Hon. THADDEUS J. DULSKI, EXECUTIVE OFFICE OF THE PRESIDENT, Chairman, Committee on Post Office and Civil Service, House of Representatives, Washington, D.C. DEAR MR. CHAIRMAN: This is in reply to your request for the views of the Bureau of the Budget on H.R. 20, a bill, "To amend Title 13, United States Code, to limit the categories of questions required to be answered under penalty of law in the decennial censuses of populaton, unemployment, and housing, and for othr purposes." The Department of Commerce in a report to your committee on this bill recommended against enactment of H.R. 20. We have asked other agencies that make extensive use of census data for their views on this bill and the Commerce report thereon. Without exception, these agencies agree with the Commerce position, and some of them cite additional reasons for their opposition to the proposed legislation. We are attaching copies of their replies to us, since they may be of interest to your committee. The Bureau of the Budget concurs in the report of the Department of Commerce and supports the views expressed by other Departments and agencies. In addition, the Bureau of the Budget has conducted its own review under section 209 of the Budget and Accounting Procedures Act and section 3(a) of the Federal Reports Act and specifically agrees with the Commerce report that enactment of H.R. 20 would substantially increase the cost of the decennial census and would decrease the effectiveness of a unified census in providing information needed by agencies of the Federal, State and local governments and by business, educational, and other users of census data. For the above reasons, the Bureau of the Budget recommends against enactment of H.R. 20. Sincerely yours, WILFRED H. ROMMEL, Assistant Director for Legislative Reference. Enclosures: Letters from Department of Agriculture; Department of Health, Education, and Welfare; Department of Housing and Urban Development; Department of Labor; Department of Transportation; Department of the Treasury; and Council of Economic Advisers. Hon. ROBERT P. MAYO, Director, Bureau of the Budget. DEPARTMENT OF AGRICULTURE, DEAR MR. MAYO: This is in reply to your request of March 7, 1969, for our views on the proposed Commerce report on H.R. 20 "To amend title 13, United States Code, to limit the categories of questions required to be answered under penalty of law in the decennial censuses of population, unemployment, and housing, and for other purposes." This Department strongly recommends against enactment of the proposed legislation. We concur in the views expressed by the Department of Commerce that to attempt the conduct of a census on a partly mandatory and partly voluntary basis would present serious problems in data collection and in the statistical treatment of items proposed for voluntary coverage. The taking of a complete and accurate census has never been more important to the Nation than it is today, with the multiple efforts to correct social and economic problems that Government at all levels is currently making. To place a voluntary response label on the heart of the census inquiries is to imply that these questions are somehow not meritorious, or necessary, or proper. It is inconsistent to spend the large amounts of money required for the census and then implicitly encourage nonresponse by labeling part of the response as voluntary and by implication less important. Such a course would say to the public, “Answer if you will but don't bother if you don't want to." Advertising the sample part of the census as voluntary is simply to invite nonparticipation, and under such legislation there would be nothing to prevent organized efforts at nonresponse by persons antagonistic to the purposes of the census-such as was experienced in both the recent South Carolina and Wisconsin pretests of the 1970 census. Hardly a day goes by when the Department does not refer to the Censuses of Population and Housing in the operation of programs, in reaching policy decisions, and in conducting its research. Some items are referred to more frequently than others, but we use literally nearly every item, (e.g. population numbers, residence, age, race, sex, migration, education, labor force status, occupation and industry, condition of housing, plumbing facilities, tenure) response for most of which would not be required under H.R. 20. Only a few of the extremely varied uses the Department makes of decennial census data are noted below. There are certain Federal funds channeled through the Department of Agriculture to State agencies that are allocated by law on the basis of each State's share of the total U.S. rural and farm population. This is true of funds for agricultural extension work and for research conducted by the State agricultural experiment stations. The identification of urban-rural residence is not jeopardized by current efforts to limit the scope and procedures of the next census, but farm residence is an example of an item that would be jeopardized inasmuch as response would not be mandatory under H.R. 20. An instance of the use of census data in program design can be described in the development last year of the Department's submission to the Office of Economic Opportunity of proposals for action under the special impact programs authorized by the Economic Opportunity Amendments of 1967. This legislation authorized programs directed to the solution of critical problems in particular communities, including rural areas that have had substantial outmigrations to urban areas with large concentrations of low-income people. In selecting the rural areas for development under such programs, extensive use is made of census data. Both the income and migration items that are at the heart of such analyses are among those that would be placed on a voluntary basis by H.R. 20 and therefore would be rendered unreliable because of incomplete coverage. Our housing programs, conducted by the Farmers Home Administration, require much use of population and housing census data. The areas in which the Farmers Home Administration is legally authorized to operate housing programs are largely defined in census terms. Specifically, the programs are restricted by legislation to rural areas and towns of less than 5,500 population. The allocation among the States of money available for loans is made on the basis of census statistics on the condition of rural housing of low-income families. The Housing Census data on plumbing and water supply are extensively used in operation of our water association loans program, in which small water systems to supply rural communities are financed. In the last several years, Federal agencies have employed the planning, programming, and budgeting system in the design and justification of their work. This systems approach has considerably increased the use of census statistics because of the requirement to estimate program needs and target populations. Thus the Farmers Home Administration, for example, has had to estimate the magnitude of rural housing requirements by condition of housing and by income class of the population. The Censuses of Population and Housing comprise our only source of base data in this area. One of the Department's major programs is food distribution to low-income families, accomplished in some areas by direct distribution of commodities and in others through the food stamp program. The Department has been extending food distribution rapidly, but in a particular effort to cope with the worst situations, it was decided to place programs promptly in each of those counties among the 1,000 poorest counties in the Nation that did not yet have them. The identification of the counties in which this special distribution effort is being made and in which direct USDA operation is authorized was made entirely on the basis of income data from the 1960 census. Within metropolitan cities, use is made of census tract data by income in administering consumer food programs. These data are essential in identifying localities where concentrations of eligible people live, and thus pinpoint areas where publicity and program assistance work should be conducted. Census income data are also extensively used in administering that phase of the school lunch program that provides funds for special assistance to children from low-income fam ilies. Census data on number of children in low-income families are used to allocate funds for this program among the States, and census tract data are used to plan the program within large cities. The above examples illustrate the extensive reliance that the Department places on statistics from the Censuses of Population and Housing. We would be seriously handicapped in fulfilling our program responsibilities in the absence of census data of the scope, detail, and reliability which we have had in previous censuses and which would be jeopardized by the proposed legislation to remove most of the 1970 inquiries from the normal mandatory obligation to respond. Sincerely, J. PHIL CAMPBELL, THE SECRETARY OF HEALTH, EDUCATION, AND WELFARE, Hon. ROBERT P. MAYO, DEAR MR. MAYO: This letter is in response to Mr. Rommel's request of March 7, 1969, for a report on H.R. 20, a bill to "amend title 13, United States Code, to limit the categories of questions required to be answered under penalty of law in the decennial censuses of population, unemployment, and housing, and for other purposes," and on the proposed report of the Department of Commerce thereon. The bill would require answers for only the following categories: (1) name and address; (2) relationship to head of household; (3) sex; (4) date of birth; (5) marital status; and (6) visitors in home at the time of census. The bill would allow other information to be asked for only on a voluntary basis. The Department of Health, Education, and Welfare recommends against this legislation for the following basic reasons: 1. Dividing the census questions into mandatory and voluntary would gravely reduce the quality of the entire Census of Population and Housing. The rate of voluntary responses most likely would be so low that the information probably would be useless. Moreover, a mixture of mandatory and voluntary questions might be so confusing that the rate of return of answers to mandatory items would be lowered. 2. The Federal Government should be helping State and local governments acquire more-not less-information about economic and social problems within counties and cities. The census is a once-in-10-years opportunity to obtain statistics on a block-by-block basis which can then be combined by school zones, health districts, political boundaries and other areas selected by State and local governments. These statistics are essential for the intelligent sharing of responsibilities among the Federal, State, and community levels. 3. The operation and evaluation of many DHEW programs (direct and intergovernmental) depend on data on subjects which would be placed on a voluntary basis by this bill. A major example is that of the interrelationship of income and educational problems. More specifically, title I of the Elementary and Secondary Education Act of 1965 requires the use of the number of children in families with substandard incomes as a basic factor in distributing Federal funds to local educational agencies. As another example, it would be difficult to analyze the progress of minority groups and the extent to which our programs were reaching them if questions on race and other ethnic characteristics were made voluntary. I agree with the unfavorable reply proposed by the Department of Commerce— particularly its emphasis on Federal, State, and local governmental needs for census data on poverty, education, and older citizens. We would therefore recommend that H.R. 20 not be enacted. Additional technical details and justification for our position are provided in the enclosed technical supplement. Sincerely, ROBERT H. FINCH, Secretary. TECHNICAL SUPPLEMENT Below are provided additional technical considerations justifying the Department's position in opposition to H.R. 20. The record of the Census Bureau The confidentiality of data collected by the Federal Government and nongovernmental organizations is an important issue at this time. The Department believes, however, that H.R. 20 does not really address the issue. The Bureau of the Census has earned a reputation for the inviolability of its data, which is essential to the planning function in this and many other departments. The issue of confidentiality does not rest in the type of data obtained but rather in the use to which the information is put. The census questions do not constitute an invasion of privacy when the information collected is used for statistical purposes only. The importance of a centrally administered census Some proponents of H.R. 20 have argued for greater reliance on State and local governments to collect the information presently collected by the census. The difficulty with this proposal is that the administrative expense of letting individual contracts for local censuses is far greater than that of a centrally administered survey. More importantly, there can be no data comparability with locally run surveys because of the lack of uniform questions, time factors, and collection methods. The national census insures this uniformity; data collected in one locale are comparable to that collected in another. The same data can be used for local, State and national planning purposes. In this way, compatible data are available for use in planning and negotiating at all levels of government. The problem of voluntary questions The voluntary system of answering questions, successfully used by private survey organizations is not practicable for the decennial census. 1. The size of the decennial census is not comparable to that of private survey organizations. The intensive interviewing and followup procedures which are a normal part of a small voluntary survey are impossible with a survey the size of the decennial census. 2. Most sampling surveys involve marketing questions aimed primarily at the middle and upper economic classes. Many of DHEW programs are aimed at lower income groups where we expect a response problem if a voluntary census were instituted. The ghetto population has an innate suspicion of all Government officials and consequently would tend to refuse to answer any voluntary question. The underenumeration in the 1960 census is heavily concentrated in the ghetto18 percent on nonwhite males between the ages of 20 and 35 were missed in the 1960 census. Poorer results can be expected if the voluntary response system were instituted. If, for example, there were a total nonresponse rate of 15 percent to voluntary questions of a social and economic nature, we could expect a much higher rate for the ghetto population, resulting in a drastic underestimate of the needs of this group. This would impair the planning functions of many Government agencies, including HEW. The uses of census data in Department of Health, Education, and Welfare 1. Ceneus figures provide the Department with certain indicators that measure the number of persons with certain types of needs which are addressed by specific departmental programs; for example, the number of families headed by women with an income of less than $3,000; the number of families headed by men with health disabilities, the number of adults who have completed less than 12 years of education. Data routinely collected in the operation of the program does not provide all the information required to determine the need for and effectiveness of the program. Program data describes the number and characteristics of persons benefiting from a program. It can provide no information on the numbers and characteristics of persons needing but not receiving program benefits. An operating program could appear to be operating efficiently with a modest cost per recipient, but it might be ineffective if it is dealing with an insignificant proportion of the population in need. The universe of need for Government programs can only be determined through a sampling on enumeration of the entire population. 2. In addition to being useful program measures, certain variables-such as educational attainment, income, and occupation—have an additional significance in their impact upon other variables. For example, the extent to which unemployment varies with educational attainment, or the utilization of health services with educational attainment, or the utilization of health services with income or education, are questions of great importance to the operation of departmental programs. The use of most social services is so affected by these influence variables that the utilization of services becomes meaningful only in conjunction with utilization rates for various ranges of income or education or types of employment. Failure to collect these kinds of data will deprive the Department of Information needed to plan across the spectrum of its operations. Census variables of particular interest to Department of Health, Education, and Welfare 1. Years of school completed. This is important for use in education and manpower programs, and because of its strong influence on a wide variety of other characteristics of individuals. It permits the determination of the variations in educational achievement for different parts of the country, and the resulting variations in available manpower. 2. School or college enrollment. This is important for assessing the present effectiveness of our educational system and in determining the future manpower pool. 3. Vocational training completed.-Vocational training is becoming an increasingly important adjunct to formal education as a means of upgrading the manpower pool. Information from this item will enable the determination of how certain nonprofessional and lower professional occupations are changing their requirements and upgrading their personnel to meet these new requirements. It will permit an assessment of the effectiveness of vocational training in order to determine the extent to which the Federal Government and other governments should contribute to such programs. 4. Income. This is perhaps the single most important socioeconomic variable collected in the census. It is used as a measure for distributing Federal funds to States and localities. It provides crucial information about families in need of public assistance, and indicates the effect of various types of transfer payments, including social security, veterans payments, and public assistance, on the income of families in need. The 1970 census will allow an analysis of such need by geographical areas and demographic characteristics which has not been possible since the 1960 census. 5. Employment status and hours worked; weeks worked last year.-Manpower programs and welfare programs are both heavily dependent upon measures of unemployment and underemployment in small areas. It is important to determine where the unemployment problems are most severe, and for which groups. Generally, sample surveys do not provide sufficient data to do this analysis on a small enough geographic basis. 6. Occupation, industry, and class of worker.―This data is important in determining the labor force available within given geographic areas, information highly important to economic development programs. When this data is combined with educational attainment we can establish the different educational levels required by differing occupations, thus analyzing future education requirements resulting from changes in the occupational structure. 7. Children ever born.—This data, in connection with demographic data on the mother, is extremely important in determining fertility rates for different populations and areas in the country. It is needed for demographic projections as well as for the planning of certain Federal programs such as maternal and child health care. Data on the number of children in the household is not adequate for fertility studies, since some children may have died or moved out of the household. 8. Presence and duration of disability.—At present, our information on disabilities in the general population is very sparce. Such data is needed by Government and voluntary agencies in order to design more effective vocational rehabilitation programs. 9. Mother tongue.-By stratifying on race we can obtain rather good measures of the differentials between Negroes and other Americans in such important characteristics as income and occupation. It is also important, however, to determine the differentials of other significant minorities, especially Puerto Ricans and Mexican-Americans. The latter stratifications can be done most readily by use of this variable. |