Summary of June 13 2003 COPAFS Meeting
COPAFS chair Don Muff opened the meeting, and we went directly to Ed Spar for his executive director's report.
Ed started with the budgets-describing as "fairly good news" the fact that the House allocations are pretty close to requested levels. However, NCHS continues to have budget difficulty, as bio-terrorism is still the major focus of CDC funding. Turning to the economic census, Ed reported that the returns-which looked good at first-have now dropped to levels experienced in 1992, so a strongly worded letter is being sent to nonrespondents. Next, Ed noted that Robert Lerner will be nominated to become the next commissioner of the National Center for Education Statistics, and that we can expect hearings soon. There was comment from an attendee that, while Lerner is a strong researcher, he holds some highly controversial views, and we can expect a contentious confirmation process. Ed had e-mailed COPAFS reps the announcement of the new metropolitan areas, and commented that he is fascinated by the lack of press coverage so far. COPAFS will host a seminar on the new areas November 4 in Alexandria. The conference is limited to 150, and additional information can be found on the COPAFS website. Ed also noted that the first Census 2000 PUMS product is currently being released, and closed by announcing the June 18 ceremony at BLS to present Rich Allen with the Jeanne Griffith Award.
Dates for next year's COPAFS meetings are March 12, June 11, September 17, and December 12.
Issues Relating to Language Translation in Health Surveys
Jennifer Madans. National Center for Health Statistics
Jennifer explained that NCHS has been getting questions on the translation of its questionnaires to languages other than English, and is accountable to the institutional review board on this matter. They do not really know how well they are doing, and Jennifer came to COPAFS not to explain their work, but to promote discussion of the issue. As she described it, there has been work in this area-a conference on translation and a Census Bureau workshop on best practices for census surveys-but no coalescing around standards or best practices that could serve across agencies.
NCHS seeks advice on how to address translation issues-not specific methods, but broader suggestions on how agencies can deal with the challenges of translating their surveys. One attendee noted that there are organizations that certify translators. Jeffiner agreed that certification might be sufficient for some purposes, but wondered what the use of a certified translator really means. When the possibility of an OMB role was raised, Hermann Habermann (currently with census, and formerly with OMB) cautioned that he would be reluctant to start a big OMB process before understanding more about the need.
Jennifer spent much of her presentation describing the questions NCHS is wrestling with, and COPAFS attendees reinforced that these are indeed tough questions, and raised additional questions. For example, there was discussion of the importance of translating to different dialects of Spanish. It was noted that dialect-specific translations could improve response rates and data quality, but there was also comment that the additional cost and disruption could be detrimental, especially if mandated for all surveys. The view was that translations should be tailored to the target population of the survey, and it was suggested that, for some surveys, translations to different dialects of English could be important.
As the session concluded, it was clear that we had not resolved NCHS's issues, but that Jennifer had done much to improve our awareness, and had succeeded in generating a good discussion.
Making Data Accessible to People with Disabilities
Marianne Zawitz. Bureau of Justice Statistics
Laurie Brown. Social Security Administration
Laurie started by describing "accessibility" as analogous to the ability to get into a building-the provision of ramps for those unable to use stairs, and so forth. In terms of information on the web, obstacles to access result from many temporary and permanent disabilities, but with so many obstacles relating to sight impairment, this was the focus of the presentation. Laurie described the Web Accessibility Initiative, which defined standards for web accessibility. These standards are the basis for Section 508, which is part of the Rehabilitation Act Amendments of 1998.
Final Section 508 standards were published December 21, 2000, and took effect June 21, 2001. A primary objective is to eliminate barriers in information technology, and Section 508 has established 16 "web requirements." Some of these are probably best understood by those fluent in HTML, but the challenge is illustrated by the requirement that a text equivalent be provided for every non-text website element that provides information or facilitates site navigation-including images, graphics, audio clips and video clips. Information in data tables must also be provided in audio form. This is especially challenging for complex data tables, where each cell of information must reference a unique and multi-layered set of row and column headers and sub-headers. Spreadsheets can be even more difficult-as cells have formulas and complex interrelationships, and labels often span multiple cells. And in presenting response options in online questionnaires, something as straightforward as a five-point agree-disagree scale may not translate easily from HTML to audio.
Marianne continued the presentation by noting that all federal agencies, and federal contractors, must deal with the Section 508 requirements. The agencies are concerned about burden, and frustrated by a lack of guidance and automated solutions, and FedStats-somewhat by default-has taken the lead in promoting discussion of these issues. As part of this effort, FedStats held a June 2002 workshop on the presentation of tables, charts, formulas and forms. The workshop was attended by 40 federal agencies, as well as researchers, vendors, and representatives of the disabled community.
The workshop's findings and recommendations are available in a white paper, but those described in the presentation reflect the agencies' frustration. For example, there is concern that translations of large complex tables may not be usable to disabled users-that even if translated content can be comprehended, sight-impaired users would not have the ability to actually use (or manipulate) the data as intended. There was also concern that making tables accessible should not make them less usable to sighted users, that meeting the 508 standards poses a resource problem for statistical agencies, and that there are no methods for validating accessibility. There was particular concern that the information in complex charts and maps cannot be communicated according to 508 standards, and a recommendation for guidance on how agencies can provide assistance in lieu of accessible alternatives (e.g., a telephone contact for verbal explanation of maps and charts).
It was when describing the obstacles associated with charts and maps that Jennifer and Marianne noted that for now, agencies are expected to show a "good faith effort" with respect to 508-that they will be judged on performance, not compliance. Again, one might have to be a website designer to fully appreciate the challenges posed by Section 508, but the high level of complexity is apparent. COPAFS attendees were impressed by and commended the efforts to promote the accessibility of federal data on the Internet, but also expressed concern that there needs to be a balance between the needs of sighted and non-sighted users.
The American Community Survey: An Update Nancy Gordon.
U.S. Census Bureau
Nancy-who arrived fresh from a meeting on Capitol Hill-reminded us that 2003 funding did not permit full ACS implementation, and that full implementation is now scheduled to start July 2004. She described this as a blessing in operational terms, as it gives the Census Bureau time to staff up and train ACS staff, and to address issues raised by Congress.
One issue is mandatory response, so the Census Bureau is conducting a test of a voluntary ACS. In March and April, questionnaires were mailed in four panels. Two panels received instructions describing ACS response as mandatory, and two described it as voluntary-one of these in matter-of-fact terms, and the other stressing that response was voluntary. Results are not available yet, but we can expect to find that voluntary response leads to lower response rates, higher costs, and possibly reduced data quality. Nancy reported that the Census Bureau is now comparing ACS and long form data, and will release the results in a series of reports. For users interested in making their own comparisons, the Census Bureau is preparing modified versions of the Census 2000 long form data-with group quarters removed to make them more comparable to ACS test data (which do not include group quarters).
Nancy then described the May 13 hearing on the ACS held by the House subcommittee chaired by Adam Putnam (R-FL). Census Bureau Director Louis Kincannon and Undersecretary of Commerce Kathleen Cooper testified first (obviously in support of the ACS), followed by a panel of five data users, who testified in support of the ACS. Nancy conveyed that the Census Bureau was pleased with the support expressed at the hearing, and remarked that there has been much positive ACS news in recent months.
When asked to clarify the ACS budget numbers for 2004 and 2005, Nancy explained that she could not cite numbers for 2005 (there has been no formal request), but noted that the 2004 cost of $64.8 million includes the ramp up for full implementation in the fourth quarter. And based on estimated costs, she indicated that something on the order of $150 million (plus some expenses deferred from 2004) would be a good guess for full implementation in 2005. In response to another question, Nancy confirmed that these cost estimates assume a mandatory ACS. Nancy went on to cite an estimated total savings of $868 million that the ACS would provide versus a traditional census with a long form.
Patty Becker has promoted the creation of ACS tract aggregations, which would enable the more timely release of ACS data for sub-areas of large government units (aggregations meeting the 65,000+ and 20,000+ thresholds). Nancy responded to Patty's renewed inquiry by noting that the Census Bureau is working on this.
The Sub-National Estimates Program: An Update Lisa Blumerman.
U.S. Census Bureau
Lisa noted that this is the first COPAFS presentation on the Census Bureau's estimates program, and explained that by "sub-national," she was referring to estimates at the state, county and place levels. The Census Bureau produces estimates mandated by Title 13, and which are used for purposes of funds distribution, survey controls, and program planning. She displayed a partial organization chart identifying Signe Wetrogan as head of the estimates and projections branch, with Lisa heading up estimates, Gregg Robinson on evaluation and demographic analysis, Rodger Johnson responsible for administrative records, and Greg Spencer heading up projections. It was in Nancy Gordon's presentation that we learned that Lisa will soon leave the estimates branch to take a Census Bureau position in the policy area.
Lisa explained that the Census Bureau's sub-national estimates are done in collaboration with the Federal-State Cooperative Program for Estimates (FSCPE), and provided a basic description of methods. County population estimates are produced with a component method-involving separate estimates of births, deaths, and net migration, while the place (or sub-county) population estimates are now produced with a distributive housing unit method, driven by estimates of residential construction, mobile home placements and housing unit loss. All estimates must be consistent, such that estimates for small areas sum exactly to the estimates for larger areas. Lisa noted that they are now using new methods and sources-ACS data combined with administrative records-to estimate international migration
Next, Lisa described the evaluations of their 2000 estimates against the counts from the 2000 census. For example, percent error for national level population was -2.4 percent, and mean absolute error was 3.4 percent for counties, and 12.4 percent for places. Also evaluated for perspective were alternative estimates based on 1990 PES adjustment, constant growth assumption, and "do-nothing" estimates based on the assumption of no growth since 1990. As expected, estimates based on constant and no-growth assumptions did not fare well. There was also a tendency for the PES adjusted estimates to be more accurate, but this could be largely a function of the PES adjustment adding population to estimates that tended to be too low. Estimates of proportional distribution of the population also tended to show lower error-a favorable finding for applications involving funds distribution.
Looking to the future, Lisa explained that the estimates branch hopes to further improve the estimates of international migration-focusing first at the national and then sub-national levels. They are also looking into ways to incorporate state-specific local data into sub-state estimates, as well as the use of updated sub-county boundaries, and the integration of ACS results in the estimation process. The improvement of state and county estimates of population by age, sex, race and Hispanic origin is another priority. The estimates branch also seeks to achieve a more timely dissemination of its estimates, and to increase the involvement of the FSCPE and state data centers in its work.
Concerns of COPAFS Constituents
No concerns were raised, and the meeting was adjourned.
- Don Muff, Muff Consulting Services
- H. Habermann, U.S. Census Bureau
- Howard Leathers, AAEA
- Sara Zapolsky, AARP
- Stephen Tordella, Decision Demographics
- Ken Hodges, PAA and Claritas
- Richard Forstall, Assn. of American Geographers
- Mel Kollander, Temple University ISR
- Clyde Tucker, ASA
- William Kandell, Rural Sociological Society
- Gerald Sroufe, AERA
- Lee Herring, American Sociological Assn.
- Leslie Scott, ESSI
- Gerry Hendershot
- Jeremy Wu, BTS
- Chuck Waite, NABE
- John Thompson, NORC
- Joe Garrett, Mathematica
- Vince Iannacchione, RTI
- Bill Lidington, Beyond 20/20
- Fred Cavanaugh, Sabre Systems
- Judie Mopsik, Abt Assoc.
- Thomas Brown, I-ASSIST
- Jeff Newman, BEA
- Colleen Flannery, Census
- Ralph Rector, Heritage Foundation
- Linda Jacobsen, APDU
- Maurine Haver, NABE
- Margaret Martin, COPAFS
- Ed Goldfield, Census, CNSTAT
- John Munyon, SeekData
- Lu Jeppersen, SeekData
- John Leffler, SeeData
- Gooloo Wunderlich, IOM
- Jerry Fletcher, AAEA
- Patricia Becker, APDU/SEMCC
- Ed Spar, COPAFS
- Kelvin Pollard, PRB
- Lisa Blumerman, Census
- Nancy Gordon, Census
- Paula Schneider, Census Ret.
- Chet Bowie, Census
- Nancy Potok, NORC