-6

QUESTIONS RELATING TO EMPLOYMENT, OCCUPATION, ETC.:

Of the 32 population items, ten relate to employment, occupation, place of work, etc. For convenience, these items have been tabulated separately in Table E, attached. The table shows both priority rating and geographic level desired.

ITEMS OF LOWEST PRIORITY:

Population Items

-

The following eight population items were rated of lowest priority by those recommending their inclusion. A relatively high percentage of respondents also recommended their omission.

Housing Items -- The following eight housing items were rated of lowest priority by those recommending their inclusion. A relatively high percentage of respondents also recommended their omission.

Tables A and B show the percentage of respondents that recommended omission of each item from a mid-decade census. For convenience, the following summary indicates for the complete count and sample groups the highest and lowest percentage of respondents recommending omission:

(7) 5% Items: Vocational training completed..
Occupation or industry, 5 years ago.
Year of immigration...

24.6

24.6

40.1

(5) 20% Items: Number of units in structure and whether

8.7

30.1

13.1

28.7

17.5

29.2

17.5

35.0

GEOGRAPHIC LEVEL FOR WHICH DATA WOULD BE DESIRED
IN A MID-DECADE CENSUS

Section II of the FSUC questionnaire asked respondents to indicate the geographic detail for which they would require data in a mid-decade census, based on the following levels:

Tables A and B, attached, indicate the percentage of respondents who would require mid-decade information by geographical levels, according to the breakdown shown above, for each of the population and housing items included in the 1970 census.

Need for Small Area Data -- Levels 3 and 4 represent the need for data on a small area basis. The following summarizes the averages of respondents requiring data at these levels for the complete count and sample items in the population and housing censuses:

It is obvious from the above that there is an extremely high degree of interest and definite need for data at the small area level in a mid-decade census. More than 50 percent of the respondents indicated a need for data at the county, urban place, or census tract level, for every single population and housing item.

Although it is recognized that block data can be provided only from information collected in the 100 percent items in the census, the survey shows that there are some respondents who desire block data for each of the subject items included in the 20, 15 and 5 percent samples. In fact, it is important to note that two respondents felt that all population and housing information items were vitally needed at the 5-digit zip code level, for the five-year census.

As would be expected, state and local governments indicated the highest degree of need for data at the block level. To illustrate, on the subject of color or race,' 31 percent of all respondents desire such data at the block level, whereas 48.6 percent of state and local government respondents indicated such a need. The next highest degree of need for block data was by nonprofit research organizations and universities. And, as also would be expected, trade associations and labor unions indicated little or no interest in block data.

RESPONSES BY CLASS OF USER:

Because of the variations in responses between classes of users, special tabulations have been made of the responses for both population and housing items by the six major classes of users covered in the survey. These responses are shown for each of the 32 population items and 33 housing items in Tables F and G, attached.

SUMMARY AND CONCLUSIONS

This survey of user views on a mid-decade census is the most definitive survey ever conducted by the Federal Statistics Users' Conference, not only because of the large number of subject items covered, but for the first time it accurately pinpoints how users rate statistical subject areas in terms of priority and the geographic level desired.

The 59 percent rate of response was the highest of any FSUC survey. The returns indicate that respondents gave considerable time and thought to answering the six-page questionnaire.

Although the survey was designed primarily to determine user views regarding a mid-decade census, it is obvious that the results will also be beneficial in defining user interests and needs in relation to current statistical series or special sample surveys.

-9

With regard to criteria for evaluating statistical programs, FSUC's Long Range Program for Improvement in Federal Statistics states: "Statistical programs which serve multiple and widely felt needs should have priority over those which serve limited purposes...As a corollary to this criterion, every statistical program, existing or proposed, should be considered in terms of possible uses to which the information can be put and should be designed to provide optimum usefulness for these different purposes. In other words, programs should provide the greatest good for the greatest number. On the other hand, this FSUC survey indicates that, in some cases, there are considerable variations in the interests and needs as between groups of users. For example, one group may have little interest or need for a specific type of statistical data, whereas another group will consider the information highly essential for their purposes. The survey shows that there are

often considerable variations in needs (and obviously uses) as between state and local governments and university users.

By the same token, consideration should be given to the differences in needs of non-Federal government users and the Federal government. To illustrate, information on "basements" was given a comparatively low rating by respondents to FSUC's survey. This item, first collected in 1960, is considered of great importance to the Federal government for civil defense purposes. The question on number of children ever born has been collected in most censuses since 1890. In the FSUC survey, 15.7 percent of the respondents recommended its omission, 28.3 considered it of low priority, while 39.8 percent rated it of high priority. Statistics on children ever born are unique in providing significant information on the current and future trends of population growth through births and how the composition of the population is changing through differences in fertility of various population groups. Such data are used for projecting the age of the future population. In turn, these projections are needed by Federal, State and local governments and private industry for the planning of various short- and long-range projects. So far as FSUC respondents are concerned, what this may mean is that a relatively small number are individually concerned with making population projections; while the remainder may have a high interest, but rely on projections made by others, especially those of the Federal government.

With regard to a mid-decade census, the survey conclusively shows an extremely high percentage of respondents favoring both a mid-decade census and mandatory response. On the subject of penalties, the conclusion is that severe penalties are not required.

From the standpoint of priorities, it is obvious that age, sex and color or race would rank high in the population category. Other high priority items include: employment, income, occupation, marital status, education, place of work, means of transportation to work, relationship to head of household, place of residence 5 years ago, and year moved into this house.

In the housing category, the following items had the highest ratings: number of units and whether a trailer, value, contract rent, year structure built, vacancy status, rooms and tenure. Following in importance is information on: automobiles, flush toilet, bathrooms, complete kitchen facilities, months vacant, farm residence and water supply.

The survey results support the statement made by the Executive Director of FSUC in testimony on proposals for a mid-decade census before the Subcommittee on

-10

Census and Statistics of the House Post Office and Civil Service Committee that a full replication of the decennial census is not necessary or desired. The FSUC survey results should provide some guidance in the planning of a scaled-down census, if and when legislation for a mid-decade census is passed.

The survey also provides overwhelming evidence of the great need for small area data by all classes of users. For every single population and housing item, more than 50 percent of the respondents indicated a need for data at the county, urban place, or census tract level. For block data provided from the 100 percent iteme for both population and housing, an average of approximately 27 percent of user respondents indicated a need for this level of data. For these two small area geographic levels, the range of need is from 64 percent to 86 percent.

JOHN H. AIKEN
Executive Director

October, 1971