Minutes of the Quarterly Meeting
Friday, June 7, 2002
9:00am to 3:00pm
Bureau of Labor Statistics Conference Center
Report from the Board of Directors & Update on Recent Developments and Council Activities
COPAFS chair Maurine Haver opened the meeting, and turned it over to Ed Spar for the Executive Director's report. Ed Spar offered congratulations to Dick Kulka, Mike Cohen, and Carolee Bush on their election as Fellows of American Statistical Association.
Ed started with the budgets noting that cuts in the NCHS budget have put the National Health Interview Survey at risk. NCHS director of Health Statistics, Ed Sondik was present, and said the NHIS is already being affected. He described NCHS funding as inadequate for the reworking of the NHIS to Census 2000, and commented that if the situation does not change, they will have to make significant changes in 2004.
Nadine Hamilton from BTS said that their reauthorization group is progressing. Note was made that the Education Dept has a flat budget, and while reauthorization is going on NCES structure could change significantly.
Other announcements: Ed mentioned that the November 6-7 COPAFS seminar (Challenges to the Federal Statistical System in Fostering Access to Statistics) was coming up and that registration was likely to fill up early. In personnel news, Bill Barren will be leaving Census, John Thompson will be leaving also to go to NORC in Chicago.
Noting the contents of the packets that were distributed for the meeting, Ed mentioned a table on races & Hispanic status, pointing out that "Spanish Other" has increased as a percent of total. In 2003 the Census Bureau is going to revisit the whole question of Spanish ethnicity and race. Census may want to eliminate the "other" line in the race item, but says they won't test a single question. This will likely be a large project.
Also in the packets, for 10:30 presentation, was the Opportunities & Issues sheet for the ACS. Ed pointed out that these comments have been vetted with Nancy Gordon at Census.
Ed relayed the word from the Census Bureau that they do intend to release five-year average ACS data for block groups-not just census tracts, as had been understood by some data users. And because there may have been confusion among some users on another point, Ed reminded us that ACS data for areas with 65,000+ population would exclude many counties-since so many have populations below that threshold.
Ed finished with mention of two brochures in our package (one on Kids Count and one on transportation statistics), and a recent New York Times op-ed piece describing how the OMB resisted pressure from the administration to relax Directive 3. The directive prevents administration officials from receiving advance releases of key federal statistics or commenting on them until an hour after the release.
The next COPAFS meetings are Sept 20, and Dec 13th 2002.
New Proposed Initiatives at the Bureau of Economic Analysis:
Steven Landefeld, Bureau of Economic Analysis
Steven explained that the GDP is the BEA's flagship product. The full GDP measure involves many components, but BEA produces preliminary GDP estimates based on partial data. The early estimates provide a good overall picture of the economy, and have a large following. However the subsequent revisions can cause problems for those who do economic or budget forecasting-such as the Congressional Budget Office. Steven noted that this became an issue recently when some of the optimistic early estimates-based on economic growth in the late 1990s-were revised downward.
Presented Strategic plan.
- Updating national accounts. BEA Plans Updating the NIPA's and addressing data gaps.
- National Accounts Improving consistency and Integration with other accounts.
- Improving consistency with international standards.
- International Account: Updating and closing data gaps in the accounts.
- Industry accounts: Updating and closing data gaps in the accounts.
- Regional Accounts Updating and accelerating accounts.
- Wish list: Expanded accounts Satellite or supplemental accounts.
BEA is currently updating the national accounts, and addressing data gaps. Objectives include the introduction of the quality dimension to price data, improved measures of services and international trade, better estimates of employee stock options, improved timeliness of BEA data products, and the implementation of NAICS. BEA also hopes to improve the consistency of its data with "other accounts," such as the Federal Reserve's flow of funds accounts. They are also working to improve consistency with international accounts, and to close data gaps-for example, with respect to international services and international capital flows. BEA is also accelerating its regional accounts, and improving the accuracy and coverage of its regional accounts source data. Immediate objectives for 2003 include the release of more timely data, an upgraded statistical processing system, and improved data on services and electronic business.
The American Community Survey-Opportunities and Concerns
Nancy Gordon, Associate Director Demographic Programs.
Nancy started with ACS background that is familiar to most COPAFS attendees-describing the three main components of Census Data (ACS, MAF/TIGER and re-engineered 2010 census). As part of her introductory remarks, she mentioned the expectation that the innovations of a re-engineered census would make the ACS/short form census combination "cost-neutral" with the traditional census with a long form.
Next, Nancy reviewed ACS "opportunities and challenges"-similar to the "opportunities and concerns" list developed by COPAFS. Among the opportunities Nancy described were:
- A simplified 2010 census enabling a more accurate population count.
- More current data-of value to users in all sectors.
- More accurate targeting of funding-both public and private sector.
- Enhanced federal surveys.
With respect to the federal survey opportunity, Nancy explained that the ACS would not replace surveys such as the CPS, SIPP and AHS, but would contribute enhanced sampling frames for these and other federal surveys.
This point led to the first "challenge"-the concerns of some private sector data collection firms. Nancy stressed that the Census Bureau has no intention of taking work away from the private sector, and described the challenge as that of taking advantage of the ACS's potential contributions to federal surveys, without undermining the private firms. Private sector concerns seem to come primarily from firms that do survey work under contract to federal agencies.
Another challenge is to increase data user comfort levels with the ACS-as some users are still uneasy with the idea of five-year average data, and not convinced that the ACS will adequately replace what they use from the census long form. As Nancy described it, increasing user confidence levels will require 1) experience with how to use annual microdata, 2) assurance that ACS data quality will be equal to that of the long form, and 3) experience with disclosure avoidance limitations. Funding is another challenge, and not just for the ACS itself, as all three components of the 2010 plan must be funded for the ACS to succeed. And ACS requires both early funding for development, and then stable funding over the long run.
Moving to questions, Nancy addressed one on the objections of private survey firms. She explained, (with Janet Norwood providing some clarification) that firms doing survey work for federal agencies are concerned that they cannot compete with a federally funded sampling frame from the ACS. Nancy indicated that the Census Bureau cannot provide the master address file to private survey firms (some have asked for it), but said the Census Bureau plans to provide block level counts of housing units (counts of MAF addresses). Nancy indicated that this was nothing new, but it was news to many attendees, and it generated discussion. Actually, block level address counts might be of limited value to the private survey firms (they really need addresses), but could be of significant value to the firms that produce small area estimates. Chip Alexander entered the discussion to confirm that such counts could be provided, but cautioned that, given Census Bureau work loads, we should not expect to see such counts before 2004.
There followed a heated discussion on the C2SS PUMS file-which provides records for only a sample of the C2SS sample. Some COPAFS attendees pressed the Census Bureau presenters on why the Bureau did not release the full C2SS microdata sample. Nancy explained that the long form PUMS is a sample of a sample, and invoked arguments of confidentiality and levels of geographic detail. The critics countered that C2SS PUMS shows very limited geographic detail, and that other federal surveys, such as the CPS, provide microdata for the full sample. There was informal discussion after the session, but the debate ended with the Census Bureau justifying the C2SS PUMS sample, and the critics not buying these explanations. The C2SS PUMS critics also took the opportunity to express displeasure with the structure of the product, which among other things, lists variables alphabetically, thus scattering variables by subject.
In response to a question on adding questions to the ACS, Nancy noted that (congressional scrutiny notwithstanding) ACS content is pretty well established for now, and that the next opportunity for new or revised content is 2006.
Special Celebration: Before breaking for lunch, Ed Spar distributed cupcakes, and staged a small ceremony acknowledging the 90th birthday of COPAFS lifetime member Margaret Martin.
Federal Statistical Agencies' Responses to Data Quality Legislation
Nancy Kirkendall, Energy Information Administration
Nancy described the Data Quality Act (DQA) as having a potentially huge impact on federal agencies-perhaps especially those that are not statistical agencies, and which have less experience with data quality issues. As Nancy described it, the "Quality Act" passed as a rider to the Treasury appropriations act for FY2001-without hearings or legislative history. Reduced to its basics, the act requires federal agencies to take responsibility for the quality of the information they disseminate-an easy enough idea, but one that is not so easily implemented.
To assist agencies in complying, OMB has issued guidelines that are now in a period for public comment, but time is short, as agencies must issue their own quality guidelines by October 1, 2002. Among the requirements of the OMB guidelines are that 1) agencies shall adopt basic standards for data quality, and hold themselves accountable to performance goals; 2) agencies shall develop a process for reviewing the quality of information before it is disseminated; and 3) agencies shall facilitate "citizen review, and the timely correction of information that does not comply with OMB or agency guidelines.
Again, Nancy described the statistical agencies as perhaps better prepared to deal with the DQA. In response, they have formed an Interagency Council on Statistical Policy to pursue a consolidated approach. The member agencies have issued a joint Federal Register notice (available on fedstats.gov), which provides a statement of philosophy, and a list of statistical activities. Nancy remarked that "you wouldn't believe how long it took the agencies to agree on this list," but noted that, through their team approach, the statistical agencies seem to have a big head start on the DQA process.
Nancy then described some conceptual issues related to the Data Quality Act. For instance, data "quality" is described as relating to "utility" (usefulness), "objectivity" (accuracy and lack of bias), and "integrity" (security from unauthorized access). Also important is the extent to which data are "influential," or have a clear impact on public policy and/or private sector decisions. Apparently, there is debate about the importance of this concept, as not all data are "influential." Data quality standards also address the extent to which data are "transparent" (having well documented methods and sources) and reproducible (could be replicated by others if they had access to the required resources). The implication seemed to be that the more "influential" the data, the more important it is that they be "transparent" and "reproducible."
Nancy finished with a reminder that the draft guidelines are out for comment, and a recommendation that we take a look at them. In response to a question about timeliness, Nancy said she regards timeliness as a dimension of quality, but said there has not been much discussion of it. She then distributed a list of useful web links (see below).
Updating the Consumer Price Index - Latest Developments
John Greenlees, Bureau of Labor Statistics
John's focus was on CPI developments since his last COPAFS presentation in June 1999. These developments are summarized under the "CPI Improvement Initiative Projects," and include
- An expanded consumer expenditure survey sample
- An accelerated expenditure weight update system
- A superlative CPI series
- Hedonic qualitative adjustments and new goods.
John presented a slide showing previous and updated weights, reflecting the relative importance of various product groupings to the CPI. Relative importance changes over time, and as an example, John noted that "food away from home" is approaching the importance of "food at home." As one might expect, personal computers and cell phones also are gaining in relative importance.
Next, John described a new index that is to be introduced in August as an alternative measure. The new C-CPI-U does not replace the CPI-U (whose applications will continue), but unlike the standard measure, is subject to revision. Changes to the CPI-U itself include the hedonic quality adjustments, which are being expanded to include products, such as VCRs, DVDs, televisions, personal computers, and even washers and dryers. Additional work being undertaken as part of a "CPI Continuous Updating Initiative Project" relates to outlet rotation, within-outlet rotation, and computer system replacement.
John then described a report produced by a CNSTAT panel that was asked to look at the conceptual issues in cost of living measurement. The panel made 15 recommendations-all for research, and none of the "do this" variety. For example, the panel recommended expanded development and testing of hedonic methods, but also recommended more cautious integration of hedonic quality adjustments. John closed with a discussion of CPI bias. He explained that BLS has no official position on the persistent questions concerning an upward bias in the CPI, but cited recent research suggesting lower bias than was indicated in earlier research. In other words, recent CPI revisions seem to have reduced an upward bias that BLS has never formally acknowledged.
Concerns of COPAFS Constituencies
The meeting adjourned at 3:00 pm.
Submitted by Sarah Zapolsky, AARP, Secretary COPAFS with gratitude to Ken Hodges for providing significant input and content.
- Bob McGuckin American Economic Association
- Bill Lidington Beyond 20/20
- Dick Kulka RTI
- Don Muff Muff Consulting Svcs.
- Dorothy Harshbarger NAPHSIS
- Douglas Skuta ESRI
- Ed Godfield CNSTAT, Census
- Ed Sandil NCHS/CDC
- Ed Spar Copafs
- Fred Cavanaugh Sabre Systems, Inc
- John Knapp AUBER
- John Cromartie AAG
- Judy Mopsik Abt Associates
- Ken Hodges PAA
- Kent Hargesheimer SRC
- Kelvin Pollard PRB
- Linda Jacobsen Claritas
- Margaret Martin COPAFS
- Mark Nord RSS
- Nadine Hamilton USDOT/BTS
- Nicholas Zill AAPOR
- Pat Doyle ASA
- Patricia Becker APDU/SEMCC
- Rick Ayers ESRI
- Sarah Zapolsky AARP
- Stephen Tordella Decision Demographics
- Steve Landefeld BEA
- Tim Tang NEA
- Tim Jones Census
- Ralph Rector The Heritage Foundation
- Jane Smith FenestraTech
- Maurine Haver NABE
- Ken Bryson ACS Census Bureau