Association of Public Data Users
BACKGROUND ON THE CENSUS ADVISORY COMMITTEE
The 2010 Census Advisory Committee (CAC) provides advisory input on the design of the 2010 census, the American Community Survey, and related programs. Committee members represent a range of census stakeholders, and APDU’s seat on the Committee provides a channel for APDU members to comment from the data user perspective.
Ken Hodges is your APDU representative on the 2010 Census Advisory Committee, and this report describes the most recent meeting of that Committee. Reports on these meetings are designed to keep APDU members informed on census activities, and to encourage feedback. Contact Ken anytime at khodges@claritas.com with comments, questions, or suggestions.
MAY 17-18 2007 MEETING OF THE 2010 CENSUS ADVISORY COMMITTEE
Nancy Gordon, the Census Bureau’s Designated Federal Official, started the meeting with welcoming remarks, and announced that a joint meeting of all advisory committees is planned for November. The one day meeting will focus on the integrated communications plan for the 2010 census. Gordon also described some staff changes at Census. Among them, Jay Waite is the new Deputy Director, and Teresa Angueria is the Acting Associate Director for Decennial Census. Gordon also introduced Marilia Matos, who has joined the Census Bureau (from Interior) as the new Associate Director for Field Operations.
Committee Chair Mark Neuman noted that we were meeting amid concern that a lack of funding threatens to delay census partnership activity until 2009 – thus impairing the effectiveness of these programs. Neuman explained, however, that the Census Bureau requested partnership funding, and appealed when it was removed. The issue is in the hands of Congress now, and there is little to be gained by pressing Census officials on this issue at this time. Neuman added his welcome to Marilia Matos, noting that he has worked with her, and knows she will be highly capable in running the field operation.
Census Bureau Update
Census Bureau Director Louis Kincannon remarked that people were pleased when he agreed to serve beyond his November resignation, but that soon people will start asking “are you still here?” Turning to business, Kincannon explained that Commerce Under Secretary Cynthia Glassman was unable to attend, but has been active in census matters—including trips to the field offices to see how census work is done. Kincannon also reported that the Census Bureau has sent LUCA (Local Update of Census Addresses) alerts to communities, and held hundreds of meetings in preparation for what he described as the most important contribution that local areas can make to the census. Invitations to participate in LUCA are to be sent in August, and the director was pleased that the LUCA timeline is well ahead of where the 2000 program was at this stage.
Kincannon noted that this is probably the last time he would address us as Census Director, and he took a few moments to recall his five years plus in that role – with accomplishments including the American Community Survey, the strengthening of the confidentiality commitment, and the new building. But he cautioned that, while 2010 census preparations are going well, confidentiality issues remain, and the budget perils of recent years cannot continue. He assured that Congress and OMB understand the Census budget requirements, and we have a lot of good will, but it will take vigilance to keep things on track. Kincannon closed by saying that he has enjoyed working with the advisory committee, and thanked the representatives for their service.
Deputy Director Jay Waite reported that there has been much success – the ACS is funded, the 2008 dress rehearsal is getting underway, and the 2010 census has been supported by extensive and rigorous research. Waite reiterated that LUCA notices have been sent, with formal invitations to follow in August. He said they are making it easier for locals to participate, and that the process should be more efficient than in 2000. Waite spoke highly of Teresa Angueira, Acting Associate Director for Decennial Census, and Marilia Matos, the new Associate Director for Field Operations, and noted the critical roles they will be playing in the 2010 census.
The short form 2010 census will enable the Census Bureau to focus on coverage, and as Waite put it, 2000 demonstrated that they cannot “fix” the count, so the emphasis has to be on getting it right the first time. The hand held computers are critical to this effort, and after some initial glitches, work with the new computers is said to be going well. Waite is confident that this census will make great strides in automation, and make it clear that we can never go back to paper-based operations.
Waite reported that they submitted census topics in March for congressional review, and that the next step is to submit question wording. And by the November meeting, they will have selected a contractor for the communications program.
Teresa Angueira reported that, as of May 11, MAF/TIGER enhancements had been completed for 2,653 (out of 3,141) counties, and they are on schedule to complete all counties by April. Angueira also reported that the CAUS program, which updates addresses during ACS field work, was suspended for 2007 because of the budget delays. ACS data are proceeding on track, both for households and group quarters, and the 2006 one-year data are due for release in August.
Angueira noted that census content is set, and reflects the adoption of some, but not all of the recommendations from last November’s joint advisory meeting. Noting that “we are in implementation mode now,” Angueira explained that they adopted only options that had been tested, and presented for consideration in advance of that meeting. They are completing analyses of the 2006 tests, and proceeding with automation, LUCA, the awarding of contracts, and the opening of local census offices. As Angueira put it, “The census has begun – we are on the critical path.”
However, because of the census ramp up, a lengthy delay in passing this year’s Census budget would be equivalent to a budget cut of 50 percent – possibly scuttling the dress rehearsal, and introducing great operational risk to the census. The Census Bureau is being proactive in making sure Congress understands the situation.
A new Data Access and Dissemination System (DADS) also is in the works, and the Census Bureau hopes to award a contract for its development by September. They will reach out to user groups for input on the new system, with Nancy Gordon spearheading the outreach effort.
Open Question and Answer Session
Noting the importance of the partnership programs to the reduction of differential census coverage, Committee Chair Mark Neuman asked what the consequences would be if 2008 funding for these programs is not restored. Census Director Kincannon did not list detailed consequences, but described how much had been planned for 2008 and 2009, and noted that Congress knows the situation.
Karen Narasaki (Asian American Justice Center) described the partnership program as a top priority for the Asian American and Civil Rights communities, and stressed that it takes time to organize the relevant populations. We cannot afford to lose the advance effort, and she is hopeful that funding can be restored. Narasaki also expressed concern with a Census memo indicating a lack of funding for work on Asian language census questionnaires. Teresa Angueira responded that Asian languages will be a focus for language assistance, but that Spanish bi-lingual questionnaires still provide the greatest improvement to response rates.
Ken Hodges (Association of Public Data Users) reported APDU’s interest in raising the issue of the ACS sample size, which is considerably smaller than that of the census long form. The 2000 long form yielded responses from about 16.4 million households, but even after combining five years of data, the ACS reflects only about 10.5 million responses – a number that will remain static even as U.S. population continues to grow. As a result, ACS data have greater sampling error than the long form data they replace – especially for small areas and small populations. Hodges acknowledged that increasing the ACS sample would require funding, but said APDU views this as an important long term issue that we need to start addressing now. Jay Waite responded that there are things they hope to do to reduce ACS variance levels, but assured that the Census Bureau is aware of this issue, and agrees that the ACS sample will have to keep pace with U.S. population size.
At the November meeting, the CAC had recommended that the race and ethnicity questions provide response options that would be easier for non-Spanish speaking indigenous populations (such as from Peru or Mexico) to identify with. Ilene Jacobs (California Rural Legal Assistance) expressed concern that the absence of such response options in 2010 could lead some to believe they are not to be counted by the census. Howard Hogan, Associate Director for Demographic Programs, said the notion that people might not respond if they do not see a response option of this type is troubling, and assured that the Census Bureau will work hard to ensure that everyone understands that they are to be included in the census count. Mark Neuman described this as a big issue that has kind of crept up on us, but commented that “the Census Bureau gets it now” at the very highest levels.
Jacqueline Byers (National Association of Counties) commented that it was nice that LUCA alerts have been sent, but suggested that sending invitations in August might dampen response, since it is the slowest time of the year for many governments. She asked when in August the invitations would be sent. Jay Waite could not cite an exact date, but quipped that when the Census Bureau promises something in a given month, they usually mean the last minute of the last day of that month. Byers said that would be good, as late August would yield a better response than mid August.
Arturo Vargas (National Association of Latino Elected Officials) expressed concern that advisory committee input often is sought too late to incorporate its recommendations. His remarks reflected the disappointment of several reps that more could not be done to incorporate the recommendation (from the November 30 meeting) for race response options for indigenous populations. Teresa Angueira explained that they limited final census content (including race response categories) to options that had been tested, and noted that the options presented last November had been identified as part of a process that the advisory committees had been part of. Looking ahead, Jay Waite promised that next November’s meeting would not merely present a final communications plan. The message was that CAC input from that meeting could inform decisions yet to be made.
Kimball Brace (Election Data Services) wondered which jurisdictions would be going into LUCA without enhanced TIGER files. Teresa Angueira said they can provide a list of which counties would and would not have TIGER enhancements by a given time, and she noted that priority has been given to large counties that have experienced a lot of change.
Karen Narasaki (Asian American Justice Center) echoed the concern about ACS sample size, saying they have heard reports of suspect ACS data on the Asian American population. Helen Samhan (American-Arab Anti-Discrimination Committee) said they have similar doubts about the ACS data on Arab Americans, and share the concern over ACS sample size.
Kimball Brace then recognized (APDU board member) Terri Ann Lowenthal, who was in the audience, described her excellent census-related work as a congressional staffer and consultant, and the fact that she is leaving Washington for her native Connecticut. Mark Neuman added his praise, and expressed the shared sentiment that Terri Ann will be very much missed as a colleague and friend.
American Community Survey Content Development Update
Lisa Blumerman started by noting that the 2007 budget eventually provided funding for ACS data collection (both households and group quarters), and a 2007 Methods Panel test that will evaluate alternative questionnaire designs in an effort to reduce data collection costs. The 2008 budget request includes funding for field staff training related to content changes, and the production of the first three-year data products. A 2008 Methods Panel test would focus on ways to increase response rates in Puerto Rico and from non-English speaking households.
ACS content changes are planned for 2008 – the culmination of what Blumerman described as a long process, involving an interagency committee, Federal Register notices, and extensive testing. The changes will make ACS short form items consistent with the 2010 census, and “long form” content is guided by the results of the 2006 ACS Content Test. The plan calls for adding three questions (health insurance coverage, marital history, and veterans service-connected disability), and dropping two questions (seasonal residence and veteran years of service). Changes are proposed for five housing topics (year built, number of rooms and bedrooms, plumbing/kitchen/telephone, food stamp benefits, and property value), and seven population items (citizenship, school enrollment, education, migration, disability, employment status, and weeks worked). A Federal Register notice on the proposed changes has been submitted, and the Census Bureau is working to get OMB approval in time for 2008 data collection.
Blumerman also reported that ACS group quarters data collection is going well, and that a multi-year estimates study is available to help users prepare for the first ACS multi-year data this summer. The study includes files of 1-year, 3-year, and 5-year data from the ACS test programs, and provides data for areas as small as block groups. Karen Narasaki (Asian American Justice Center) recommended that the Census Bureau also offer training when the first multi-year products are released.
Joan Naymark (U.S. Chamber of Commerce) asked why the seasonal residence question was being dropped, and Blumerman said it was because it is not used internally, and is not required by law. The question on veteran years of service was dropped because it is no longer needed by the Department of Veterans Affairs. Ilene Jacobs (California Rural Legal Assistance) commented that if the seasonal question is dropped, the ACS would lose the promising potential for providing improved data on the migrant worker population. Naymark returned to the issue of ACS sample size, and suggested the possibility of a test – examining 2000 long form results based on only 10.5 million responses to determine the impact on error. The Census Bureau’s David Whitford commented that they are doing work on the sample size question, but there was agreement that Naymark had proposed an interesting and potentially valuable test.
Karen Narasaki then asked if there was any evidence of drops in ACS response among specific populations, as the ACS lacks the promotional support of the census long form. Blumerman explained that they track ACS response rates closely (although not by race), and have not seen declines that would suggest the need for a closer look.
Concurrent Working Group Sessions
Following the break, the CAC reps split into working groups on the topics of Communications, Language, and Coverage Improvement and Coverage Measurement. Your APDU representative was assigned to the group on Coverage.
Coverage Improvement and Coverage Measurement
This working group heard first from the Census Bureau’s Dave Sheppard on coverage improvement, where goals include the identification of missed persons and the elimination of erroneous enumerations. Research reflects two approaches. The first involves efforts to get the count right the first time – by helping respondents include the correct people on their census form. The second “fix it later” approach involves the identification of households with coverage problems, contacting them, and making corrections.
The 2005 National Census Test was the vehicle for testing coverage improvement efforts. The test had a mail response of 54 percent, so the results relate to a sample of that mailback universe. The test explored alternative versions of the introductory material explaining the residence rules, and who should be included in the household. A “principle-based” approach, with bullet points describing who to include, performed somewhat better than the “traditional” approach (side by side presentation of “include” and “exclude” conditions), and a “worksheet” approach providing step by step instructions. For example, the worksheet approach required more roster changes, as respondents were including too many people as household members.
The test also examined coverage questions designed to identify households that may have mistakenly included or excluded persons. Sheppard described two versions of the under-coverage question, which asks if there were people at the address on census day who were not included in the response, and two versions of the over-coverage question, which asks if any included persons sometimes stay at another location. He then described how the versions of each question were evaluated, and how they performed.
Sheppard’s presentation prompted discussion from the reps on the application of residence rules for a variety of hypothetical situations.
Next Donna Kostanich presented on coverage measurement. She explained that, in contrast to coverage improvement, which seeks to make the census more accurate and complete, coverage measurement is about measuring the accuracy and completeness of the census. Coverage measurement does not affect the census count itself.
Kostanich recalled that the 2000 census coverage measurement effort – the Accuracy and Coverage Evaluation (A.C.E.) did not detect persons counted more than once as well as it needed to. She also contrasted the 2000 focus on net coverage with an emphasis in 2010 on the components of coverage – separate measures of persons missed and erroneous enumerations. The 2010 measures also will focus on housing coverage in addition to population coverage. As Kostanich put it, we still want to know the net coverage, but it does not tell us much about how to improve the census count.
In response to questions from several CAC reps, Kostanich confirmed that the component measures of coverage would be specific by race and ethnicity (as well as other characteristics). She explained that the coverage measurement process begins with the 2006 census test, and plans are underway for coverage measurement in the 2008 dress rehearsal – with the results probably becoming available by late 2009. Asked when we could expect coverage measurement results for the 2010 census to be available, Kostanich said that timing similar to the dress rehearsal would be the best assumption for now. And when asked how coverage measurement work is different now that it is out of the political spotlight, she commented that the context is a lot less intense, but cautioned that it is only 2007, and things could change.
We concluded with discussion of the identification of erroneous enumerations, where there is reason to expect improvements over 2000. For example, the search for duplicate enumerations will be nationwide – in contrast to 2000, when it was limited to a relatively small search area. This observation raised a question about the risk of the erroneous identification of duplicates (erroneous erroneous enumerations?). Kostanich expects that this should not be a big risk, as the follow up process should confirm if enumerations really are duplicates.
Working Groups Report Out to Full 2010 CAC
The full committee reconvened so the working groups could report on their meetings.
Jacqueline Byers (National Association of Counties) reported that the Communications group discussed the importance of hiring communications specialists at a level closest to the people, and using them effectively. There was also agreement on the importance of starting the census in the schools program earlier (than for 2000), and on the need for special communications materials for Hurricane Katrina victims. The group also was concerned that the ACS should not be collecting data while the 2000 census is in the field. The concern is that having both in the field could confuse people and diminish response. There was little follow up discussion of this concern, but your APDU rep’s recollection is that the Census 2000 Supplemental Survey (C2SS) was, at least in part, intended to test the effect of collecting ACS data while the decennial census was in the field. If so, now might be a good time to revisit the results. The group also called for the “portability of census ads” – the ability of locals to use census ads without royalties or other barriers. And they agreed that communications should stress the confidentiality of the census.
The Language working group was a small one, so Terry Ao (Asian American Justice Center) presented a list of what she described as essentially her own recommendations. These included the following recommendations for the Census Bureau.
1. Circulate to the CAC a list of the 49 languages for which language assistance guides are to be provided.
2. Circulate the language guides to advocacy communities so they can review the translations in time for input.
3. Expand primary language assistance to more than five languages.
4. Develop glossaries of terms and other census documentation in alternative languages.
5. Make hiring diversity more consistent across regions.
6. Create a protocol on how to translate census materials.
7. Use technology, such as video PSAs, to help minimize language barriers.
8. Do not delay in seeking a waiver on the hiring of non-citizens for census work.
Jungmiwha Bullock (Association of Multi-Ethnic Americans) reported on the meeting of the Coverage Improvement and Coverage Measurement group, which is described above.
Day Two
REAC Update
Day two of the meeting started with an update from the Race and Ethnic Advisory Committees (REAC) by K.V. Rao. Rao explained that REAC consists of five subcommittees representing specific race and ethnic groups. REAC had met the previous week, and Rao noted their concern that partnership activities need to be funded for the coming year, and that privacy concerns and recent lapses pose a challenge to the 2010 census. He also expressed REAC’s appreciation of Census Director Kincannon’s support and service.
Rao noted that REAC had reached a number of resolutions/recommendations at their recent meeting, and listed them as follows.
1. The federal budget process needs to make an exception for census funding in order to guard against the negative impact of another continuing resolution (budget delay).
2. Recommend funding for advertising and other outreach to small communities and populations that might be missed by national census (communications) contractors.
3. Recommend that minority owned businesses be assured an adequate share of 2010 census contracts.
4. Make the activation of 2010 partnership programs an immediate priority.
5. REAC meetings should be extended to a second day.
In response to a question about whether the Census Bureau has responded to these recommendations, Jeri Green, Chief of the Census Advisory Committee Office, indicated that the response is awaiting final approval, and will be provided soon.
Public Comment
With no one seeking to speak in the period for public comment, Terri Ann Lowenthal took the opportunity to thank the Committee for its warm remarks the previous day, and to express appreciation for its colleagueship and friendship over the years.
At this point, Ilene Jacobs (California Rural Legal Assistance, Inc.) inquired about changes to small area census geography that have been proposed for 2010. With no one from Geography Division present at the time, Ken Hodges (Association of Public Data Users) summarized the recent Federal Register notice on the proposed criteria for defining block groups for the 2010 census.
From a data user perspective, the most significant change is the proposal to increase the minimum size of block groups from 600 to 1,200 persons, and with 54 percent of current block groups failing to meet that condition, the number of block groups defined for 2010 could drop significantly from the current total of about 208,000. Such a drop would impair many small area data applications, and could give the appearance of a retreat from the promise of ACS block group data. In response to suggestions that the increased minimums are aimed at preserving confidentiality, Hodges noted that no size limits are applied to government units – many of which are smaller than block groups, and for which both census and ACS data will be reported. He observed further that the Federal Register notice points to the small ACS sample size, and sampling error as the reason for the proposed increase in block group minimums. The issues of block group size and ACS sample size are linked.
CAC Chair Mark Neuman brought us back to the partnership issue, and asked if we, as individual representatives, want to express support for restoring funding for partnership activity in 2008. The real issue, Neuman noted, is the contribution these programs make to the reduction of differential undercount. Terry Ao (Asian American Justice Center) recommended that we advocate for the restoration of the $18 million in funding for partnerships, and specified that it be in addition to full funding of the administration’s current request – as opposed to a shift of funds from another part of the Census budget. The CAC was unanimously in favor of this position.
Congressional Update
The Committee normally hears from congressional staff from both the majority and minority sides, so Mark Neuman explained that John Cuaderes, minority staffer from the House Committee on Oversight and Government Reform, had a schedule conflict, and was unable to attend.
We heard from Tony Haywood, a new staffer with the Subcommittee on Information Policy, Census and National Archives. Haywood explained that he has 13 years of experience on the Hill, and nine years with the Committee on Oversight and Government Reform, but is new to the census. He assured the reps that he hears the CAC consensus on partnership funding, and is part of that consensus. More specifically, he reported that they have indicated formally, in a letter to the appropriators, their support for full funding plus the additional $18 million for partnership activity in 2008.
Jacqueline Byers (National Association of Counties) recalled that prior to the 2000 census, stakeholders conducted briefings to bring congressional staff up to speed on census issues, and with the recent turnover, she wondered if it might be a good idea in advance of 2010. Haywood said it is a great idea, and would be appreciated.
Fred Riley (National Conference of Black Mayors) described as insensitive, Census plans to count Katrina evacuees where they are living in 2010, rather than at the neighborhoods from which they evacuated – and to which they might return shortly after the census. Haywood responded that he understands the concern, and will take it back to the Hill. Mark Neuman observed that if significant numbers moved back to New Orleans after the census, it should be reflected in the Census Bureau’s population estimates, which factor into the distribution of funds.
Neuman then asked about the perspective of the congressional members now responsible for census oversight – do they seem to grasp the challenges involved in the census? Because he is new to census oversight, Haywood was reluctant to say, but he suggested that it is probably best to assume that they do not, as there has been turnover among both staff and members.
Vice Chair Lee Adams observed that technological advances, such as hand held computers, can help the census, but cannot see into people’s houses. Haywood was quick to agree with Adams’ point that technology is no substitute for people, and that that is where the partnership programs are important.
Ilene Jacobs (California Rural Legal Assistance) urged that upcoming hearings focus on differential undercount, and issues related to immigration, and the difficulty in counting immigrant populations. She also stressed the importance of the waiver—permitting the hiring of non-citizens—that was so helpful to the 2000 census. Mark Neuman echoed the thought, noting that it is about having Census workers who look like the population they are counting – including non-citizens.
Committee Action Items
CAC Chair Neuman asked for comments on the format for future meetings. With numerous reps not attending the second day, he wondered if a one-day meeting would be sufficient. The general sense (among those attending day two) was that the second day is helpful, and that we should not limit the Committee’s agenda to fit a one day meeting. Neuman also suggested that the CAC reps give thought to new populations or communities that may emerge as difficult to count, and where advance consideration might help. New immigrant populations in the South and parts of the Midwest were cited as possible examples.
Joan Naymark (U.S. Chamber of Commerce) brought us back to the issues of small area data and block group size, and urged that the CAC needs to be more on top of these issues. Terry Ao (Asian American Justice Center) urged that we follow up on the hiring waiver – recalling that the DCAC (the predecessor committee) had advocated for this waiver, and asking if the CAC should do the same.
The discussion concluded at this point, and the meeting was adjourned.