COPAFS
 

Minutes of the March 10, 2006 COPAFS Meeting

COPAFS Chair Sarah Zapolsky started the meeting with a vote to approve Joe Garrett of Market Strategies, as a new at-large COPAFS board member. Following the vote in favor, Sarah introduced Ed for his executive director’s report.

Ed Spar. Executive Director’s Report.

Ed noted the passing of Joe Waksberg of Westat. Joe was a longtime and strong supporter of COPAFS.

Ed then reported that Jeffrey Sedgwick is the new nominee to head the Bureau of Justice Statistics. Sedgewick’s appointment would fill the one remaining vacancy among statistical agency head positions.

Drawing our attention to the 2007 budget request numbers, Ed noted that the Census request provides for full ACS funding, but that there is a flap over the Census Bureau’s plan to eliminate the Survey of Income and Program Participation in response to a mandated budget cut. This is a hot issue, and the subject of one of today’s presentations. Ed explained that COPAFS continues to work hard for the NCHS budget, but at best things are holding steady. In more positive news, the Bureau of Justice Statistics request includes an extra $10 million to restore the sample size of the Crime Victimization Survey. There had been suspense over whether the Bureau of Transportation Statistics could proceed with the Commodity Flow Survey, but the budget proposal appears to be sufficient, and the survey is in the BTS plan.

Dates for this year’s remaining COPAFS meetings are June 9, September 15, and December 8.

New Outreach Initiatives from the National Archives and Records Administration.

David McMillen. National Archives and Records Administration.

After years with congressional committees, McMillen has taken the new job of External Affairs Liason at Archives. NARA dates back to 1935, and as McMillen put it, has been falling behind ever since. The original mission was primarily about storage, but the agency is re-inventing itself, and now has a greater focus on access. McMillen’s position is part of the new focus, and he expressed particular satisfaction with the commitment to increased collaboration between Archives and the statistical community. The job has McMillen working with professional associations, and he is currently organizing a conference on the skills archivists will need in the 21st century. He is also initiating a conversation on the archiving of websites—a novel idea for an agency that has viewed websites as beyond its jurisdiction.

McMillen commented that the archiving of electronic data is one of their biggest challenges, and that the information explaining the billions of archived electronic records is as important as the records themselves. The challenge is in preserving electronic data such that they can be read accurately and authoritatively 50 or more years in the future. Asked what Archives means by “authoritatively,” McMillen recalled the wax seals once used to verify the source of important documents, and explained that the objective with electronic records is to ensure that they be preserved to look like they did when they first entered the system.

In response to a question about the prospects for automating their work (with Google-like capabilities), McMillen noted that a major part of the challenge is in deciding what to preserve. Google is good at finding things, but does not decide what to preserve, and does not do any actual preservation. Asked how electronic data are preserved, McMillen explained that it involves a lot of migrating of data from medium to medium. For example, Archives is now converting all of its audio tapes to digital media. To the surprise of some, McMillen described magnetic tape as one of the more stable media for electronic data—more so than CDs and DVDs. And in response to a whimsical remark about punch cards, he noted that they have just purchased a punch card reader. McMillen acknowledged that Archives is not on the cutting edge of technology, but it seems the trailing edge continues to be critical to their work.

Asked about the relationship between Archives and the Library of Congress, McMillen described it as “pure competition,” – noting that “we are Administration and they are Congress.” He called it a spirit of friendly competition, and noted that until 1935, the Library of Congress was Archives. Archives has classified as well as declassified and reclassified material, but access to most information is just a matter of finding out what is available. Asked how one does that, McMillen said their website is a good place to start. One can even learn what is available through an old fashioned in-person visit with an archivist, but this option works best if one has a narrowly defined topic.

State and Sub-state Estimation with the National Survey on Drug Use and Health.

Douglas Wright. Substance Abuse and Mental Health Services Administration.

Wright explained that his agency provides state and even sub-state estimates of drug use and health based on a sample survey that is now representative by state. The household survey selects one or two persons (age 12 and older, with age 12-25 oversampled) from each sampled household, and also includes persons in non-military group quarters. The computer-assisted interviews take about one hour, and are largely self-administered. About 68,000 respondents are interviewed each year.

The data are used to monitor the prevalence of substance abuse in the US, identify areas with high rates, and support SAMHSA policies and programs—such as those related to underage drinking and the availability of treatment facilities. Wright noted that while state agencies are interested in the state estimates, the real push for state data is the federal need to monitor the impact of state laws and policies on illicit drugs, and to administer block grants.

The data themselves come in the form of direct estimates for large states with large samples, and model-based estimates in the other states. Wright described their hierarchical Bayes estimation methods, and measures including those related to substance use, perceived risk, dependence/abuse/treatment need, and serious psychological distress. He then presented a series of maps showing state level patterns of cigarette and marijuana use. There was comment from some attendees that states do not vary widely in these rates, and questions concerning the standard error for state estimates.

The objective of the sub-state estimates is to provide data for state and local planners. The sub-state regions are defined by the state users, and often relate to treatment planning or block grant areas. The estimates are produced by accumulating additional data from the survey records. There are about 330 sub-state areas, most are county-based, and due to the large volume, the sub-state data are reported only on the website. Wright presented a series of maps comparing state and sub-state patterns for several of their measures. He finished by describing some methodological concerns. For example, the survey results often are inconsistent with state data collected using different methods (such as telephone interviews), and with data from school surveys, where it is suspected that student respondents may over-report use. There is also concern over the misinterpretation of their data—by users who do not consider the “prediction intervals,” or error, involved.

An Update on the Survey of Income and Program Participation.

Howard Hogan. U.S. Census Bureau.

Hogan, the new Associate Director for Demographic Programs, explained that the Census Bureau has been asked to take a $40 million reduction in its 2007 budget. They have decided against across-the-board cuts, and are protecting the 2010 census and the American Community Survey. The Survey of Income and Program Participation (SIPP) has been selected to take the hit. Hogan acknowledged that SIPP is a survey with passionate users (“and boy, don’t I know it”), but he described it as a survey already in need of re-engineering.

Remaining funds will allow SIPP data collection through September of 2006, but there will be no collection in 2007. Because the data are so highly valued, the plan is to resume data collection as part of a new survey of income and wealth dynamics. To minimize the gap between the old and new surveys, Hogan said data collection needs to begin by 2009 – since there is little chance of initiating a major new collection activity while the 2010 census is in the field.

Hogan expressed the view that the new program should be more than a reconstructed SIPP, and should leverage administrative records and American Community Survey capabilities to provide the types of information previously provided by SIPP. But there seems to be some question concerning what the new system should emphasize. As Hogan put it, every SIPP user is in love with a different part of SIPP – some with the annual data, some the cross sectional data, some the longitudinal data, and so forth. He described the replacement of SIPP micro-data as especially challenging if there is to be more reliance on administrative records. How do you take administrative data, he said, append them to ACS, and get the product by the Disclosure Review Board?

As part of the discussion, Hogan noted that there are no federally mandated uses of SIPP data. However, he stressed that these developments are not an end to the Census Bureau’s commitment to the collection of information on income, poverty, and program participation, but an opportunity to restructure the collection of such data in ways that make sense—apart from budget reduction mandates.

An extended discussion with COPAFS attendees followed. In response to concerns about the data collection gap, Hogan agreed that it is unfortunate, and noted that with adequate funding, they would have overlapped the programs. And in response to skepticism about the resumption of data collection by 2009, he asserted that they do hope to accelerate the new survey. One attendee commented that a “pause” might actually be in order given the limitations of SIPP data, but expressed concern that the Census Bureau maintain its ability to collect longitudinal data. In response to a suggestion that the Census Bureau keep at least some longitudinal collection going (at minimal cost) during the pause, Hogan indicated that they could definitely consider that option.

Continental Drift: Canada and the United States Population Trends and Statistics.

Barbara Torrey. Population Reference Bureau.

Promising a more upbeat presentation (than the one on SIPP), Torrey opened with the observation that the US statistical system is “endearing” compared to the Canadian system. Torrey has been studying the U.S. and Canada, and notes that these differences are growing. And because the two nations are so close in geography and history, these differences can be viewed in the context of a natural experimental design from which much could be learned. For her COPAFS presentation Torrey focused on fertility and immigration.

US and Canadian total fertility rates were similar until about 1975, but have since diverged, with US rates increasing, and Canada’s decreasing. Obvious explanations, such as US race/ethnic diversity, do not hold. The differences also are not explained by benefits (which are higher in Canada), nor do they trace to differences in housing or female labor force participation. In fact, US/Canadian fertility differences are every bit as strong in the US states that border Canada. Torrey also noted that the percent of births out of wedlock is about the same in Canada and the US, overall, but the rate in Quebec is much higher. Common law unions are more common in Canada, and are measured more directly by the Canadian census. As part of the discussion Torrey described the curious US policy of considering Canadian “aboriginal” Indians to be US citizens—eligible for and often recruited for service in the US armed forces.

Turning to immigration, Torrey noted that both nations have an increasing foreign born population, but that growth in Canada is more rapid. And while most US immigrants enter with family eligibility, economic eligibility predominates in Canada. In Canada, 74 percent of immigrants who are eligible for citizenship become citizens, compared with only 37 percent in the US. Torrey reported that the US and Canada have started longitudinal surveys of immigrants. However, the US survey asks about 50 questions on topics such as income and use of benefits. The Canadian survey has a few questions on income, but focuses more on immigrants’ satisfaction with government assistance, and with their healthcare providers. The Canadian survey also asks if immigrants like their jobs, whether the job makes use of their skills, and if they want further training.

Torrey contrasted the centralized Canadian statistical system with the multi-agency US system—which the Canadians regard as statistical anarchy. She commented that during her time at the Census Bureau, Statistics Canada was held up as a paragon of virtue, but she argued that, to the contrary, the US system pays off in healthy competition. For example, Torrey described the Census Bureau and other agency websites as much better than that of Statistics Canada—which she complained has indecipherable text, little data, and charges money if you want any of it. She also criticized Statistics Canada for a lack of timeliness. At a time when the US was providing mortality data for 2001, Canadian data were available only for 1998, and reflected outdated classifications. Calling the Canadian system “slightly dysfunctional,” Torrey said their privacy regulations “make Title 13 look like a breeze.” Despite a “data liberation” initiative now underway, they still withhold a lot of data, and as a result, the Canadians do very little quantitative research.

There was comment that Torrey’s characterization are a bum rap with respect to economic data, which were described as extensive and timely. Torrey did not contest the assertion, and noted that her remarks were limited to demographic and social data.

Concerns of COPAFS Constituents

No concerns were raised, and the meeting was adjourned.