STANFORD UNIVERSITY SCHOOL OF MEDICINE
STANFORD MEDICAL CENTER
300 PASTEUR DRIVE, PALO ALTO, CALIFORNIA

DEPARTMENT OF GENETICS November 12, 1963 DAvenport 1-1200

Ext. 50532

Dr. Bertie H. R. Woolf

Executive Secretary

Human Ecology Study Section Reference: NIH grant HD-00045-01
Division of Research Grants

National institutes of Health

Bethesda 14, Maryland

Dear Dr. Woolf:

| was happy to learn from Dr. Bodmer of the constructive results of his

visit ‘with you in company with our colleagues at the Bureau of the Census

and the National Institute for Child Health and Human Development. | believe
that you brought up the question of a site visit, and I hasten to say that we
would welcome this as a most constructive means of clearing up any of the
questions that you and the study section might have, and also as a means of
eliciting specific constructive comments on the specifications of the census
tabulations which are the main substance of our request.

i am particularly encouraged by the cooperative spirit of the discussions

that Dr. 8odmer reported on. This should be a promising sign for establish~
ment of a meaningful understanding involving the federal bureaus and academic
scientists like ourselves interested in the efficient application of popula-
tion data for medical problems. After all, the census is unquestionably the
most extensive source of data on human biology, and Jtwould be foolish not

to utilize reasonable opportunities for taking advantage of the very large
investment that goes into the production of this file for a wide variety of
important health-related research questions. .As indicated in our application,
we are most interested in establishing output files that would be of the
greatest usefulness for further research, and also in relying on the advice
of a number of our colleagues in establishing these specifications. Dr. Bodmer
has already taken steps in this direction, for example, during his attendance
at the recent Macy Conference on human genetics.

| enclose some remarks that Dr. Grabill from the Bureau of the Census has
summarized on the utility of childspacing information for the analysis of
fertiiity. | would second these remarks and ask that they be included,
together with his letter, as part of our application, but | would put special
Stress on item 5 of his outline, the significance of these studies as a control
base for other health statistics with special reference to (1) mental retarda-
tion, (2) the perinatal morbidity study now under way under the auspices of the
NINDB, especially a more detailed analysis of the relationship between maternal
age and parity of birth, and duration since the last previous birth. In our
Or. Bertie H. R. Woolf November 12, 1963
Page 2

application | have already alluded to the connection between this analysis
and the understanding of the etiology of mongolism, which is so strongly
influenced by maternal age.

Our application also instances several other studies which have been clearly
enough formulated, for example, family structure with respect to sequence of
sexes and with respect to the effect of sex of a given child on the Interval
to the next one.

A very important by-product of any of the specific tabulations which are
produced as the basic purpose of this research will be the production of
pre-tally data tapes which contain most of the essential information for the
further study of distributions, but from which all confidential information
has alreedy been deleted, and which can therefore be made more conveniently
and widely available than is possible for the original census file.

Finally, as our discussions have initialled, our cooperation with the Census
Bureau will make it possible to produce statistical measures of distributions
of family data for research purposes which are painfully lacking in published
tabulations which are used primarily for meeting requirements for information
on a different basis than for demographic research. These measures, prin-
cipally the series of moments of the distributions in the tabulations, are
important not only for an understanding of the data as they are usually pre-
sented, but because of the adjustments and poolings that are necessarily
involved in ‘the practical production of published tables, are not often
readily re-obtainable from the final output tables themselves.

There are a number of limitations in census data, as we are very well aware,
and there are, of course, a number of sources of bias that must be taken
account of. It would be most constructive to discuss these in more detail
with a site visiting group; for example, as cannot be reiterated too often,
the census. data refer to households rather than families, and measures must

be taken to account for families with incomplete membership in the household.
It is not sufficient to merely count those households which have complete
families, although this criterion can be established by comparing the reported
total number of children ever born to the number of children presently resident,
since there can be a significant bias, for example, in the sex of absent
children, both from sex=dependent differential mortality and age at departure
from the family group. Since the main bulk of the conputational effort in
this project will be borne by the Bureau of the Census, it is perhaps in the
elucidation of problems of this kind that most of our own creative effort

will be spent jin cooperation with them, .

Whatever the limitations, in certain respects, of the present data, they
represent too valuable a file to overlook in studying these problems. Fur-=
thermore, only such a study can establish the basis on which future census
enumerations can be conducted in a biologically more meaningful context.

Sincerely yours,

Joshua Lederberg
Professor of Genetics