Heading 1
Methodology for Historic Storm Dashboards
Goals: The goal of the dashboards is to present FEMA data in a way that helps disaster survivors, advocates, and public officials understand how a disaster affected their community and whether FEMA’s response was effective and equitable. We identified four storms to analyze in states and territories represented by our partners; Superstorm Sandy in New Jersey in 2012, Hurricane Maria in Puerto Rico in 2017, Hurricane Harvey in Texas in 2017, and Hurricane Ida in Louisiana in 2021. Due to Harris County’s size and population - 4.7 million people in 2022 - a separate dashboard was created for Hurricane Harvey in Harris County.
On the state/territory level dashboards, data is presented at both the county or municipio level and at a smaller geography level. County-level data give an overview of impact at the state level but can obscure potentially significant differences within counties. For instance, though Harris County as a whole recorded 29 applications per 100 households, some zip codes recorded as many as 98 applications per 100 households.
Data Source: The principal data source we used is the FEMA Individuals and Households Program Valid Registrations Dataset, available from OpenFEMA (the IHP dataset). This dataset records application data from FEMA’s Individuals and Households Program (IHP), which provides financial and housing assistance to eligible households affected by a disaster with losses and serious needs that are not covered by insurance. IHP provides Housing Assistance in the form of rental assistance, home repair funds, and direct housing assistance like travel trailers or manufactured homes. IHP also includes Other Needs Assistance (ONA) for disaster-related needs like medical costs, funeral costs, and transportation.
As of February 2024, the IHP dataset included over 20 million applications, making it among the most robust datasets for evaluating recovery. In addition to an identifier for the disaster code, each row includes:
Geographic data: state, county, and ZIP code
Socioeconomic Data: household income; household size, housing tenure (owner/renter), type of housing unit
Damage levels: total verified loss for damage to personal and real property
Application outcomes: eligibility, amount of assistance, reason for denial. Eligibility and amount of assistance are provided for both Other Needs Assistance and Housing Assistance. A household is marked eligible for Housing Assistance only if they receive financial assistance; eligibility for Transitional Sheltering Assistance is considered separately, and Direct Temporary Housing assistance is not indicated. A reason for denial is only given for applications referred for Housing Assistance.
The major storms we reviewed had between 250,000 and 1.3 million rows of data. For each dashboard, we filtered data by disaster code to be specific to the storm and state or territory.
Disaster Code | Storm Name | Affected Area | Year |
|---|---|---|---|
4611 | Hurricane Ida | Louisiana | 2021 |
4339 | Hurricane Maria | Puerto Rico | 2017 |
4332 | Hurricane Harvey | Harris County | 2017 |
4332 | Hurricane Harvey | Texas | 2017 |
4086 | Superstorm Sandy | New Jersey | 2012 |
For the Harvey in Harris County dashboard, we also filtered data by the field “County = Harris.”
With the exceptions noted below, the raw data was used.
Notes to Individual Assistance per 100 Applications
The goal of this map is to show the geographic spread of storm impact. To do this we mapped aggregate applications adjusted by the total number of households. We consider raw applications a more reliable indicator of need following a hurricane because of program and implementation changes across disasters. Each FEMA IHP application is a statement of need by a household and mapping the aggregate distribution of applications shows impact throughout the region.
We aggregate applications at the zip code and county geography levels because those were the two standardized geographies available. The IHP dataset also records the city the application was filed in, however, city limits are less precise and applicants may identify themselves as living in a city even though they are not technically within its census-designated boundaries. The total number of applications per geography was divided by the number of households in that geography according to the US Census, using table DP02 from the 5-year American Community Survey at the 5-Digit ZIP Code Tabulation Area (ZCTA5) and county geography levels. The ACS data year chosen was the year of the disaster to reflect impact at the time the disaster hit. This is particularly relevant in parts of Harris County which have experienced significant population growth as well as parts of Puerto Rico that have lost population, where results may have been skewed by the population change.
A small number of applications were filed from zip codes corresponding to P.O. Boxes. This led to anomalous results because the Census does not record households for P.O. Boxes. If a P.O. Box had more than five recorded applications, the number of applications was added to the zip code that surrounded it; P.O. Boxes with fewer than five applications were not added to adjacent zip codes: for each storm this amounted to fewer than 100 total applications not reflected on the map. The applications were included in total application numbers and all other tables.
Finally, there were a small number of applications with a zip code in a state different than the one in which the disaster occurred. These were not mapped, though they were included in application totals and other tables. For instance, for Hurricane Harvey, 890,923 applications were recorded from zip codes in Texas, one was recorded from a zip code in Louisiana, and one was recorded from Missouri.
Social Vulnerability Index by Census Tract
The Social Vulnerability Index allows advocates and planners to identify communities that may need support before, during, or after disasters. There are multiple indices available: we chose to use the Centers for Disease Control and Prevention/Agency for Toxic Substances and Disease Registry (CDC/ATSDR SVI) because the data is publicly available. The CDC/ASTDR SVI is also used by the National Risk Index. The CDC/ATSDR SVI estimates vulnerability using 16 variables, in four themes: Socioeconomic Status, Household Characteristics, Racial & Ethnic minority Status, and Housing and Transportation Types. A raw score is calculated for each theme, and then a composite score. To calculate the final score, a percentile is assigned to each census/county tract based on the ranking by composite score, with 0 being the least vulnerable and 1 being the most vulnerable. For instance, in a state with ten counties the county with the lowest SVI would have a score of 0, the second lowest would have a score of .1 etc.
For each census tract and county, we used the estimated percentage of the population that the SVI defines as having "Racial and Ethnic Minority Status", which combines seven variables, divided by the estimated total population, to calculate the percentage of the population that is Black, Indigenous, and other people of color (BIPOC). The DEDP uses the term BIPOC instead of "minority".
The year used for SVI was 2022, the most recent available when the dashboards were created. We used the statewide geographic comparison, wherein tracks are ranked only with other tracts/counties in that state or territory rather than with all tracts/counties in the United States.
Notes to Bivariate Map of Social Vulnerability and Damage
The goal of the bivariate map is to combine social vulnerability and storm impact to highlight the communities that were most impacted by a disaster but may have the fewest resources and least ability to recover. As with the other state-level dashboards, we provided data at a large and small geography.
On the bivariate map, the small level geography units are units we created and not existing geographic units like zip codes or census tracts. This was necessary because the metrics were only available at different geographic levels; applications per 100 households at the zip code level and SVI at the census tract level. We used the intersect command in ArcMap 10.8 to overlay zip codes on census tracts to create smaller polygons, each corresponding to one zip code and one census tract. These polygons were coded according to whether the zip code was in the top or bottom half of the number of FEMA applications per 100 households and the top or bottom half of social vulnerability, i.e. SVI greater or less than .5. This analysis created 4 categories: 1) high damage, high vulnerability, 2) low damage, high vulnerability, 3) high damage, low vulnerability, and 4) low damage, low vulnerability.
Counties were similarly coded and categorized.
Notes to Equity and Outcomes:
Unless otherwise noted, the complete IHP dataset for the disaster was used, and the denominator for each rate is the total number of applications.
Eligibility rate is defined as the number of eligible applications (ihpeligible = 1 in the IHP dataset) divided by the total number of applications. An application is marked as eligible only if the applicant receives financial assistance, either financial Housing Assistance or Other Needs Assistance. In the IHP dataset, FEMA does not mark applicants admitted into a direct housing program, such as Temporary Sheltering Assistance (TSA), as eligible unless they also receive financial assistance.
Rejections for failure to prove ownership: If an application was referred for housing assistance, the field “hastatus” records the outcome of an applicant’s housing application status, including rejection codes if the application was rejected. An application can be rejected for more than one reason. This rate includes the total number of applications with “IOWNV - Ineligible, Ownership Not Verified” as a rejection code even if the application has more than one rejection code. The denominator is applications from owners that were referred for housing assistance. If applicants were denied Housing Assistance but eligible for Transitional Sheltering Assistance FEMA does not record a rejection code for that application. FEMA does not record a denial code for applications that were not referred for Housing Assistance, so the reasons applicants were denied Other Needs Assistance are unknown.
Applications by Owner/Renter - Less than 1% of applications have “Unknown” in the field ownrent. For visual clarity, these applications were excluded from the pie chart; however the applications are still included in the total numbers and all other tables.
Eligibility by Owner/Renter - Applications with “Unknown” in the ownrent field are excluded from the chart for visual clarity.
Applications by Income - The FEMA data have 7 income categories: 1) More than $175,000, 2) $125,000 to $175,000; 3) 60,000 to $125,000; 4) $30,000 to $60,000; 5) $15,000 to $30,000; 6) Less than $15,000 and 7) 0. FEMA’s metadata explain that “A value of 0 indicates applicants who either refused to provide their income, or in some cases reported they were self-employed.” Accordingly, rows with a 0 are summarized as “not reported/self-employed.”
The categories of reported income have been combined from six into three categories - $0 to $30,000; $30,000 to $120,000; and more than $120,000. Note that these categories do not correspond to any official designations of high, medium, or low income; rather, this was the simplest way to represent the data. Because there are only 7 categories, IHP data cannot be easily compared with HUD metrics like Area Median Income.
Average Assistance Amount - This is the amount of money received by applicants who have been approved. The denominator is only applicants who received financial assistance from FEMA.
Citations
Federal Emergency Management Agency (FEMA), OpenFEMA Dataset: Individuals and Households Program – v1 Retrieved from https://www.fema.gov/openfema-data-page/individuals-and-households-program-valid-registrations-v1 on March 17, 2024. This product uses the FEMA OpenFEMA API, but is not endorsed by FEMA. The Federal Government or FEMA cannot vouch for the data or analyses derived from these data after the data have been retrieved from the Agency's website(s).
Centers for Disease Control and Prevention/ Agency for Toxic Substances and Disease Registry/ Geospatial Research, Analysis, and Services Program. CDC/ATSDR Social Vulnerability Index 2020, Database Texas. https://www.atsdr.cdc.gov/placeandhealth/svi/data_documentation_download.html. Accessed on March 30, 2024.
U.S. Census Bureau; American Community Survey, 2013-2017 American Community Survey 5-Year Estimates, Tables DP05 & DP02. Retrieved by John Laycock, March 18, 2024
FEMA. “Individual Assistance Program and Policy Guide (IAPPG)” Version 1.1 May 2021.
Retrieved from: https://www.fema.gov/sites/default/files/documents/fema_iappg-1.1.pdf 9/24/2024
How to cite these dashboards:
Any citation should include
- the authors: John Laycock & Maddie Sloan
- website name: Disaster Equity Data Portal
- Name of the dashboard being cited
- URL of the dashboard
- Date Retrieved.
Example:
Laycock, John & Sloan. Maddie. Disaster Equity Data Portal. "Hurricane Sandy Impacts in New Jersey." Retrieved from https://public.tableau.com/views/HurricaneSandyImpactsinNewJersey-BETA/Welcome?:language=en-US&:sid=&:redirect=auth&:display_count=n&:origin=viz_share_link" 9.25.2024