Embed Size (px)
Citation preview
The ACS: Fulfilling its Promise to Data Users
Alfredo Navarro
US Census Bureau
APDU 2010 Annual Conference
Washington, DC
September 21, 2010
2
Outline• Discussion of issues affecting the
reliability of ACS estimates, specifically for small areas and population groups- Sample Design- Mail/CATI Response Rates- Population Controls
• Successes and Challenges
3
ACS Sample Design
• Sample size – about 2.9 M addresses on an annual basis
• Stratification and sample allocation – Similar to census long form
• Unlike census long form, only a sample is selected for personal visit during non-response follow-up
4
Sample Size Effect on Reliability
Census 2000 LF Sampling Rate = 1- in –6Planned ACS Sampling Rate = 1- in - 8
Reliability of ACS estimates relative to LF
1.25
5
Non-response Follow up Sampling Differential Response Rates
• Mail/CATI response rates continue to decline below the levels assumed during the planning phase of the survey
• Substantial variation exists in mail response rates by geography
• Census tracts with high proportions of African American and Hispanic origin populations tend to have lower mail and telephone response rates (Griffin 2005)
Mail/CATI Response RatesCensus Tracts
Source: 2008 and 2009 ACS
Size (Occ. HU’s) Median (%) Range (Q3 – Q1)
0 - 399 32.5 42.3
400 - 999 45.6 32.3
1000 – 1999 56.2 27.4
2000 – 3999 58.4 23.4
4000 – 5999 57.2 18.6
Over 6000 59.5 15.9
6
Effect of Non-response Follow up Sampling (CAPI)
• Estimated from the 2005 -2009 ACS preliminary weighted files
• The range of the median effect is
1.25 – 1.28
7
8
Effect of Population ControlsSource: Small Area Estimates Research
• Research on 1999-2005 ACS data showed that standard errors for small areas – census tracts and places - were much higher than anticipated.
• Lack of tract-level controls identified as a leading cause contributing to an increase of between 15 – 25 percent.
(Starsinic – 2006)
ACS ReliabilitySource: 2005 – 2009 ACS
• The survey was designed to produce estimates with CV about 1.33 the CV of corresponding long form estimates
• The most current results show that the CV is about 1.75 the CV of corresponding long form estimates
9
10
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0-400 400-1000 1001-2000 2001-4000 4001-6000 6001+
CV
Size of Tract (Occupied Housing Units) Source: 2005 – 2009 ACS 5-year Data
Preliminary Tract CV Distribution for % Persons in Poverty
11
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1.1
0-400 400-1000 1001-2000 2001-4000 4001-6000 6001+
CV
Size of Tract (Occupied Housing Units) Source : 2005 – 2009 ACS 5-year Data
Preliminary Tract CV Distribution for % Unemployed
12
ACS Today – Successes and Challenges
• Developed and implemented a sampling plan that over-samples census tracts with lower than average response by mail and phone.
• Developed a model-assisted estimation application that relies on auxiliary information to reduce variance for census tracts and medium-sized places.
• Use of experienced interviewers (including bilingual) facilitate high levels of survey response.
13
Concerns for Estimates of Total Resident Population of Substate Areas
Group Quarters sample designed to produce state-level estimates of characteristics of GQ population
Estimates of characteristics of the GQ population contribute to tabulations of the total resident population
Change in GQ sample design will not have a significant effect on 2010 and 2011 estimates
14
Tracts with GQ Sample (2006‑2009 ACS Sample)
Frequency
Tracts with GQs 43,367
Tracts with ACS GQ sample 21,596
Tracts without ACS GQ
sample
21,771
15
Elmore County, AL, Poverty Rates
2006 Total Household Group Quarters
Poverty Universe 70,428
68,452 1,976
Number in Poverty
9,881 8,103 1,778
Percent in Poverty 14.0%
11.8% 90.0%
2007 Poverty Universe 71,71
271,712 0
Number in Poverty
7,456 7,456 0
Percent in Poverty 10.4%
10.4% 0%
Group Quarters Small Area Estimation Research
• Improving the GQ estimation process, specifically for small areas such as census tracts
• NAS Panel on measuring GQ population
16
Additional Improvements for FY11
• Pending Congressional action the ACS sample will be expanded from 2.9 M to 3.54 M housing unit addresses
• Reallocate the sample through an improved stratification to produce a more equitable distribution of tract level estimates with respect to reliability
17
Additional Improvements … 2
• 100 Percent Follow up of unmailable addresses in American Indian areas with high concentrations of Native Population
• Enhanced Variance Estimates will reduce margins of error by 3- 5 %.
18
