U. S. DEPARTMENT OF COMMERCE BUREAU OF THE CENSUS WASHINGTON 25, D. C. April 22, 1963 IN REPLY REFER TO: [ATR MAIL Dr, Joshua Lederberg, Director Kennedy Laboratories for Molecular Medicine Stanford University School of Medicine Stanford Medical Center 300 Pasteur Drive Palo Alto, California Dear Dr. Lederberg: This is in response to your letter of April 4. We conduct, each month, a national sample survey of the population, cover- ing some 35,000 households across the Nation. This project, entitled the Current Population Survey (CPS), is used to collect information on a wide variety of demographic and associated matters, and also yields employment and unemployment data published by the Bureau of Labor Statistics. Sup- plementary questions are added to the CPS schedule virtually every month, at the request and expense of sponsors outside the Census Bureau. The cost normally runs in excess of $25,000 for items of fairly limited scope, Fortunately, in two reasonably recent months, we obtained month-of-birth data which may be useful to you even though they are subject to the limit- ations mentioned below, The cost would, of course, be lower if we only have to tabulate existing data rather then gathering new data. Copies of “the schedules used in the two pertinent months--August 1959 and March 1962-- are enclosed. In August 1959 we obtained the month-of-birth of the first four and the last (if not one of the first four) children ever born to each ever-married female, If the female was the wife of the head of the household, the occupation (if any) of the head (not necessarily the father) can be obtained by constructing a new tape relating heads of households and their wives. If the head is an ever-married female, her occupation (if any) will be avail- able on her card, The month of birth of the ever-married females (but not males) will also be available. You will not be able to get spouse-to-spouse birth month relationships, but the other birth month relationships you want will be available for most children ever born to ever-married females who were in households in August 1959, The March 1962 survey offers a somewhat different alternative. Here the birth months are available only for each person of any age still living in the household in March 1962, not for children ever born, 2. However, birth month intercorrelations are available for all persons still in the household, assuming all children of the head are siblings, (This assumption is not necessary with the August 1959 data for the mother-child and sibling intercorrelations.) The March 1962 data also will require recasting in a new type of record to meet your needs, In either case, the cost of the tabulations is almost certain to run over $5,000. If you wish to consider either of these operations further, please let us know and we will develop cost estimates based on detailed specifications which you would supply. In conducting the CPS, enumerators mist revisit every household in the sample a number of times during the year. To keep track of the household the enumerator uses a record known as the control card. The control card contains name, sex, relationship to head of household and year and month of birth of all household members, These cards are returned to the Bureau and put in storage when a household leaves the survey panel. The cards could be used to construct a tape file for computer operations. We could link the control card data to the information gathered in a given survey for any month to obtain occupational information, However, the main value of this control card file would be to provide repeated observations of the Nation at varying times showing birth month interrelationship for household members, Any use of these control cards would involve clerical effort and would be more costly than processing the March 1962 data which are already on tape, However, a large number of cases would be obtained at a price lower than the cost of a new field study. , With reference to work on quarter of birth interrelationships based on the 1960 Census, here the processing problems are much more easily solved because the records for all members of each household are already grouped on the tape in a form designed for further analysis, You could choose between a 25-percent sample, a 5-percent sample or a 0.1l-percent sample and obtain any cross-tabulation of population data you desire based on the information on the 1960 Census sample schedule. The costs of programming will be roughly equivalent for any size sample but the computer running time will vary considerably from sample to sample, If you can use information based on quarters of birth, the best way to proceed would be to specify exactly what you want out of your tabulations and then let us evaluate the request. Enclosed are Technical Memoranda describing the layout of the records with which you might work, No, 29G describes the 25-percent sample in its detailed form (about 2,800 reels), 29H plus 29M describe the 25-percent sample in its abbreviated form (about 1,300 reels), 29W describes the 5-percent sample (about 300 reels). The Technical Documentation for the 0.1l-percent sample describes that record, As mentioned in the small manila leaflet, the 0.1-percent sample is available for immediate purchase for $1,500 and could, therefore, be processed at your installation. There are seven reels in the IBM format and thirteen reels in the Univac format. If you had a limited 3. number of tabulations we might be able to perform them on the 0.1-percent sample at a cost lower than your purchase price plus your programming costs, Please let us hear from you regarding which, if any, of the alternative paths you wish to follow. Sincerely yours, 7 a “ “ce foe , c CO Le Le (als Howard G,. Brunsman Chief, Population Division Bureau of the Census Enclosures