226 



National Resources Committee 



menting with methods of estimating a correction factor 

 to be reported along with the original data. Likewise, 

 the Division of Manufactures is seeking ways of pub- 

 licly reporting the apparent degree of completeness in 

 its coverage. 



Difficulty of getting additional funds for researrch. — 

 The inability of the Bureau of the Census to get more 

 funds for research aimed at improvement of its data 

 is a problem undoubtedly shared by many other Fed- 

 eral statistical agencies. The reason for the difficulty 

 may differ with different agencies. Assuming that the 

 will to improve exists within a statistical unit which 

 is a division inside of a i-egulatory bureau, most such 

 units still face the problem of "selling" the bureau's 

 administrator, on the importance of such research. If 

 he is "sold," the problem of getting money may then 

 be easier than it is for the Bureau of the Census, be- 

 cause there are indirect as well as direct ways for the 

 statistical unit attached to a regulatory agency to get 

 funds. If he is unsympathetic, however, the difficulties 

 of getting research funds may be even greater than it 

 is for the Bureau of the Census. Although tlie Bureau 

 of the Census is one of many agencies in the Depart- 

 ment of Commerce, and must take its share of budget- 

 ai'y cuts which the Department receives, the Director 

 of the Census has much more freedom of action than 

 the chief of a statistical division subordinate to a regu- 

 latory bureau. However, in going before the Bureau 

 of the Budget and Congress, the Bureau of the Census 

 Ijays the penalty of past neglect of research, because 

 such work now calls for new expenditures rather than 

 for a continuance of past allocations whose usefulness 

 is an accepted tradition. 



Proposal to smooth out work load. — A further step 

 toward an orderly development of research work within 

 the Bureau would be possible if a proposed bill, drafted 

 within the Bureau, could be enacted into law. This 

 bill, described in detail in tlie Journal of the American 

 Statistical Association, March 1938, pp. 230-31, seeks 

 jirimarily to smooth out the work load and make sys- 

 tematic the schedule of operations. One of its most 

 beneficial results would be the reduction in the ratio of 

 temporary to permanent employees. Other important 

 features are provisions for a regular quinquennial Cen- 

 sus of Population ; an annual sample Census of Agri- 

 culture, with complete coverage every 5 years; and, 

 similarly, an amiual sample Census of Manufactures, 

 with complete coverage every 5 years. Statutory au- 

 thority is sought to stagger various reports in such a 

 way as to permit a larger proportion of the aggregate 

 census appropriation to go to permanent rather than 

 temporary employees. The indispensability of a quin- 

 quennial Census of Population — not only to such agen- 

 cies as the Social Security Board, but also to any Gov- 

 ernment agency allocating funds to States or cities on a 



basis of population, as well as to research workers on 

 population problems — has been proclaimed almost uni- 

 versally. Various reports of the National Resources 

 Committee have been emphatic in recommending it. 

 Similarly, the proposed annual sample surveys of Agri- 

 culture and Manufactures are necessary to Government 

 administration and to private research as well. 



Summary 



It was pointed out that the first great responsibility 

 of the Bureau of the Census is to produce the raw 

 material for further processing by statisticians every- 

 where, in and out of the Government. Beyond that 

 vital routine operation, how much analytical research 

 should the Bureau of the Census do? 



This question was considered by distinguishing three 

 types of analytical work conducted by Government sta- 

 tistical organizh,tions : 



1. Investigations of the accuracy of the basic data, with the 

 aim of finding ways of Improving its collection and tabulation and 

 helping to determine what new questions to ask on schedules. 



2. Research on the social and economic significance of re- 

 lationships derived from basic tabulations, with the general 

 aim of making additions to knowledge. 



3. Quick analysis of data with the aim of providing guidance 

 to an administration facing a particular administrative problem. 



The first type, it was maintained, was clearly within 

 the province of the Bureau of the Census; the third 

 type, generally not, being best performed by analysts 

 directly attached to a given administrator. The ques- 

 tion of responsibility for the second type is perhaps 

 more open to debate. 



With respect to the second type of analytical re- 

 search the Bureau can point — especially between 1880 

 and 1910 — -to much good performance. AVhen univer- 

 sity research in the social sciences was in its infancy 

 the Bureau produced a large volume of research of 

 high quality. Most of this appeared as textual ac- 

 companiments of the decennial census reports. The de- 

 lays in publication of the basic data, while the analysis 

 was awaited, caused considerable criticism, and in 1920 

 the text accompanying the tables was reduced to a 

 bare introduction. The analytical studies were trans- 

 ferred to separate monographs in selected fields. In 

 1930 the text in the regular volumes also was brief, 

 but the separate monographs did not represent so much 

 actual analytical research as they did the presentation 

 of additional tabidar material which for one reason or 

 another had been excluded from the census. 



Arguments were presented with respect to the re- 

 sponsibility of the Bureau of the Census for analytical 

 work of the second type: (1) That the Bureau of the 

 Census should not, do such work (a) because this is 

 not the function of a Government agency; (b) be- 

 cause, if it is the function of Government, it should 



