Fritz Scheuren, Urban Institute

Even though my role was simply that of chair today, let me comment briefly on the three excellent papers at this session, particularly making connections with other work. My comments follow the order of presentation. There were also important floor comments and highlights from these are given below as well.

This paper gives a fine review of what is going on internationally on the possibilities of register-based censuses and how these might mesh with traditional enumerations and surveys. Blums work has a strong personal and professional interest for me (Scheuren 1999a). In my opinion, her paper should be of great significance to official statisticians (e.g., Prevost and Leggieri 1999).

What is so good about this emerging area is that it has problems enough for all of us. Many policy issues exist, including privacy concerns (Scheuren 1999b), plus computational challenges since the files being used are often quite large. Of course, as has been pointed out elsewhere at these meetings, there are plenty of interesting mathematical statistical issues too (e.g., Thomsen and Zhang 1999).

Ezzati-Rice, Cynamon, Blumberg and Madans Paper

This paper introduces us nicely to a major new survey effort. The surveys name, State and Local Area Integrated Telephone Survey, while descriptive, may well have been chosen in part for its delightful acronym, SLAITS. While SLAITS still has experimental elements, it promises to improve our state-by-state knowledge of health and welfare issues enormously.

SLAITS is clearly an excellent addition to other surveys that cover portions of the same ground, including the National Survey of American Families which I currently direct (e.g., Scheuren and Wang 1999). Another reference that may be useful as background is Brick (1999) which examines 10 major national US surveys that partially overlap SLAITS. In any case, the authors make a convincing case for the need to continue and expand their survey. They also touch on some areas of research in telephone surveys, including advances in how to deal with nontelephone households. To their presentation, I would add references to similar work found in Report No. 16 of the National Survey of American Families Methodology Series.

Johnson Paper

In 1997 the National Agricultural Research Service was granted the responsibility of conducting the US Census of Agriculture. This significant improvement in the federal statistical system, while leading to some growing pains, has been a real boon in efficiency and in methodological innovation.

The Johnson paper continues the sharing of information on these changes. Unlike the other two papers at this session, Johnson does not focus on demographic information but on economic data. One of the nicest aspects of the Johnson presentation was the presence of Don Bay -- NASS Administrator during the entire transition of responsibility for the Agricultural Census. Don, who retired a few days after the Conference, has very much deserved the appreciation and recognition he has received for encouraging efforts like that described in the Johnson paper.

Floor Discussion

The papers were all well received by those attending and many clarifying comments were made by the authors. I also wove in the comments shown above. Two issues engendered some general discussion and so are reported on below specifically.

During the floor discussion M.P. Singh, from Statistics Canada, questioned the characterization of the SLAITS response rates as good when, using the CASRO definition, they were somewhat under 70%. This lead to a general discussion of nonresponse bias issues and the growing (and legitimate) concerns that exist in using telephone surveys because of trends in response. I did mention that the CASRO rates may be too low because of the way ring-no-answers were handled. Recent work that I have been a part of suggests that, in telephone surveys like SLAITS (with many, many callbacks), response rates could be a good bit higher than reported (For this alternative calculation method, see Brick, Montaquila, and Scheuren 2000).

Miron Straf, of the National Academy of Science, returned us in his floor discussion to the themes that John Bailar emphasized in his keynote address on the first day of the Conference. Questions are what really matter, John had said. In the long run, therefore, it is what questions government needs to get answer too that will define the federal statistical system.

Now the federal statistical system is almost certainly going to lose information market share as the internet age expands. Managing this inevitable outcome will be a major challenge. For example, with the expansion in the use of administrative data, the federal statistical system almost certainly will increasingly become an integrator of information and less and less a provider of primary data. This is implied by the Blum paper, among others, at the Conference.

David Moores 1998 ASA Presidential Address on statisticians being the Wrong Kind of Nerds deals with a related point and his remarks may have a special force in government. It might be added that in the private sector the name statistician is being replaced by quant. Of course, the root for the word STATISTICIAN is the German STAAT for state. This change in vocabulary may be no coincidence.

References

Brick, M., Montaquila, J. and Scheuren, F. (2000) Methods for Estimation of Residency Rates for Undetermined Numbers to be presented at AAPOR 2000. Also submitted to Public Opinion Quarterly.

Brick, P (1999) A Review of Ten National Surveys, to found in Report No 16 of the National Survey of American Families Methodology Series, available on line.

Prevost, R. and Leggieri C. (1999). Expansion of administrative Record Uses at the Census Bureau; A Long-Range Research Plan, 1999 Federal Committee on Statistical Methodology Proceedings.

Scheuren, F. (1999a) Administrative Records and Census Taking, Survey Methodology.

Scheuren, F. (1999b) Discussion of Papers by Bissett; Thomsen and Zhang; and Prevost and Leggieri, 1999 Federal Committee on Statistical Methodology Proceedings.

Scheuren, F. and Wang, K. (1999), National Survey of American Families; A Mid-Course Status Review, to be found in Report No 16 of the National Survey of American Families Methodology Series, available on line.

Thomsen, I. And Zhang, L. (1999) The Effects of using Administrative Registers in Economic Short Term Statistics: The Norwegian Labour Force Survey as a Case Study, 1999 Federal Committee on Statistical Methodology Proceedings.