Minutes of the September 11, 2009 COPAFS Meeting
COPAFS Chair Judie Mopsik started the meeting, by announcing the December Board election, and inviting those interested in serving on the Board to express their interest via e-mail or other means. We then went to Ed Spar’s Executive Director’s report.
Ed Spar. Executive Director’s Report
Spar recalled the success of recent seminars hosted by COPAFS, and said the organization would continue pursuing seminar opportunities. He directed our attention to the upcoming FCSM conference, noting that almost 300 have registered for it already.
The next COPAFS meeting will feature a discussion on the Census Bureau’s Longitudinal Employer-Household Dynamics (LEHD) program, which Spar hopes will receive additional funding, and which is noteworthy for its work with synthetic estimates.
On budgets, Spar said not much is happening, and a continuing resolution appears inevitable. The Census Bureau is expected to be exempted with an anomaly from flat funding, but that the CR could pose problems for NCHS and other agencies. Spar offered Steven Landefeld (BEA) and Michael Sinclair (BJS) the opportunity to describe the implications of a CR for their agencies, but they responded with polite but guarded remarks. Howard Silver (COSSA) commented that we are seeing a process that plays out every year in which the House takes care of Justice and the Senate takes care of Commerce, with the differences worked out in conference.
The final COPAFS meeting of 2009 is December 4.
GDP and Beyond: Alternative Measures of the Economic Status of Households and Business
Steven Landefeld, Bureau of Economic Analysis
Steven Landefeld explained that the notion of going beyond the basic Gross Domestic Product (GDP) measure dates back to the formation of the accounts in the 1930s. The concern is that the basic measures focus too much on mere economic transactions without reference to the social context, and Landefeld described a disconnect between GDP and the things that people see as important in their lives.
Commenting that BEA has been burned by dabbling in alternative measures (such as Green GDP), and that they will not go so far as the “index of happiness” considered by some, Landefeld assured that they are actively exploring alternatives to the basic GDP. As examples, he cited work on measures of economic progress (how people or businesses are doing relative to the basic GDP), and described what he called the issue of sustainability.
Landefeld showed numerous slides illustrating alternative measures and the differences they suggest for the periods 2000-2007 and 2008. For example, much of the growth in spendable cash income is in the form of supplemental sources, which if removed from the basic measure, show a much less favorable picture of personal income.
Turning to sustainability, Landefeld described net versus gross measures, including net versus gross domestic income and net versus gross domestic product. GDP does not account for things such as depreciation and the investments required for future production, and thus does not tell us how sustainable economic activity is over a long period of time. Landefeld showed a number of graphs to illustrate his point. One was a time series of household net worth and savings. Savings had been dropping while net worth increased, largely due to the bubble in housing values and stock prices. With perceived wealth, people felt less need to save, but the bubble was not sustainable. Another tracked GDP, profits, and equity prices. Equity prices experienced dramatic increases without corresponding increases in profits and GDP, and were therefore not sustainable.
Landefeld said their work on alternative measures comes at a fortuitous time, given popular doubts about the value of economics and economists, in the wake of the economic downturn. Economists have been criticized for not foreseeing the recent economic problems, and Landefeld sees BEA’s job as that of providing economists with the tools to help them do a better job. He closed by asserting that BEA has done a good job of fulfilling its current mission with the basic measures, but acknowledged that the agency can provide useful additional information with alternative measures.
New Initiatives for FY 2010 at the Bureau of Justice Statistics
Michael Sinclair, Bureau of Justice Statistics
BJS Acting Director Michael Sinclair commented that the agency’s staff has been working hard to take a fresh look at its programs, including the National Crime Victimization Survey (NCVS), which accounts for most of BJS’s expenditures, but is just one of its many data products.
Sinclair started with an overview of the NCVS, which has two major objectives – first, to produce estimates of the incidence of victimization (of both reported and unreported crimes), and second, to measure the characteristics and consequences of crime. Sinclair described these objectives as competing, in that they would be best accomplished with different survey designs. BJS has worked to balance these objectives, but they hope to make improvements with a re-design focusing on restoration and renovation.
Restoration includes near-term objectives such as improving the base program (within budget constraints), and increasing sample sizes to improve data precision. They also hope to improve the Census Bureau’s operations with upgraded laptops, additional staff, and field interviewer training. Long-term objectives include the creation of a quality improvement program, and the addition of new content.
Renovation objectives include small area estimation, research on synthetic estimates, questionnaire revisions, and different sampling strategies. Plans call for field testing of new methods in parallel with current methods, and implementation of the new program is targeted at 2013.
Sinclair summed up the NCVS work by noting that they look to maintain the core of the current survey, but add a supplemental strategy involving new sampling methods and new data collection procedures. The supplements could piggy-back on the existing survey, but they are also considering the possibility of a parallel survey for direct measures, and the use of administrative data for synthetic estimates.
Initiatives beyond the NCVS include surveys of local sources to get an early read on changes in advance of the Uniform Crime Reports, and parallel surveys of law enforcement and the public to better understand police-public interactions. Consideration also is being given to the re-design of the State Court Processing Survey and the National Judicial Reporting Program – perhaps combining the two in order to reduce costs. Sinclair also described initiatives related to indigent defense, the conversion of criminal history records into research databases, and the surveys related to PREA – the Prison Rape Elimination Act.
Sinclair wrapped up by describing the improved and more user friendly BJS website, and said they are looking forward to user input on it.
Demographic Analysis for the 2010 Census
Victoria Velkoff, U.S. Census Bureau
Jason Devine, U.S. Census Bureau
Because Demographic Analysis is a team effort, Tori Velkoff started by introducing Kirsten West and Gregg Robinson, who have worked on DA for many years. There is a long history of DA at the Census Bureau, but they are working on changes for 2010, and are looking for feedback. Velkoff also explained that the DA estimates are unique among Census Bureau estimates in that they are not based on census counts, and they are important because they provide the first measures of census coverage.
Jason Devine described DA as one of two methods the Census Bureau uses to measure census coverage. The other is the micro-level survey-based estimates based on case-by-case record linkages (as in the 2000 A.C.E.). In contrast, DA is a macro-level approach that compares census population counts with estimates based on administrative data. The DA estimates are independent of the census counts. The 2010 DA estimates will build on the 2000 DA estimates, not the 2000 census counts.
The DA method tracks the size of population cohorts based on administrative data. Specifically, population less than age 65 is estimated with the basic demographic accounting equation which adds births subtracts deaths, adds immigrants, and subtracts emigrants. For population 65 and above, the estimates reflect the Medicare count plus an estimate of persons not enrolled in Medicare.
DA estimates measure coverage by single year of age, but are limited to national level, and Black and non-Black race. The Census Bureau is researching the possibility of adding DA estimates of Hispanic and non-Hispanic, but the limited race/ethnic data collected in past administrative data makes this a challenge for older birth cohorts. The survey-based coverage estimates provide greater race and geographic detail, but involve their own challenges (see the next presentation). Divine also contrasted the core DA with Demographic Bechmark Analysis (DBA), which compares estimates of population and housing against census counts for various geographic levels.
Devine then presented a number of slides illustrating 2000 DA results – such as population by age. The DA and census numbers are impressively close, but Devine pointed out differences, such as age heaping. Self-reported census age has spikes for ages ending in 0 and 5, while the DA distributions are smooth. Also striking is the census undercount of population age 5, and its over-abundance of persons around age 20.
For 2010, the Census Bureau will revise the 2000 DA estimates based on final 1999 and 2000 birth and death data, then add/subtract births/deaths for the decade 2000-2010. Preliminary data on births and deaths for 2009 and 2010 will be used, and previous assumptions about the completeness of birth and death data are being reviewed. Population age 65 and above again will be based on Medicare enrollment data, with adjustment made for estimated under-enrollment.
When the 2000 census population count came in higher than the DA estimate, it led to questions about erroneous enumerations, but also the migration component of the DA estimates. For 2010, ACS data will contribute to the DA estimates of net international migration. Pending validation, the plan is to estimate immigration based on ACS data on residence one year ago, and foreign-born emigration would be estimated with a residual method based on Census 2000 and ACS data. Estimates of native migration are a challenge, and still rely on previous research, but fortunately, this is a very small component. The Census Bureau plans to look at IRS data for this purpose – the thinking being that native-born persons who emigrate should file a return the following April. Asked if they could use Department of Homeland Security data for this purpose, Velkoff noted that they are looking at DHS data, but that they do not track native emigration. There has also been talk of a CPS question asking about family members who have emigrated, but Velkoff observed that this approach does not seem to work very well.
Velkoff and Devine wrapped up with a review of the 2010 census timeline as follows.
December 2010 – Apportionment data released.
December 2010 – DA estimates of coverage will be developed.
January – March 2011 – Redistricting data released on flow basis.
2012 – Survey-based estimates of coverage will be released.
2010 Post Enumeration Estimats Using Census Coverage Measurement
Vincent Mule, U. S. Census Bureau
Vincent Mule started by stressing that the purpose of Census Coverage Measurement (CCM) is not to adjust the census counts. Rather, it is to estimate the net coverage of the census and the components of census coverage for persons in housing units and for housing units (CCM does not cover group quarters).
The design involves a sample of the population (the P sample), and an independent sample of housing units in block clusters (the enumeration, or E sample from the census itself). The CCM program is underway now with the independent listing of addresses in the sample areas (August – December 2009). Following the census, person interviewing is scheduled for August – October 2010, followed by person matching (October 2010 – April 2011), and final housing unit matching (April – July 2011). The timing is later than 2000, when A.C.E. results were required in time for potential use in adjusting the census counts. Release dates for the 2010 CCM results are still under consideration, but Mule indicated that they would be out sometime in 2012.
Mule then provided some details. As in 1980, 1990, and 2000, CCM for the 2010 census will be based on dual system estimation, but will use regression modeling instead of post-stratification. The 2010 design also will introduce correlation bias adjustment for males age 18 and above.
Mule explained the basics of dual system estimation with the familiar 2 by 2 table with cell properties as follows.
N11 Person in P sample and in E sample.
N12 Person in P sample but not in E sample.
N21 Person not in P sample, but in E sample.
N22 Person not in either sample.
As Mule described it, CCM measures census coverage based on numbers in the first three cells, but is unable to account for persons (or housing) in the N22 cell.
For 2010, Mule said they have been instructed to provide more on the components of census coverage. For persons in housing, CCM will identify 1) correct enumerations, 2) erroneous enumerations, 3) census imputations, and 4) omissions. For housing units, CCM will identify 1) correct enumerations, 2) erroneous enumerations, and 3) omissions.
Mule then described the difference between net error and components of error, and elaborated on some definitions. For example, erroneous enumerations can be either duplications or persons who should not have been counted at all (such as those born after census day, or who died before census day). There are no direct estimates of omissions, but since net error is equal to omissions minus erroneous enumerations, omissions are estimated as net error plus erroneous enumerations. In addition, 2010 CCM has been asked to identify whether omissions were in a housing unit that was identified or not.
Plans call for estimates broken out for three major domains: 1) demographic or housing characteristics, 2) geographic areas, and 3) census operations. For example, the intent is to produce census coverage measures (correct, erroneous, imputation, omission) specific not only to race/ethnic or age groups, but specific to census operations, such as mail response versus non-response follow up. Similarly, housing coverage will be classified by characteristics such as occupancy status, tenure, type of structure, and operational categories, such as address canvassing, LUCA, or field verification.
For substate areas, estimates of net error would be released if they meet a to-be-determined statistical criteria. For persons in housing who should have been counted elsewhere, an additional table has been proposed that would identify the correct residence as being in the same county, different county in same state, or a different state. Another priority is to determine ways to reduce non-sampling error in the 2010 CCM. One possibility is to reduce the CCM sample size – reducing non-sampling error at the expense of some increase in sampling error.
A lively discussion followed. Asked how big the CCM sample size was, Mule recalled it was about 300,000 housing units, and is planned to be the same for 2010 (unless reduced as described above). John Thompson, who was with Census at the time, recalled the 1990 sample as about 150,000.
To shared amusement (and no real answer), Bill O’Hare asked why the Census Bureau calls CCM something different every census (PES, then A.C.E, and now CCM).
Asked if the CCM results would contribute to improvements in the 2020 census, Mule described several applications related to things like determining the impact of operational issues, and learning more about duplications. And when someone posed a hypothetical two percentage point increase in undercount, and a resulting outcry for adjusted census counts, Mule played it straight, saying they are just doing what they have been charged to do.
Concerns From COPAFS Constituencies
No concerns were raised, and the meeting was adjourned.