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 new Census Advisory Committee. 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.
Arnold Jackson, the Census Bureau’s “designated federal official,” started the meeting and introduced the Committee’s new chairman, A. Mark Neuman, and vice chairman, Lee Adams.
Neuman recalled his time with the Census Bureau’s Executive Staff, and the 2000 Census Monitoring Board, and remarked that one of his priorities will be to make the advisory meetings more open and available to those outside the Washington area. Initial steps might include the posting of meeting transcripts, and other materials. The new chairman also signaled his intent to keep us on schedule by noting that we were already behind, abbreviating his remarks, and moving to the census director’s report.
Census Bureau Director Louis Kincannon welcomed everyone to the “final days in building number 3” – a proclamation that drew applause from some Census Bureau staff. The move to the new building (adjacent to the current building) is set to begin this August.
Turning to business, Kincannon commented that the $878 million proposed FY 2007 budget continues this administration’s strong support for the census. The proposal calls for over $500 million for decennial preparation, $180 million for the American Community Survey (ACS), and $74 million for MAF/TIGER enhancement – proportions that Kincannon described as consistent with the Census Bureau’s priorities.
In response to across the board cuts mandated by Congress, Kincannon explained that they do not like across the board cuts, and instead decided to suspend the Survey of Income and Program Participation (SIPP) – which was considered in need of renovation anyway. The plan is to discontinue SIPP in its current form, and re-engineer it to an updated form that takes advantage of new resources. It was news to some CAC reps that SIPP was being suspended and not permanently eliminated.
Kincannon then reported briefly on the first release of annual ACS estimates (due in August), plans for the 2008 Census Dress Rehearsal, and the 2006 census tests. He described his visit to the Cheyenne Indian Reservation test site, where one can travel 100 miles to contact just two households, and the testing of handheld computers in the Travis, TX test site. The director also described the Census Bureau’s report on the implications of counting prisoners at pre-incarceration residences (higher costs and questionable data), and the proposed constitutional amendment that would require that the census count only U.S. citizens for purposes of apportionment.
Jay Waite, Associate Director for Decennial Census, reported that the TIGER enhancements are progressing one county at a time, but are on time for an April 2008 completion. The work is critical to the use of handheld devices, and for the accurate geocoding of housing and group quarters. Waite described the ACS as going well, with funding that is established, at least for now, and with data collection now including Puerto Rico and group quarters. He recalled that, “in the old days,” the decennial census never did a very good job of collecting long form data from group quarters, and observed that it is proving difficult and expensive in the ACS. Waite said there will have to be a dialog between the Census Bureau and data users on what group quarters data are of most importance.
Waite described the upcoming release (in June) of ACS data for the areas impacted by hurricanes Katrina and Rita. The data will provide a before and after picture for these areas by presenting separate data for the January–August and September–December time periods. The Census Bureau also is doing research and evaluation on the ACS. As Waite put it, they do not want to change the questionnaire every year, but are looking at question wording and other issues for possible improvements down the road.
Noting that the 2010 census is not far away, Waite said they are looking at the results of the 2005 content tests, which included a test of a second mailing. The test indicates a 7 to 10 percent response to a second mailing – a “big help” as Waite described it. The use of bilingual questionnaires also is being tested, and appears to increase mailback response. The use of hand held devices in the Travis, TX test is going well despite some bugs, which are being addressed by the private contractors. For example, when field workers had trouble remembering passwords for their handheld devices, the contractors switched to finger print identification. As Waite described it, the technology is new for the Census Bureau, but not for the contractors. The Census Bureau is now highly confident that the handheld device program will work very well.
A key contribution of the hand held devices is expected to be the entering of late returns – preventing the wasted effort and ill will that results from enumerators visiting households that have responded to the census. Out of 12,000 late returns in the Travis, TX test, 11,000 were intercepted before an enumerator went for a visit.
Next on the agenda was a period for questions from the CAC representatives. The session went something like this.
Q. What about the on-going update of the Master Address File?
A. Jay Waite referred to the CAUS (Community Address Updating System) process,
in which field staff update the MAF while in the field collecting ACS data –
a process already underway in the hurricane impact areas. The quarterly updates
to the USPS Delivery Sequence File are another source of updated addresses,
but updates in rural areas are still a challenge.
Q. With the problems in collecting ACS data from group quarters, would it make
more sense to collect only short form data in group quarters?
A. Waite said they are wondering about this, and want to hear from data users.
Q. What about the Internet response option?
A. Waite said they are seriously wondering if it makes sense. Not many use the
option in census tests, and indications are that most who use it would respond
anyway, so there is little to be gained for the added expense. There also is
concern that someone could put something online that looks like a site for census
response, but is really for Internet “phishing.” Deputy census director
Hermann Habermann added that outside experts agree with this concern, and have
recommended against the Internet response option.
Q. How are response rates in the census test areas?
A. Waite acknowledged that response in the Travis, TX site is a bit lower than
expected.
Q. For what geographic levels will the ACS Katrina data be published?
A. The Census Bureau will publish as much as they can for the impacted areas.
This includes states and groups of counties such as metropolitan areas. Some
large counties might be expected, but users are cautioned to understand that
data collection in the impacted counties has been difficult, and the number
of responses greatly reduced. In addition to the ACS data, users can look for
a set of highly tentative post-hurricane population estimates for counties.
Q. Have you given up on block group data from the ACS? As has sometimes been
the case, today’s Census Bureau descriptions of 5-year data indicated
only “tract data.”
A. The Census Bureau confirmed, once again, that block group data are still
the plan, but noted that those who are “variance minded” continue
to worry about such data.
Q. Are there any noteworthy changes in plans for ACS data products since the
last meeting?
A. Lisa Blumerman reported that the products will be expanded in scale—reflecting
full implementation—but would look much like what we saw for 2004.
Q. By 2010, the ACS will be reporting as many as three different numbers for
the same data item—a one year, three year, and five year number. What
is the thinking on this?
A. The Bureau sees the major question as how best to educate users. The discussion
then drifted somewhat to the question of the population estimates, and the work
remaining to be done on the Program for Integrated Estimates. As we have heard
before, the estimates will make use of the ACS and the ACS will make use of
the estimates, but there is not much written on this subject yet.
Q. With the elimination of the long form in 2010, is it realistic to expect
that the differential undercount (by race) can be virtually eliminated?
A. Director Kincannon agreed there is reason to hope and set that objective,
but he was reluctant to make a prediction. Jay Waite offered that if the differential
undercount is completely eliminated “I’ll retire.”
Chief Privacy Officer Gerald Gates described the efforts to instill the privacy ideals and objectives across Census Bureau staff and contractors, and a campaign to educate all staff on the Census Bureau’s privacy obligations. Gates also reported that the Bureau has finalized its policies on custom tabulations. The new policies emphasize openness, and call for the public reporting of all custom tabulations. The first report listing recent custom tabulations has been issued, and is on the Census Bureau website.
Gates identified four major privacy/confidentiality initiatives. The first is to explore in depth the issues identified in last year’s workshop on privacy and data use. The second is research on how privacy issues impact census response, and how these issues can be addressed to improve response. Third is the promotion of internal privacy awareness—in particular, educating Census Bureau management on the privacy impacts of their decisions. Fourth is external communication and partnership – being more transparent, and better communicating the Census Bureau’s commitment to and actions to protect confidentiality. As part of this initiative, the Census Bureau is developing a response strategy for when confidentiality issues arise. They also are looking to identify the privacy concerns of specific minority populations, and are soliciting help in this effort. And finally, Gates expressed the Census Bureau’s hope that knowledgeable users will speak out publicly in the Census Bureau’s defense when confidentiality issues arise.
Some CAC representatives who had been sharply critical of the Census Bureau following the Homeland Security episode (described in the November 2004 APDU DCAC report) praised the Census Bureau for establishing the Chief Privacy Officer position, and commended the Bureau for the seriousness with which it takes these issues. They indicated that, so long as the Census Bureau continues to be this up front on this issue, they would publicly stand up in its defense.
At the fall meeting, several CAC representatives expressed interest in working groups, and prior to this meeting, the Census Bureau informed us that it was establishing working groups on the following topics.
Communications
Language
Data Collection for Special Populations
Coverage Improvement and Measurement
Content, Including Race and Hispanic Origin
Your APDU representative is assigned to the working group on Data Collection for Special Populations.
The working groups met during early afternoon, and reported to the full committee during the late afternoon session. The working group reports are summarized below.
Jackie Byers (National Association of Counties) reported that this group heard the latest on census outreach activities from Jeff Taylor, who reported that the Census Bureau plans to award the major advertising and outreach contracts by May 2007. Media, and how people consume media have changed since the last census, and the Census Bureau understands the need to account for these changes into the advertising and outreach for the next census. Other concerns include the awarding of contracts to minority firms, how to reach immigrant populations, coordination with partners, the communication of privacy/confidentiality assurances, and the effective use of the latest technology.
A new concern for the 2010 census is how to advertise a short form only census, since much of the pitch for previous censuses highlighted popular uses of long form data.
Karen Narasaki (Asian American Justice Center) commented that language is an issue for both the decennial census and the ACS, and that the ACS lacks the public relations that benefits the decennial. The working group suggests the need to build knowledge and awareness of the ACS among schools and other community organizations. The group expressed specific concern that advance letters are only in English, as are the packages of materials sent after these letters. The bi-lingual forms—a single form with English and Spanish language options—is viewed as a positive step, but there is concern that English/Spanish is the only such form to be sent to targeted areas. The working group argued that bi-lingual forms are needed for other non-English languages, but acknowledged that the other non-English language populations are not as easy to target as the Spanish speaking population.
Kimball Brace (Election Data Services) reported that the group heard an update on the LUCA (Local Update of Census Addresses) program, with important improvements, such as sending letters to more than just the highest elected official in each jurisdiction. But the group suggested that the letter could be harder hitting – warning local officials that if they do not participate in LUCA, their census count will be lower than they expect. There was discussion of the possibility of LUCA having access to other resources, such as voter records and USPS change of address data. And concern was expressed that the testing now underway is aimed primarily at 2020. Why not 2010, they wondered.
Asked if the group had discussed the overcount of some populations, Brace said they had not, and there was discussion of the need to ask specific questions to identify duplications—for example, asking college students where their parents live. A question was raised concerning the potential use of administrative records in the reduction of differential undercount. Jay Waite commented that this is one way to address hard to count populations, but cautioned that a portion of the world is nervous about the Census Bureau getting and using administrative data.
Clark Bensen (POLIDATA) explained that the Content group looked at the use of administrative records (such as Social Security records to fill in missing items such as age and sex. The use of administrative data prior to resorting to imputation makes sense on paper, and likely would yield better results, but the group had serious concerns with public perception. The group also expressed concern that such measures would improve the data collected for populations that are not under-privileged.
There was much discussion of the race/ethnicity/ancestry options that are being tested. The three part question (race – ethnicity – ancestry) is still a possibility, but it has met resistance, and is generating questionable data. There are also questions on how to interpret the data for redistricting purposes. As one rep put it, the data are getting too complicated for judges to understand. Jay Waite remarked that the race question comes down to 1) how much space can be used on the short form, and 2) how to get people to answer the race question. He assured that the Census Bureau will tabulate the test data, and present the results to the CAC for the fall meeting. Before leaving the topic, someone made a case for allowing multiple response on the Hispanic origin question—for example to accommodate someone whose father is Cuban and mother is Puerto Rican. There was agreement that this is an important issue, but realization that it raises further issues on tabulation and reporting.
Your APDU representative presented the summary for this group, which received updates on the counting of migrant/seasonal farm workers and persons in group quarters. To enhance the enumeration of migrant farm workers, the LUCA process will be conducted earlier this time, and will integrate housing and group quarters. These had been handled separately in the past, and the integrated approach is expected to provide better coverage and less duplication. The Census Bureau also is developing a list of counties where concentrations of migrant farm workers can be expected—and where the Bureau can focus training and canvassing efforts. Additional steps include the use of cultural facilitators, waivers allowing the hiring of non-US citizens as enumerators, and focus groups exploring how best to get migrant farm workers to respond to the census.
The working group was asked for suggestions on how to enhance outreach to the migrant farm worker population, where to locate “be counted sites,” and how best to disseminate information to this population. The unanimous response was that we need Ilene Jacobs (California rural Legal Assistance, Inc.) to address these questions. Ilene is assigned to the working group, but was unable to attend the meeting. As part of the discussion, it was noted that due to its seasonal nature, and the different ACS and decennial residence rules, the migrant farm worker population may look different in the two sources.
We then heard about the initial efforts to collect ACS data from group quarters (GQ) populations. Sampling from a frame based on Census 2000 plus updates, a two-stage collection process starts with a contact person at the sampled facility, followed by interviews with a sample of 10 to 15 residents. As Jay Waite had mentioned, the process has been more difficult and expensive than expected, with gaining access to some facilities proving a major obstacle. The Census Bureau is seeking suggestions on how best to overcome this resistance, and how to control costs. One working group suggestion called for work with trade associations (such as those involving universities) and advocacy groups (including those for the homeless, which have been the source of some resistance). There was only brief discussion of the need for long form data from GQ populations, but it was agreed that it is a question for the group to address.
The group expressed that it is not clear on the definition of “special populations,” and the Census Bureau’s view of the scope of its work. The focus today was on migrant farm workers and group quarters populations, but are there others to be considered? It is hoped that this question can be clarified before the next CAC meeting.
The second day started with follow up to some topics from the first day. For example, Jay Waite clarified that, since ACS group quarters data collection did not begin until this year, the first 5-year ACS data including group quarters are not due out until 2011. Waite confirmed that 5-year data minus GQ would be released in 2010, and remarked again that the question of group quarters data from the ACS is somewhat up in the air. And following up on the previous day’s discussion of hiring waivers, we learned that for Census 2000, the Bureau hired 32,000 non-citizen workers for Census 2000. Chairman Neuman commented on the importance of this waiver, and urged the Census Bureau to get an early start on ensuring it for 2010.
K.V. Rao, the CAC rep from the Race and Ethnic Advisory Committees updated us on their activities. Rao noted that the committees have experienced a lot of turnover, but is a good group. Privacy and confidentiality continue to be a concern. The committees are always interested in complete counts and accurate data on their populations, but are concerned with proposals for the use of administrative records matching in place of imputation. They have asked for a list of the administrative sources currently used, and talking points explaining how and why such resources are used. As Rao described it, administrative records applications run the risk of a public relations nightmare.
The REACs also are concerned with language issues, and have pointed to technology that is now able to determine the language of telephone callers, and recommends that the Census Bureau explore the use of this technology in assistance operations. There is also concern that minority companies tend to be too small to bid on large Census contracts, and that the Bureau needs to make a special effort to ensure that minority firms are awarded a share of the 2010 contracts. The REACs also recommend that the Bureau start moving now on 2010 partnerships, and the waiver needed for hiring non-citizens. The REACs also see a need for more minority representation in senior Census positions, and recommend work on promotions and opportunities at the upper levels.
Joanne Caldwell, from the Census Bureau’s Congressional Affairs Office, observed that the census appropriations hearings went well, and the House and Senate mark ups on the proposed budget are expected in June. She promised to provide any news of note to the CAC. Awareness of the ACS is increasing, and Caldwell noted that the Census Bureau has been asked to expand the survey to include a variety of items (such as Internet access, immigration, and others) related to proposed legislation.
John Cuaderes, majority staff with the Subcommittee on Federalism and the Census, confirmed that members of Congress understand the importance of census products, but pay little attention until about two years prior to the census. But Cuaderes argued that critical decisions will occur in 2006, and the members need to be educated that “redistricting starts today.” So members need to be concerned with the success of the census tests and dress rehearsal, and understand that the success of the ACS is now critical to the success of the decennial census.
Congressional staff are believers in the value of the ACS, and (as Caldwell noted) are trying to add questions. But Cuaderes cautioned that the ACS is still fragile, and we cannot simply load questions on it. And still there are many members who know little more of the ACS than that their constituents complain about having to respond to it. Most members become supportive when they understand what the ACS is about, but still Cuaderes expects a fight for 2007 ACS funding, and said friends of the census will have to raise their voices, and let their support be known.
Cuaderes also told users to be prepared for the “new kid on the block” – Senator Tom Coburn (R-OK), who now chairs the Senate oversight committee. As Cuaderes put it, Coburn can be a powerful ally if you get to him early, but if you don’t, you will wish you had. Private conversations with other observers confirmed that the Census Bureau will need to take care in dealing with Coburn.
Mark Stephenson, minority staff with the Committee on Government Reform, recalled the tensions over sampling leading up to the 2000 Census, and was pleased to note that the 2010 lead up has been relatively bi-partisan. Stephenson also described proposed legislation that would give the District of Columbia a vote in the House, and provide an additional seat to Utah—permanently increasing the House, and the redistricting landscape, to 437 seats.
At this point, one representative asked that the Census Bureau provide CAC members with notice of bills affecting the Census as they are introduced. Caldwell agreed to do this. There followed a question on the status of the proposed Constitutional amendment to require the counting of US citizens only. It seems there are two approaches to this objective – one through amendment, and another through legislation. Cuaderes commented that the legislative approach would be faster, but if passed, the bill in question would go directly to the courts. There is much interest in these proposals, but Cuaderes does not see sufficient support for either to pass. Asked to describe some of the additional questions that have been proposed for the ACS, Cuaderes identified questions on college degrees, food insecurity, and marital history. The process for adding questions is such that if Congress directs a new item through legislation, the Census Bureau must add it, but Cuaderes stressed the need for a deliberate approach, such as that involving testing through the ACS methods panel.