2005 American Community Survey Special Product for the Gulf Coast Area
QUESTIONS AND ANSWERS
The U.S. Census Bureau has released a special data product for Gulf Coast areas affected by Hurricanes Katrina and Rita. This data product is based on data collected from the American Community Survey (ACS). We are providing some responses to questions that may be asked about this special data product.
Do these data give us an official Census Bureau estimate of the post-hurricane population size of New Orleans and other affected areas?
No. The ACS produces estimates of the characteristics of the population and the housing units in which they live. The Census Bureau’s Population Estimates Program produces official Census Bureau estimates of the population size of geographic areas. Concurrent with the release of these data, the Census Bureau is releasing the January 1, 2006 Special Population Estimates for Impacted Counties in the Gulf Coast Area to provide information and indications of the impact of the hurricanes on the population size of affected counties.
What geographic levels does the Gulf Coast Area Special Product cover?
Several geographic levels are available in the 2005 ACS Special Product for the Gulf Coast Area:
1) entire state (for the states of Alabama, Louisiana, Mississippi, and Texas);
2) FEMA-designated Individual and Public Assistance (IPA) area of the state;
3) balance of the state;
4) Metropolitan Statistical Areas;
5) groupings of counties and parishes; and
6) individual counties and parishes.
Did you release data for areas, like Houston, that gained population as well as areas, like New Orleans, that lost population?
The Gulf Coast Area Special Product includes data for FEMA-designated areas that gained or lost population— identified in the response to Question 2, above. For example, areas such as Harris County, Texas, which includes most of the city of Houston, are covered in this release.
Does this product include people in the affected areas who lived in disaster shelters such as the Ernest N. Morial Convention Center in New Orleans, the Domino Sugar factory, or other temporary housing that was intended solely for people displaced by the hurricane?
No. The Gulf Coast Area Special Product universe only includes addresses for housing units that existed before Hurricanes Katrina and Rita struck the Gulf Coast area. The 2005 ACS sample, which was selected prior to the hurricanes, includes only housing units that were in place at the time the sample was selected. The 2005 ACS does not include people living in group quarters or in a temporary group quarter facility such as the Ernest N. Morial Convention Center.
Does this product include people in the affected areas who lived in temporary living quarters such as trailers or tents at housing units addresses?
In some cases. For example, if a trailer was temporarily located at an address that was selected for the survey (e.g., on the site of a house that was demolished), we collected information from the people who were living there.
6. What data products did the Census Bureau release? What characteristics of the affected population are covered in this data product?
The 2005 ACS Special Product for the Gulf Coast Area contains two snapshots:
the first eight months of 2005 (January through August); and
the last four months of 2005 (September through December).
Estimates of general demographic characteristics (e.g., sex and age), selected social characteristics (e.g. educational attainment and residence one year ago), general economic characteristics (e.g., labor force participation and poverty rate), and selected housing characteristics (e.g., percent of housing units that are vacant and median gross rent) are available for both time frames for all published geographies.
How do these special data products differ from the regular 2005 ACS data products that will be released later in 2006?
The regular 2005 ACS data products consist of one-year period estimates. The 2005 ACS Special Product for the Gulf Coast Area represents period estimates for the first eight months and last four months of 2005. For 2005, only the last four months of data collection reflected the impact of the hurricanes. As a result, the regular ACS data products will be affected only to a limited extent by changes taking place in that four-month period. To understand the effects of the hurricanes, data for two periods of time are needed ─ one that reflects the general characteristics in the eight-month period prior to the hurricanes, and another that reflects the general characteristics in the four-month post-hurricane period. The single-year ACS estimates are developed in such a manner to make them consistent with the Census Bureau’s population estimates for July 1 of that year. This step was omitted in developing the four-month and eight-month estimates for the Gulf Coast Area Special Product.
What measures did the Census Bureau take to protect the confidentiality of data about people in the hurricane-affected area?
In advance of the release of any data products, the Census Bureau’s Disclosure Review Board conducts a careful review of proposed data tabulations. In the case of the Gulf Coast Area Special Product, that review helped determine, for example, whether data could be reported for one county, or if, instead, the data for several counties had to be combined and data reported for the combined area. The Census Bureau also used statistical methods to protect individual information, just as it does to protect individual information for all its data products.
9. What limitations do I need to be aware of in understanding the Gulf Coast Area Special Product?
There are three factors that should be considered in understanding these products. The four-month estimates are based on less than one-third of the sample required for ACS single-year estimates. This is because only four of the twelve months of sample are used to produce these estimates. In addition, the sample sizes in these areas were further reduced because of an increase in the number of sample addresses that, when visited, were determined to be ineligible for the survey. Sample ineligibility (or out-of-scope addresses) will occur when units that have been identified in the initial sample are determined to be commercial, nonexistent, or demolished. The level of housing unit destruction caused by these two hurricanes resulted in a large increase in the rate of out-of-scope sample addresses.
A second limitation is coverage error. The final ACS estimates are normally adjusted for coverage error by controlling survey estimates to independent population estimates. For this product, we did not use population controls. In 2004, the coverage rates in these four states were about 92 percent. This means that we expect that the estimates include this level or greater levels of coverage error.
The third type of error, nonresponse, can affect the estimates at the unit or item level. In some parishes and counties, levels of unit nonresponse (the inability to conduct an interview) increased. Response rates, however, for higher levels of geography (e.g., state) remained high (over 90%). Item nonresponse levels are fairly consistent for the two time periods suggesting that when an interview was conducted, equally complete data were collected for the two periods of time.
10. Where can I find the data?
The data are available on the ACS home page of the Census Bureau’s web site: <http://www.census.gov/acs/www>.
11. Are the estimates of population from the Gulf Coast Area Special Product different from the Special Population Estimates for Impacted Counties in the Gulf Coast Area?
Yes, because they were developed in different ways. The ACS estimates are based on data collected from housing unit addresses as part of a survey conducted every month. They describe two period estimates in 2005. The Census Bureau designed a special methodology to produce the January 1, 2006 Special Population Estimates for Impacted Counties in the Gulf Coast Area. These estimates are a blend of the Census Bureau’s extrapolated household population estimates to January 1, 2006 without any impact of the hurricanes and cumulative net migration estimates as of January 1, 2006, derived from a change of address file from the United States Postal Service (USPS).
12. Why aren’t the data for each county or parish shown individually rather than being combined with the data for other counties or parishes?
In order to protect the confidentiality of respondents in the affected areas and maintain a high level of data quality, we required a minimum number of interviewed people in the September through December 2005 sample for each geographic area. If the population of a given county or parish did not meet that threshold, we combined its data with data from other geographic areas, and presented data for the combined area.
13. How did you choose the specific geographic areas included in this special data product?
We decided to use the geographic boundaries of the FEMA-designated Individual and Public Assistance (IPA) areas as the universe for this set of products. We consulted with local and state partners to identify counties, parishes, and grouping of counties and parishes of interest. We assessed the feasibility of producing products for these areas considering our commitments to protecting confidentiality and maintaining data quality.
14. Did you use the same data collections; data processing, and estimation methods that you used to produce regular ACS data tabulations?
In the months immediately following the hurricanes, the Census Bureau modified its data collection practices. In the areas most affected by hurricanes, we did not mail a replacement ACS mailing package to the September sample addresses. For the October sample, we delayed or did not mail any of the mailing items. For the November sample, the ACS resumed the standard mail procedures. We modified the prenotice letter starting with the October sample, and modified the reminder cards and the letters sent with the two mailing packages starting with the November sample. These modifications clarified to respondents that they should be included on the ACS questionnaire if they were displaced. We also modified our computer assisted telephone interview (CATI) procedures, and computer assisted personal interview (CAPI) procedures based on information about emergency zones provided by FEMA.
We followed standard data processing methods, but editing and imputation were done separately within each of the two data sets. We made two changes to the methodology and estimation methods. The weighting method was modified to exclude the final weighting adjustment based on the use of the Census Bureau’s official population estimates. Therefore, estimates included in this data product do not include the usual adjustment to correct for housing and population coverage.
15. How should I interpret the four-month and eight-month estimates?
All estimates are period estimates designed to describe either a four-month or eight-month time period.
16. Did the Census Bureau produce an analytic report summarizing these results?
No. These data were produced to assist the Gulf Coast area in identifying emergency recovery needs. We hope that analysts in these areas will produce their own reports. We provided data, maps, and supplementary explanatory materials as guidance on the use and interpretation of the Gulf Coast Area Special Product.
17. Did you release data for group quarters (e.g., colleges, prisons, and nursing homes)?
No. The universe for the 2005 ACS Special Product for the Gulf Coast Area consists only of living quarters that meet the definition of a housing unit. Characteristics of persons living in group quarters facilities such as colleges, jails, and military housing are not reflected in the Gulf Coast Area Special Product. The ACS did not begin data collection from group quarters until January 2006.
18. Did you collect information about vacant housing units?
Yes. The ACS procedures include data collection from both occupied and vacant housing units. We collected as much information as we could about vacant housing units in the sample.
19. Did you collect information from everyone living in the area affected by the hurricanes?
We only collected information for housing unit addresses included in the sample. We collected information from everyone living in these areas who was eligible to be interviewed under the ACS residence rules. This included people in housing units who had lived in the area for at least two months and people who were temporarily living in a sample housing unit because they had no other place to live.