulnerability index score for the tract. This calculation was carried out for each of the 7,049 census tracts in the state. For tracts with missing data, we created the index by averaging the available observations. Lower values of the index mean lower vulnerability, and higher values indicate greater vulnerability.
To compare social vulnerability among areas in the state, we grouped the tracts into thirds. Those tracts with index scores in the bottom third were considered “Low Vulnerability.” Those in the middle third are “Medium Vulnerability,” and the highest third is “High Vulnerability.”
| Vulnerability Factor | Indicator | Data Source |
|---|---|---|
| Households with air conditioning | Households with an air conditioning unit | Roberts 2011a |
| Population over 25 with a diploma | People over age 25 who have a high school diploma | U.S. Census, American Community Survey (2005-9) |
| Born outside the U.S. | People who were born outside the United States | U.S. Census, American Community Survey (2005-9) |
| Impervious areas | Land in the area that has an impervious surface (e.g. sidewalk or roof) | EPA 2001 |
| Residents living in institutions | Population living in “group quarters,” including institutions like correctional facilities, nursing homes, and mental hospitals, college dormitories, military barracks, group homes, missions, and shelters. | U.S. Census, American Community Survey (2005-9) |
| Households with limited English | Population 5 years and over who answered that they speak English less than “very well” | U.S. Census, American Community Survey (2005-9) |
| Households with no vehicle | Percentage of households with no vehicle available | U.S. Census, American Community Survey (2005-9) |
| People of color | People identifying as any other race or ethnicity besides white. | U.S. Census, American Community Survey (2005-9) |
| Households in poverty | Households with an income that is below 200% of the official federal poverty level | U.S. Census, American Community Survey (2005-9) |
| Pre-term births | Infants that were born before completing 37 weeks (about 8.5 months) of pregnancy | Roberts 2011b |
| Renter-occupied households | Percent of households where people are renting | U.S. Census, American Community Survey (2005-9) |
| Over 65 and living alone | Percent of households occupied by someone over age 65 who lives alone | U.S. Census, American Community Survey (2005-9) |
| Tree canopy cover | Land covered by tree canopy | Calculated by Jesdale (2011) using data from Multi-Resolution Land Characteristics Consortium (2001) |
| Under age 18 | Population under age 18 | U.S. Census, American Community Survey (2005-9) |
| Unemployment | Population 16 years and over able to work who are unemployed | U.S. Census, American Community Survey (2005-9) |
| Have jobs working outdoors | Percent of workers who work in agriculture, forestry, mining, or construction | U.S. Census, American Community Survey (2005-9) |
| Pregnancy | Percentage of women 15 to 50 years old who had a birth in the past 12 months | U.S. Census, American Community Survey (2005-9) |
| Food access | Access to full-service supermarkets according to Low Access Area measurement tool | The Reinvestment Fund 2010 |
| Youth fitness | Fraction of children that are overweight or obese in tract (i.e. fraction over 85th percentile for age and gender based on the CDC growth curves. | Ortega Hinojosa 2011 |
Environmental Protection Agency. 2001. National Land Cover Data. Multi-Resolution Land Characteristics Consortium. Washington, D.C. http://www.epa.gov/mrlc/nlcd-2001.html
Ortega Hinojosa, A. M. 2011. Digital data files sent by email. Doctoral Candidate, UC Berkeley School of Public Health, Division Environmental Health Sciences.
The Reinvestment Fund. 2010. TRF Supermarket Study of Low Access Areas. Accessed July 2011 from http://www.trfund.com/TRF-food-access.html.
Roberts, E. 2011a. Analysis of California Energy Commission survey of California household air conditioning unit. Personnel Communication, Research Manager, California Environmental Health Tracking Program, California Department of Public Health.
Roberts, E. 2011b. Preterm birth rates by Census Tract for California, 2006. Personnel Communication. Research Manager, California Environmental Health Tracking Program, California Department of Public Health.
United States Census Bureau (2010). 2005-2009 American Community Survey 5-Year Estimates Summary File Tracts and Block Groups. https://explore.data.gov/Population/2005-2009-American-Community-Survey-5-Year-Estimat/jhya-8c2t