Minutes of the JUNE 4, 2010 COPAFS Meeting

Ed Spar. Executive Director’s Report

Spar observed that the new COPAFS website has been live for a while, and expressed appreciation for the positive feedback received so far. He was also pleased to report that the Census Bureau’s Debbie Griffin has received the Jeanne Griffith mentoring award. Spar also described, and passed around a copy of Bill Frey’s updated “State of Metropolitan America” publication, available now from the Brookings Institution.

Turning to budgets, Spar said not much is known at this point, but he noted that, if approved, the budget provides for Census Bureau work (with BEA and BLS) to test enhanced applications of administrative records data. Spar also described the supplemental poverty measure, and the controversy surrounding it. The supplemental measure is to be available by September of next year, and there is a Federal Register notice seeking comments on it now. Working with the Census Bureau, BLS is moving ahead on a revised Consumer Expenditure Survey, and Spar hopes to schedule a presentation on the revisions to this and other surveys later this year. Spar also recommended a Stephen Landefeld article “GDP and Beyond,” which is available online, as well as a recent New York Times article (Rise and Fall of the GDP) reflecting controversies surrounding the GDP measure.

The remaining 2010 COPAFS meetings are September 24 and December 3.

Tracking the Revolution in Health Information Technology
Richard Moser. National Cancer Institute

Moser noted the rapid change in technology, the ways people access information, and the impact on health information. For example, the majority of those visiting the National Cancer Institute website are patients and their families, rather than practitioners. With information so accessible and in such high demand, it is critical that agencies be effective communicators of health information.

This context is the genesis of the Health Information Technology Survey (HINTS), which started in 1998 with the objective of providing information on knowledge, attitudes, and behaviors related to health issues. Based on a national probability sample of the adult noninstitutional population, HINTS is a surveillance vehicle (measuring, for example, how many women get mammograms), and a research vehicle. Data were collected in 2003, 2005, and 2007-2008, and a fourth survey is in the planning stages. Westat is the main contractor on HINTS.

Declining response rates are a challenge, and to boost response, the CATI/random digit dial methods used in 2003 and 2005 were expanded to dual telephone/mail collection for 2007-2008. Experiments with incentives confirm that modest financial pre-incentives improve response, but the rate is still only about 40 percent. Offering an Internet response option actually reduced response rates, as many respondents who said they would respond on the Internet did not follow through.

Moser described some HINTS questions and findings. For example, when seeking information on cancer, people go first to the Internet rather than physicians, and the Internet advantage is increasing. However, people express lower-and decreasing-levels of trust in the information acquired from the Internet. This finding prompted discussion about what it really means. For example, it was suggested that people seek different types of information from the web and from physicians, and that Internet information could include newspapers, radio, blogs and many other sources. In short, the Internet/physicians comparison may not be “apples to apples.”

Moser then described HINTS 4, which seeks to use a more collaborative or participatory approach. Dubbed HINTS - GEM (grid enabled measures), the new approach seeks to create a virtual community of scientists and researchers, and a process for vetting survey items based on an active exchange of opinions among experts. The objective is to achieve agreement on standardized, valid and useful measures for future surveys.

Looking to the future, Moser noted that they are reaching out to the Department of Health and Human Services and the Food and Drug Administration, in an effort to broaden the scope of their work.

A Review of Quarterly State and Local Tax Data
Christopher Pece. U.S. Census Bureau

Pece explained that the objective of the Census Bureau’s Governments Division is to provide uniform information on state and local governments, and to follow the activity of government units. Most data are gathered through surveys of governments, which are voluntary, and have virtually no confidentiality restrictions.

Among the "base programs" is the Census of Governments, which gathers finance, employment and organizational data on government units. There are also reimbursable programs - surveys conducted on behalf of other agencies, which compliment the base programs. Pece noted that some of the Division’s programs have long histories, even if under different names, and there is a need for modernizing and reengineering much of their work. A 2007 CNSTAT report offered 21 recommendations for improving existing programs, and Pece said Governments Division is working hard to address the majority of these recommendations. The recommendations call for improvements to the quality, relevance and utility of data, as well to dissemination and timeliness. Efforts at modernization and reengineering include a research program and data user workshops.

Pece then described progress on the redesign of the Quarterly Tax Survey. Conducted since 1962, the survey is one of the division’s most widely used data products, and provides estimates for three components of tax revenues - 1) local government property taxes, 2) local government non-property taxes, and 3) state taxes. The CNSTAT report called for improvements to this survey, and Pece described improvements being made to the three components. For example, the local property tax component is getting a new sample, and redesigned imputation methodology and edit processes. The non-property tax component is being redesigned from the ground up with a new form, and a focus on those who impose taxes, rather than those who collect them (private companies, such as cable TV, collect some taxes for local governments). The state tax component also is getting redesigned edit and imputation methods, and the data are being compared to the annual tax survey.

Next steps on the Quarterly Tax Survey include estimates of property tax with coefficients of variation (summer/fall 2010), improved edits, and classification research for hybrid taxes (taxes that do not fit traditional classifications). Next steps on modernization and reengineering include continued involvement with the user community, implementation of a Quality Improvement Program, and content review. For updated information Pece recommended a visit to their website at

Concerns From COPAFS Constituencies

In response to a Board suggestion, Spar moved the solicitation of COPAFS rep concerns up in the agenda. In response to a question about the lack of progress reports on the 2010 census operation, Spar said it was a good question, but that he was deliberately waiting to schedule such updates until later, when the Census Bureau could provide more conclusive information.

Developing a Supplemental Poverty Measure
David Johnson. U.S. Census Bureau

The proposed FY 2011 budget includes funding for BLS and the Census Bureau to develop an alternative to the long-criticized official poverty measure. The supplemental measure will be based on recommendations from an interagency technical working group, and National Academy of Sciences recommendations. It will not replace the official measure, nor will it be used for resource allocation or program eligibility.

The official measure dates to the 1960s, and defines poverty relative to a threshold (three times the cost of a thrifty food plan) that is updated each year based on change in the Consumer Price Index. Resources are based on cash income before taxes and transfers.

Among the criticisms of the official measure are that it does not:

  • Distinguish between the needs of workers and non workers.

  • Recognize variations in medical care costs.

  • Reflect geographic price variations.

  • Make regular family size adjustments.

  • Reflect increases in standard of living.

  • Reflect government policy initiatives such as in-kind benefits and tax credits.

Johnson had opened his remarks by considering the basic definition of poverty, and cited the NAS panel’s view that it is a lack of economic resources needed to obtain a minimally adequate standard of living, defined appropriately for the United States today. From this view, poverty is a relative concept that can change over time, and the supplemental measure seeks to reflect a relative definition of poverty. As such, the supplemental measure indicates higher levels and greater increase in poverty. Johnson observed that there is great resistance to replacing the official measure, but with the widely recognized need for improvement, he described the development of the supplemental measure as a "just do it" proposition.

Johnson presented a side by side comparison of the official and supplemental measures with reference to the basic "what," "who," "when," "where," and "how" questions.

What: Which resource measure is used?
Who: Whose resource is measured?
When: What time period is used?
Where: Do the measures differ by location?
How: Which summary measure and thresholds are used?

HOW: The official measure is the cost of a minimal food diet, times three to cover other expenses. The more complicated supplemental measure is based on food, clothing, and shelter, with the threshold set at the 33rd percentile of consumer units with two children plus 20 percent to cover other expenses.

WHO: The official measure has separate thresholds by family type. The supplemental measure is based on a reference family, adjusted based on a three parameter equivalence scale. The supplemental also adjusts by type of shelter (renters vs. owners with mortgage, vs. owners without a mortgage). The unit of analysis for the official measure is families and unrelated individuals. The more expansive supplemental measure includes all related persons, along with foster children and other co-residents.

WHERE: The official measure makes no adjustments by geography. The supplemental measure adjusts for housing costs by metropolitan and within state non-metropolitan areas. Johnson noted that these adjustments have a greater effect on the distribution of poverty than on the overall rate.

WHEN: The official measure updates the original 1963 thresholds annually based on change in the CPI-U. The supplemental measure recalculates thresholds each year based on the Consumer Expenditure Survey. The adjustment factors also are regularly updated.

WHAT: The official resource measure is based on before tax income from the Current Population Survey. The supplemental measure is based on this income, plus federal in-kind benefits (such as housing subsidies and tax credits), minus income and payroll taxes, and other nondiscretionary expenses (such as child-care, other work-related expenses, health insurance premiums and out of pocket medical expenses). As Johnson described it, the objective is to reflect a more complete measure of resources and expenses.

To what extent does the supplemental measure overcome the shortcomings of the official measure? Johnson answered his own question with a slide that puts a check mark by each of the shortcomings listed in his earlier slide.

Next steps include the creation of an interagency steering committee, a Census/BLS development team, a Federal Register notice (out now) to solicit methodological comments, and a December 2010 documentation of methods. The first release of the supplemental poverty measure is targeted for September 2011, on the same day the official measure is reported. Technical issues remain, such as those related to the measurement of expenses, adjustments for the uninsured, models for in-kind benefits, and the under-reporting of resources.

More information can be found on the Census Bureau website, and will soon be available on the BLS website. Johnson suggested trying the "table creator" function with which one can create alternative rates to see how the measure works.

Proposed Geographic Support Services and the Status of the Master Address and TIGER files
Timothy Trainor. U.S. Census Bureau

Trainor described a proposed budget increase for work to improve the Census Bureau’s Master Address File (MAF). The activities would build on the accomplishments of the recent MAF/TIGER enhancement program (MTEP), and support the goal of a targeted address canvassing approach for the 2020 census, versus the complete canvass conducted for 2010. The activities are a response to stakeholder and oversight recommendations, as well as internal awareness of the importance of a successful MAF.

Between 2003 and 2008, the MTEP improved the positional accuracy of TIGER, and the objective is to build on the success of that $200 million effort. The focus is on improving address coverage, and the quality of address information - while maintaining the quality of spatial data. The work will have three integrated components - address updates, street/feature updates, and quality measurement (quality addresses are those that can be found by the census).

With respect to addresses, the goal is to have complete and current coverage of addresses, even in areas without city style addresses. The challenge is how to update addresses in such areas without conducting annual census field operations. Current sources are the USPS Delivery Sequence File, and field operations associated with surveys taken throughout the year. Neither would provide the updates needed, so the objective is to get the "best available data" from partnerships and commercial files. Another challenge is to define what constitutes “best available.” With respect to features, plans call for the continuous updating of street networks and attributes in collaboration with the U.S. Geological Survey. Again, the focus is on the best available data from partnerships and commercial sources, and extensive use of imagery is anticipated to help with quality assurance. Quality improvements will apply to address and spatial data, as well as IT processes and geographic products.

Targeted address canvassing is a major objective, and would reflect a shift from canvassing each street in every area to the generation of acceptable address lists from sources such as local governments. Determining what constitutes an "acceptable" address list is a challenge, and would be done in collaboration with government units, with address quality being a major consideration. Only areas where acceptable address lists cannot be acquired would be targeted for traditional address canvassing by the Census Bureau. The Census Bureau needs to determine, by 2015, if they can do targeted address canvassing, so there is much work to be done between 2011 and 2015.

The improvements described by Trainor are currently waiting for 2011 budget approval, but Geography Division management has had planning sessions, and the Census Bureau is forming teams to address topics including sources of address and feature data, policy, partnerships, and quality assessment. The Bureau also is seeking out external expertise, such as from USGS. Asked about the savings that targeted address canvassing could provide, Trainor said the proposal identifies a range of possible savings, but suggested the savings could be on the order of $100 million.