Minutes of the June 9, 2006 COPAFS Meeting

COPAFS Chair Sarah Zapolsky started the meeting, and we went directly to Ed Spar’s executive director’s report.

Ed Spar. Executive Director’s Report

Ed reported that there is little to talk about on the budgets, but noted bad news in the form of $5 million removed from the NCHS budget – a cut that could have a negative impact on their surveys and collection of vital records. COPAFS continues to work with “Friends of NCHS” in support of the agency’s programs. Also troubling is that the Bureau of Transportation Statistics still has no director. However, the agency is assuring that its commodity flows survey is going forward.

Next, Ed drew our attention to example wording and formats for some questions being considered for addition to the American Community Survey. These include health insurance coverage, marital history, and disabilities connected with veteran status. It seems that congressional support for the ACS is being expressed in the form of proposed additional questions, but the Census Bureau is concerned that added questions could burden the still-new and fragile survey.

Ed also referred us to the Census Bureau’s new ACS Methods Paper, and a GAO report on standards for key dissemination practices. He also described a recent Brookings Institution meeting on the suspension and planned re-engineering of the Survey of Income and Program Participation (SIPP). Major issues include funding, the integration of administrative records and survey data, and the gap between the termination of SIPP and the start of its successor. The Census Bureau needs to know what users really need, and will have a meeting with data users on this in July.

Ed concluded by explaining that the Senate version of the immigration reform bill has a provision requiring the Census Bureau to investigate how to exclude illegal aliens from the census count. The proposed legislation is similar to the proposed constitutional amendment, which would require the broader exclusion of non-citizens from the census count.

Remaining COPAFS meetings this year are September 15 and December 8

An Update From the National Center for Education Statistics

Mark Schneider. National Center for Education Statistics.

When Ed introduced him as the “new, or maybe not so new” commissioner of NCES, Schneider chuckled and described the life cycle of NCES commissioners as progressing from “nominee” to “new,” “embattled,” and “former” stages. He described NCES as a very good agency, with a history of collecting survey data on students and educators, as well as administrative data on educational institutions.

Among the data collections is something called the Core of Common Data (CCD), and Schneider described graduation rates as one of the more contentious CCD measures. Graduation rates are complicated by factors such as student mobility that make the determination of a suitable denominator a challenge. The current measure calculates the number of graduates as a percent of students in grades 8, 9 and 10 two years prior.

Schneider noted that the reporting of data by states and school districts is voluntary, and that they are investigating how to impute with administrative data for jurisdictions that do not report. In some cases, he noted, imputation results (if disliked enough) motivate recalcitrant districts to provide data.

Schneider described the Integrated Post-Secondary Education Data System (IPEDS) as one of the more important NCES products, as 40 percent of the agency’s budget relates to post-secondary education. IPEDS is a highly valued database on 6,800 post-secondary education institutions, but there is debate over how to “fix” it. One IPEDS problem relates to the affordability of post-secondary education. Because so few students pay full “sticker price” tuition, and institutions do not disclose “discount rates,” IPEDS is challenged to determine what actual costs are. Separate surveys suggest a national discount rate of about 40 percent – with even the more selective institutions having a substantial discount of about 27 percent.

Another IPEDS issue is accountability, where there is resistance from academics, who argue “how could you judge what I do?” But the accountability issue gets back to graduation rates, where measurement is confounded by transfers, and the fact that 70 percent of post-secondary students are now “non-traditional.” For example, a student who transfers from a two year college to a four year school before receiving an AA degree is currently classified as a “failure.”

In response to its challenges, NCES is trying to make its data more accessible – making shopping for a college more like online shopping for other products. Schneider quipped that they would like to take this function out of the hands of US News and World Report.

Schneider then described the Schools and Staffing Survey (SASS) as a valuable tool, reporting on things like time spent on subject areas, as well as teacher pay and compensation. The measurement of teacher costs is complicated by the fact that teacher salaries are self-report, and that salaries are becoming a smaller part of teacher compensation – with some districts deferring costs by offering added benefits in place of higher immediate pay. SASS also has been plagued with a sluggish release schedule (2003-2004 data were released just this past March), so more timely releases are another objective.

Schneider concluded by noting that historically, NCES has wanted national data, but should be more receptive to opportunities to gather data on individual students and teachers—even if just from selective states. He also described plans for longitudinal surveys—one starting at the critical 9th grade, and focusing on the college selection process.

Health Insurance Data from the National Center for Health Statistics

Eve Powell-Griner. National Center for Health Statistics

Why monitor trends in health insurance? Powell-Griner listed the reasons as 1) improving access to medical care, 2) having an impact on health quality and the cost of care, and 3) evaluating the impact of healthcare policy interventions.

The National Health Interview Survey (NHIS), a major source for this monitoring effort, is a national survey with a large sample that provides information on health status, healthcare utilization, health behaviors, and demographics. About 80 questions are devoted to health insurance (respondents are not asked all questions), including type of coverage, public vs. private source of coverage, single service, and others. An early release program provides partial or preliminary data for 15 major measures in a timely online release.

Powell-Griner then reviewed some recent trends, noting that the percent of total population that is uninsured has been stable since 1997, but the rate has increased for working age adults. While the uninsured rate for children has decreased, it remains higher than for working adults, and there has been a shift to public coverage for “near poor” children. There is concern that the survey has been undercounting the Medicare population, so they have added probing questions to improve the coverage of this segment. State estimates are possible in about 40 states, and data for most other states are provided with two years of survey data.

Powell-Griner reviewed the different uninsured rates identified by three national surveys. The NHIS rate of 15.7 percent compares with a 16.8 percent rate for CPS and a 14.6 percent rate from the Medical Expenditure Panel Survey – with at least some difference expected due to the different questions and reference periods used by the surveys. Strengths of the NHIS include low non-response rates, supplemental coding, and links to other healthcare data. The NHIS also is adding questions to keep up with new developments, such as health savings and flexible spending accounts.

Further information is available at

When asked if NCHS has anything on the insurance status of the undocumented alien population, Powell-Griner was careful in responding. She said they might have some relevant information, but want to be careful in what they release because a focus on the undocumented population could negatively impact response rates.

The American Community Survey: What We Are About to Receive.

Lisa Blumerman, Doug Hillmer and Alfredo Navarro, U.S. Census Bureau.

Lisa Blumerman started with a review of the Census Bureau’s recent release of hurricane impact estimates. One component of the release consisted of special January 2006 population estimates for the 117 counties designated by FEMA for disaster assistance (for hurricanes Katrina and Rita). This is a special set of estimates (not part of the regular annual series) using methods based on USPS National Change of Address data.

The release also includes special before and after tabulations of 2005 ACS data for the disaster assistance areas. ACS products normally would provide one-year data based on data collected throughout the year, but in the hurricane impact area, data for January – August provide a “before” picture, while data collected September – December reflect the post-hurricane period. Blumerman explained that hurricane impact estimates have definite limitations (ACS data collection was thin in the impact areas), but noted that these are the first real data supporting what had been anecdotal reports of changing population size and composition.

Blumerman then provided an ACS update based on an ACS Users Guide that is being developed. She noted the start of group quarters collection in 2006, and the first release of data from full implementation – due this August. Blumerman reiterated the ACS emphasis on characteristics as opposed to counts, and touted the practical uses of the data.

Doug Hillmer then noted that the 2005 ACS products will be much like those released for 2004. Data release limitations guard against the disclosure of information on individuals, and apply checks for statistical reliability. For example, if less than half of the estimates (cell values) in a table are statistically different from zero, a collapsed version of the table is provided—so long as the collapsed table passes this condition. Hillmer confirmed that the five-year average data to be provided starting in 2010 will be exempt from the reliability checks, and resulting collapsing and suppression of data.

ACS products will include Base Tables—similar to SF3 tables, and Hillmer noted that these likely will be presented with “margins of error” rather than the upper and lower bounds we have seen in earlier releases. Other products include Data Profiles (distillations of Base Tables), Narrative Profiles (with verbal descriptions), and Subject Tables (focused on specific topics). New products will include reports by Selected Race, Ethnicity, and Ancestry Groups, and reports by other selected groups (such as population age 60+). Also planned are about 30 base tables tabulated by place of work rather than place of residence.

In response to a comment that ACS summary files are needed, Hillmer assured that such a product is in the works.

The timing of the ACS product releases is as follows:

Mid August: demographic and social characteristics.
Late August: economic characteristics, and PUMS.
Early October: physical and financial characteristics of housing.
November: selected population profiles for race, ethnicity and ancestry groups.

Additional resources include an ACS Design and Methodology Paper (over 400 pages) and the ACS Data User Training Guide (not yet available on the Census Bureau website).

Concerns of COPAFS Constituents

No concerns were raised, and the meeting was adjourned.