1 Arnold Jackson Associate Director for Decennial Census U.S. Census Bureau Texas Economic and Demographic Association September 16, 2010 2010 Census Progress.

1

Arnold JacksonAssociate Director for Decennial Census

U.S. Census Bureau

Texas Economic and Demographic AssociationSeptember 16, 2010

2010 Census Progress Report

2

Agenda

• Census Background

• Pre-Enumeration Activities

• Enumeration Activities

• Response Rates

• Nonresponse Followup

• Post Enumeration Activities

• 2020?

3

The Importance of Census Data

• Census data helps ensure that each community receives its fair share of political representation through Congressional Apportionment and Legislative Redistricting

• Census data affects how more than $400 billion per year in federal funding is allocated to local, state and tribal governments – $4 trillion over a decade.

• Census data also affects programs and decisions:

o At the local level: roads, schools, emergency services, etc. o Private sector – business decisions.o Research sector – research on demographic changes and trends.

4

How Big Is The 2010 Census?

• 309 million people

• 134 million housing units

• 235,000 group quarters (e.g., prisons, college dormitories, military barracks) to enumerate

• Over 39,000 jurisdictions (tribal, state, county, and local governments)

• 9 million census blocks

• 12 temporary Regional Census Centers and Puerto Rico Area Office

• 494 temporary Local Census Offices

• Filling approximately 1.4 million positions

5

The 2010 Census Cycle

• Research and Testing: 2002 – 2008

• Nationwide Address Canvassing: April 2009 - July 2009

• Questionnaire Delivery: March 2010

• CENSUS DAY - APRIL 1, 2010

• Nonresponse Followup: May 2010 - July 2010

• Deliver apportionment count to the President by December 31, 2010

• Deliver Redistricting Data to the States by April 1, 2011

• Release of all remaining data products June 2011-August 2013

• Complete Research, Evaluations, and Experiments Program by September 2013

6

Pre-Enumeration Activities

• Participating State and Local Governments Conducted Local Update of Census Addresses; November 2007 - May 2008.

• Opened 12 RCCs, Puerto Rico Area Office, and 150 Early LCOs to support Address Canvassing and Group Quarters Validation operations

• Conducted Address Canvassing; March – July 2009

• Conducted Group Quarters Validation; September – October 2009

• Opened 344 additional LCOs to support remaining field operations; August 2009 – December 2009

7

Address Canvassing Operation

• 38.4 million addresses were submitted as adds by Local Update of Census Addresses (LUCA) participants. 79% of those adds matched to existing addresses in our Master Address File (MAF); 21% (or approximately 8 million addresses) were added as new records to our MAF and then verified during Address Canvassing.

• Over 140,000 field workers checked 145 million addresses using handheld computers.

• The use of handheld computers gave us the ability to capture GPS coordinates for most of the nation's housing units, which should greatly reduce the number of geographic coding errors caused by using paper maps in previous counts.

• During the Address Canvassing operation, field listers were able to verify addresses, add addresses, delete addresses, change addresses, identify an address as a duplicate, etc.

• A sample of each Assignment Area’s work was checked for accuracy and all deletes were verified by a second field worker.

8

Address Canvassing Operation

• Total Address Canvassing actions applied to the MAF were 155.4 million.

o About 63% were verified “as is”o About 14% of the address records were a delete, duplicate, or non-residentialo About 7% of the address records were added to the MAF as a new recordo About 16% of the addresses involved some change to the address

• After updating the MAF with the results of Address Canvassing, approximately 134.4 million addresses were in the initial universe for the enumeration of the 2010 Census.

• Comparisons of the 2010 Census initial enumeration universe to independent housing unit estimates for the July 2009 time period, using July 2008 estimates projected to July 2009 at the national level, are 2.8 million or 2.1% higher than the independent estimates.

• Comparisons prepared for Census 2000 at a similar point in time showed the census universe to be 6 million or 5% higher than the independent housing unit estimates.

9

Enumeration Activities

• We have successfully completed all enumeration activities including :

• The Mailout/Mailback operation,

• Remote Alaska,

• Update/Enumerate,

• Group Quarters Enumeration,

• Shipboard Vessel Enumeration,

• Nonresponse Followup, and

• Vacant Delete Check.

10

Response and Participation Rates

Census Year As of Date 1 Response Rate 2 Participation Rate 2

2010 4/27/2010 64.55% 72.32%

2000 4/25/2010 67.20% 73.30%

1 Date differential is due to offset based on individual mailout start dates

2The mail response rate is calculated by dividing mail returns by the total number of forms mailed out or delivered by us in Update/Leave areas. The denominator for the participation rate subtracts forms reported as undeliverable (mostly vacant units) by the U.S. Postal Service. This rate gives a better sense of public cooperation.

11

15

13

Directing Resources Towards Populations with Low Participation Rates

• Tracts with high concentrations of Asian, Black, and Hispanic housing units show low response rates (relative to non‐minority tracts and/or their 2000 response rates).

• Advertising resources can be directed at the National level and the DMA level.

• DMAs were ranked by participation rate and the number of people who have not yet participated.

• For each of the three groups, each DMA gets an "average combined rank” of that DMA’s ranking by

– Participation rate in tracts with high concentrations of the group of interest within the DMA– The non‐responding population of interest in the DMA.

• Using these three priority lists, Census worked with consultants to determine which of the priority DMAs have media mechanisms available for the specific group in question.

• The slides that follow drill down to examples drawn from the priority list of DMAs. When the lists were created, we found the following:

• Asian: Seattle‐Tacoma was in the top 20 priority DMAs for tracts with high concentration of Asians.

• Black: New York was in the top 5 priority DMAs for tracts with high concentration of Blacks.• Hispanic: Houston was in the top 5 priority DMAs for tracts with high concentration of

Hispanics.

14

2010 Census – A Significant Achievement

• Slightly more than 72 percent of U.S. households believed to be occupied mailed back their 2010 Census forms.

• The nation responded tremendously despite trends over the past decade toward declining survey participation, a more diverse population, a difficult economic environment, and a growing distrust of government.

• To encourage mail participation, the Census Bureau challenged communities nationwide to improve the participation rates they achieved in Census 2000. It resulted in:

o 28 states met or exceeded their Census 2000 rates. o 11 states were within one point of matching their rates.

o Texas had a participation rate of 69%, a 1% increase from Census 2000.

15

• Nonresponse Followup (NRFU) was conducted May 1 – July 10.

• The NRFU operation was conducted in areas where questionnaires are mailed through the U.S. Postal Service, or hand-delivered by census employees during the Update Leave (UL) operation.

• In both situations, residents are asked to fill out the questionnaire and return it by mail. If we do not receive a questionnaire by the time field assignments are made, we conduct NRFU at that address.

Nonresponse Followup (NRFU) Overview

16

Historical Background of the Paper-Based Operations and Control System

• Initial plans for the 2010 Census called for the use of handheld computers for collection of census data in an automated NRFU.

• In April 2008, in order to reduce risk, we decided to eliminate the automated NRFU and the Census Bureau had to re-plan the entire NRFU to use paper instead and thus requiring a Paper-Based Operations and Control System (PBOCS).

• After the decision, initial plans called for the FDCA contractor to develop the PBOCS software.

• By fall of 2008, the Census Bureau assumed the responsibility for developing the PBOCS for most census operations.

Nonresponse Followup (NRFU) Challenges

17

We took the following steps to ensure NRFU enumeration proceeded on schedule despite the Paper-Based Operations Control System (PBOCS) experiencing some stability issues at the beginning of the operation. These steps facilitated NRFU training and enumeration to begin on time in all LCOs.

Printing Assignment Materials

• The PBOCS was taken down for LCO/RCC users during the period when the NRFU universe was ingested to ensure the ingest was successful.

• The LCOs began the printing of assignments first with only the even-numbered AAs; then later with the odd-numbered ones. This ensured that all enumerators had assignments to begin their work with.

• Crew leader assignments were made manually if needed, and then entered into the system as time allowed.

Nonresponse Followup (NRFU) Challenges

18

Check-In Completed NRFU Work

• The NRFU was given priority to the LCO users. Other operations, since they were nearly complete, and were either completed outside the PBOCS system or completed nightly during the hours of midnight to 2 a.m.

• The NFRU shipping functionality was removed and a contingent, automated solution was invoked.

• These steps reduced the competition on the system caused by multiple operations and functions occurring concurrently and expedited the check-in of NRFU questionnaires, thus reducing the backlog in the LCOs.

Nonresponse Followup (NRFU) Challenges

19

NRFU Workload and Staffing

• The Decennial Systems Processing Office (DSPO) started creating the initial workload. As of April 7, there were 57.7 million cases in the initial workload.

• The initial workload was reduced by 9.1 million late mail return cases received between April 7 and April 19.

• At the start of NRFU, the national workload estimate was 48.6 million housing units.

• Almost 2 million late mail returns were clerically removed during the operation.

• Estimated cost $2.1 billion.

• The national staffing will include more than 600,000 trained staff, including Field Operations Supervisors, Crew Leaders, Crew Leader Assistants, and enumerators.

20

Post Enumeration Activities

• Completed Post Enumeration Activities include:

• The Field Verification operation, which is done primarily to confirm the existence of addresses provided by respondents on Be Counted forms and through our Telephone Questionnaire Assistance Centers.

• Coverage Followup, where we re-contacted households that had some indication that people were either counted in error or were incorrectly left off the form.

• Data capture of forms to be included in the Census counts.

• Review of the 2010 Census Group Quarters results, focusing on identifying GQs missing from the census or located in the wrong census collection block, by Federal State Cooperative Program for Population Estimates demographers from participating states.

21

Post Enumeration Activities

• Ongoing Post Enumeration Activities include:• Process all of the captured data, including edits and imputation.

• Close down field infrastructure. The closeout of the Local Census Offices will begin by October 4 and is targeted to end by November 30

• Plan and conduct close down of the Regional Census Centers. RCCs will be closed by December 30, 2011.

• Complete the Census Coverage Measurement (CCM) program

• Release 2010 Census Data Products

• Complete 2010 Census Program for Evaluations and Experiments (CPEX)

22

Census Coverage Measurement

• The Census Coverage Measurement (CCM) program is our primary mechanism for measuring how well we did with the actual count.

• We are currently in the field conducting the CCM Person Interview, an independent interview of households that tries to determine who should have been enumerated.

• The final CCM field operation is the CCM Final Housing Unit Followup, which begins May 5, 2011.

• We are scheduled to release coverage estimates from the CCM program by September 28, 2012.

23

Data Products

• Schedule of Release of Numerous 2010 Census Data Products includes:

• TIGER/Line geographic shape files – Shipped on a state by state flow basis to officials beginning in late November 2010 and finished by January 31, 2011.

• Apportionment Count- Delivered to the President no later than December 31, 2010.

• P.L. 94-171 Redistricting Summary File - Shipped on a State by State basis beginning in early February, 2011 and finished by March 31, 2011.

• Map Suite and Block Assignment Files - Shipped on a State by State basis beginning mid-February, 2011 and finishing by March 1st, 2011

• Demographic Profiles - May 2011• Summary File 1 - June-August 2011• Summary File 2 -December 2011-April 2012• Congressional District Summary File for the 113th Congress - January 2013• Legislative Districts - June 2013

24

2010 Census Program for Evaluations and Experiments (CPEX)

25

Evaluations

• Evaluations measure the success of critical 2010 Census operations and processes

• 28 evaluations are being conducted across numerous program areas. As examples, we have the:

• 2010 Census Quality Survey- Measures simple response variance from an Internet questionnaire compared to that from a census paper form.

• 2010 Census Match Study- Assesses the accuracy and coverage of administrative records data compared to the 2010 Census results.

• Evaluation of the Integrated Communication Program (ICP)– Assesses the effectiveness of the ICP to increase response rates across various race and ethnic groups.

26

Experiments

• Experiments examine form content changes and new methods for the 2020 Census.

• Five experiments were conducted during the census. As examples, we have the: • Alternative Questionnaire Experiment - Tests the effects of variations

in the mail questionnaire, such as, combining the race and Hispanic Origin questions, which ultimately may lead to improving overall response to those questions.

• Paid Advertising Heavy-Up Experiment - Measures whether response from race and Hispanic Origin groups improves with additional targeted paid advertising.

• Nonresponse Followup Contact Strategy Experiment – Assesses the effects of fewer number of followup contacts for households that did not mail back questionnaires.

27

Assessments

• Assessments document final volumes, rates and costs for individual operations or processes.

• 49 assessments are being prepared across most operations. As examples, we have the:

• Mail Response/Return Rates- Provides the official response and return rates for the 2010 Census mail out/mail back operation.

• Be Counted/ Questionnaire Assistance- Provides final tallies on the distribution and disbursement of forms (by language), mostly for people who did not receive a census questionnaire in the mail.

• Census Coverage Measurement Person Matching and Followup – Provides insight into the coverage rates of the population in the 2010 Census.

28

Bridging to the 2020 Test Cycle

• 2020 planning efforts identified critical research projects, to begin in FY2012, to inform the next census design.

• 2020 planning efforts identified critical output from the 2010 CPEX to inform the FY2012 research projects.

• Schedules are being aligned so key 2010 CPEX results can be used to design research projects for the 2020 census.

29

Reasons to change for 2020

High Cost High Risk

$14$29

$40

$70

$108

$181

$-

$20

$40

$60

$80

$100

$120

$140

$160

$180

$200

1970 1980 1990 2000 2010(estd)

2020(proj)

• Early acquisition lock-in• Last minute changes• Contract management• Management systems

performance

Census Cost per Housing Unit (2010$)(Projected cost for 2020 assumes no change in design and past patterns of cost growth; it also includes the costs for

American Community Survey.)

30

Good Ideas for 2020

• American Community Survey as test bed

• Avoiding acquisition lock-in

• More tailored approach to enumeration

• Focus on cost and efficiency

3131

Operational Design Space

HybridDecentralized Centralized

Multi-mode Contact

Visit Then Records

Records Then Visit

Near Paperless Records-basedTraditional +

Internet

Establishing Where to Count – three options

Enumeration – six options

IT and Operational Infrastructure – three options

No CanvassingTargeted CanvassingFull Canvassing

Continuous Frame Updating

54 Design Options

32

Operational Design Alternatives

Targeted None

Infrastructure

Canvassing

Enumeration

Full

Continuous Frame Updating

ODA-1 ODA-2 ODA-3 ODA-4 ODA-5 ODA-6

Traditional + Internet

Multi-mode Contact

Visit Then Records

Records Then Visit

Near Paperless

Records-based

Decentralized Hybrid Centralized

Modest Improvement Radical Change

Six Options

33

Research Tracks

Expanded, Automated, and Tailored Response

Reengineered Field Structure

Continual Frame Updating and Targeting

Using Admin Records for Non-Response

General Design

Informing Operation Design Decisions